2017 Lake Michigan Ozone Study
(LMOS)
Preliminary Finding Report
LMOS 2017 Study Team
Initial release 7/19/2018; Revised release 4/22/2019
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Maryam Abdioskouei (maryam-abdioskouei@uiowa.edu)
Zac Adelman (Adelman@ladco.org)
Jassim Al-Saadi (j.a.al-saadi@nasa.gov)
Timothy Bertram (timothy.bertram@wisc.edu)
Greg Carmichael (gregory-carmichael@uiowa.edu)
Megan Christiansen (megan-christiansen@uiowa.edu)
Patricia Cleary (CLEARYPA@uwec.edu)
Alan Czarnetzki (Alan.Czarnetzki@uni.edu)
Angela Dickens (Angela.Dickens@wisconsin.gov)
Marta Fuoco (fuoco.marta@epa.gov)
Monica Harkey (mkharkey@wisc.edu)
Laura Judd (laura.m.judd@nasa.gov)
Donna Kenski (kenski@ladco.org)
Dylan Millet (dbm@umn.edu)
Brad Pierce (brad.pierce@noaa.gov)*
Charles Stanier (charles-stanier@uiowa.edu)*
Betsy Stone (betsy-stone@uiowa.edu)
James Szykman (szykman.jim@epa.gov)
Lukas Valin (Valin.Lukas@epa.gov)
Tim Wagner (tim.wagner@ssec.wisc.edu)
23
*indicates co-lead author on Preliminary Findings Report
24
This report documents preliminary results from the 2017 Lake Michigan Ozone Study (LMOS 2017).
Any opinions, findings, and conclusions or recommendations expressed in this material are those of
the author(s) and should not be construed as an official U.S. Government position, policy, or decision.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
1
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
LIST OF FIGURES AND TABLES ........................................................................................................................... 4
EXECUTIVE SUMMARY ..................................................................................................................................... 8
ACKNOWLEDGMENTS .................................................................................................................................... 12
LIST OF COOPERATING ORGANIZATIONS ......................................................................................................... 12
REVIEW ......................................................................................................................................................... 12
SECTION 1.
MISSION AND STUDY PERIOD OVERVIEW................................................................................ 13
1.1.
BACKGROUND: OZONE IN THE LAKE MICHIGAN REGION ......................................................................................... 13
1.2.
OBJECTIVES OF LMOS 2017 ............................................................................................................................. 16
1.3.
OVERVIEW OF THE LMOS 2017 MISSION ........................................................................................................... 16
1.4.
METEOROLOGY DURING LMOS 2017................................................................................................................. 21
1.5.
LAKESHORE OZONE DURING LMOS 2017 ........................................................................................................... 28
SECTION 2.
SUMMARY OF LMOS OBSERVATIONS AND KEY RESULTS.......................................................... 32
2.1.
SPACEPORT SHEBOYGAN, WI MEASUREMENTS ..................................................................................................... 32
2.1.1.
Overview ............................................................................................................................................ 32
2.1.2.
Methods ............................................................................................................................................. 33
2.1.3.
Results and Discussion ....................................................................................................................... 35
2.2.
ZION, IL, MEASUREMENTS ................................................................................................................................ 39
2.2.1.
Overview ............................................................................................................................................ 39
2.2.2.
Methods ............................................................................................................................................. 41
2.2.3.
Results and Discussion ....................................................................................................................... 43
2.3.
PANDORA/CEILOMETER MEASUREMENTS ............................................................................................................ 51
2.3.1.
Overview ............................................................................................................................................ 51
2.3.2.
Methods ............................................................................................................................................. 52
2.3.3.
Results and Discussion ....................................................................................................................... 52
2.4.
GEOTASO MEASUREMENTS ............................................................................................................................. 55
2.4.1.
Overview ............................................................................................................................................ 55
2.4.2.
Methods ............................................................................................................................................. 55
2.4.3.
Results and Discussion ....................................................................................................................... 57
2.5.
SCIENTIFIC AVIATION MEASUREMENTS ................................................................................................................ 60
2.5.1.
Overview ............................................................................................................................................ 60
2.5.2.
Methods ............................................................................................................................................. 60
2.5.3.
Results and Discussion ....................................................................................................................... 61
2.6.
MOBILE PLATFORM MEASUREMENTS .................................................................................................................. 65
2.6.1.
Overview ............................................................................................................................................ 65
2.6.2.
Methods ............................................................................................................................................. 65
2.6.3.
Results and Discussion ....................................................................................................................... 67
SECTION 3.
SUMMARY OF LMOS MODEL/MEASUREMENT EVALUATIONS .................................................. 72
3.1.
WRF-CHEM MODEL ...................................................................................................................................... 73
3.1.1.
WRF-Chem domain ............................................................................................................................ 73
3.1.2.
University of Iowa forecast system .................................................................................................... 75
3.1.3.
Post-analysis ...................................................................................................................................... 76
3.1.4.
Soil NO Emissions ............................................................................................................................... 81
3.1.5.
Future directions ................................................................................................................................ 82
3.2.
NAM-CMAQ MODEL ..................................................................................................................................... 82
3.3.
WRF-MET MODELING ..................................................................................................................................... 89
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
2
71
72
73
74
75
SECTION 4.
RECOMMENDATIONS FOR METEOROLOGICAL/PHOTOCHEMICAL MODELING .......................... 97
4.1.
PHOTOCHEMICAL MODELING............................................................................................................................. 97
4.2.
METEOROLOGICAL MODELING ........................................................................................................................... 98
REFERENCES ................................................................................................................................................ 100
76
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
3
77
List of Figures and Tables
78
79
Figure 1. Schematic illustration of LMOS 2017 measurement assets.NASA airborne remote sensing flights (blue raster pattern)
were based out of Madison, WI. .................................................................................................................................... 8
80
Figure 1.1.2. Map of ozone design values and LMOS 2017 ground site locations. .......................................................................... 15
81
82
Figure 1.1.3. Conceptual model of land/lake breeze circulations responsible for enhanced ozone production along the shores of
Lake Michigan (Foley et al., 2011)................................................................................................................................ 15
83
84
Figure 1.4.1. Schematic showing the types of measurements involved in LMOS 2017, including two types of aircraft
measurements, ship-based measurements, ground-based measurements and mobile platforms. .......................... 17
85
Table 1.3.1. LMOS Science Team PIs................................................................................................................................................. 18
86
Table 1.3.2. Summary of measurements made during LMOS 2017. ................................................................................................ 19
87
Table 1.3.3. Models being evaluated using LMOS 2017 data........................................................................................................... 21
88
Figure 1.4.1. National maps of departure from climatological means of precipitation and temperature for May 2017. .............. 23
89
Figure 1.4.2. Same as Figure 1.4.1 for June 2017 (https://www.ncdc.noaa.gov/sotc/synoptic/201706) ....................................... 23
90
Figure 1.4.3. CONUS meteorological overview for the week starting May 21. ................................................................................ 24
91
Figure 1.4.4. CONUS meteorological overview for the week starting May 28. ................................................................................ 25
92
Figure 1.4.5. CONUS meteorological overview for the week starting June 4. ................................................................................. 26
93
Figure 1.4.6. CONUS meteorological overview for the week starting June 11. ............................................................................... 27
94
Figure 1.4.7. CONUS meteorological overview for the week starting June 18. ............................................................................... 28
95
Figure 1.5.1. Time series of one-hour surface ozone concentration (ppbv) during May 22 – June 22, 2017. ................................ 29
96
Figure 1.5.2. Maximum daily eight-hour average (MDA8) ozone mixing ratios (ppbv) on June 2, 2017......................................... 30
97
Figure 1.5.3. Ozone concentrations during LMOS were representative of climatological values. .................................................. 31
98
Figure 2.1.1. Location of the LMOS ground based measurements in Sheboygan, WI. .................................................................... 33
99
Table 2.1.1. Summary of EPA Measurements at Sheboygan, WI, during LMOS 2017 ..................................................................... 34
100
101
Figure 2.1.2. Location of the 2B-Personal Ozone Monitors in relation to the lake shore and the main Spaceport Sheboygan
research site. ................................................................................................................................................................ 35
102
103
Figure 2.1.3. Time series of O3, NO2 and HCHO mixing ratios as measured at the Spaceport Sheboygan research station during
LMOS. ........................................................................................................................................................................... 36
104
105
Figure 2.1.4. Time series of the ratio of HCHO/NO2 mixing ratios and wind direction measured at the Spaceport Sheboygan
research station during LMOS...................................................................................................................................... 36
106
107
Figure 2.1.5. Visualization of boundary layer evolution (winds, temperature, and aerosol backscatter) for June 2 and 3, 2017 at
Sheboygan, WI. ............................................................................................................................................................ 37
108
109
Figure 2.1.6. Visualization of boundary layer evolution (winds, temperature, and aerosol backscatter) for June 10 and 11, 2017
at Sheboygan, WI. ........................................................................................................................................................ 38
110
111
Figure 2.1.7. Comparison of 2B-POM ozone from Sheboygan Fire Department, Kohler Public Works, and Spaceport Sheboygan
sites. ............................................................................................................................................................................. 39
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
4
112
113
Figure 2.2.1. Schematic of the chemical ionization source and four stage (S1-S4) differentially pumped interface coupled to the
time-of-flight mass spectrometer. ............................................................................................................................... 42
114
Figure 2.2.2. Hourly-averaged O3 and NOx measurements at the Zion site. .................................................................................... 43
115
Figure 2.2.3. Hourly-averaged ozone and estimated PM2.5 at the Zion site. .................................................................................... 44
116
Figure 2.2.4. Zion wind rose for the duration of the study, based on 1-min averages. ................................................................... 44
117
Figure 2.2.5. Left: Ozone vs. wind direction and speed and Zion, IL. ............................................................................................... 45
118
Table 2.2.1. Summary of ambient concentrations of PM2.5 organic and elemental carbon (EC and OC) and inorganic ions. ........ 46
119
120
Figure 2.2.6. The concentration of major bulk components of PM2.5 during the high ozone event, high ozone period, and
background................................................................................................................................................................... 46
121
Figure 2.2.7. Time series of HNO3 (top) and H2O2 (bottom) mixing ratios as measured at the Zion ground station during LMOS. 47
122
123
Figure 2.2.8. Measured mixing ratios of isoprene, MVK+MACR, monoterpenes, and toluene at the Zion field site during LMOS2017. ............................................................................................................................................................................. 48
124
Figure 2.2.9. Radiometer and sodar measurements at Zion on June 2, 2017. ................................................................................ 50
125
Figure 2.2.10. Radiometer and sodar measurements at Zion on June 11, 2017 (as in Fig. 2.2.9). .................................................. 51
126
127
Figure 2.3.1. All LMOS CL51 First Mixed Layer Height measurements (km AGL) between 15 May and 25 June 2017 by time of day
(CST, gray lines). ........................................................................................................................................................... 53
128
Figure 2.3.2. Comparison of in situ NO2 columns (x axis) and Pandora total column (y axis).......................................................... 54
129
130
Figure 2.3.3. All LMOS 2017 Pandora NO2 column measurements between 15 May and 25 June 2017 by time of day (CDT, gray
lines). ............................................................................................................................................................................ 55
131
Figure 2.4.1. Depiction oof GeoTASO sampling strategy and photo of GeoTASO instrument. ....................................................... 56
132
133
Figure 2.4.2. Flight plans within the LMOS ‘playbook’ separated by coastal mapping surveys of the western shore (left) and
emissions sampling over Chicago and Milwaukee (right). ........................................................................................... 57
134
Figure 2.4.3. Preliminary comparison of the GeoTASO DSCs from the LMOS flights to the Pandora tropospheric slant column. . 58
135
Figure 2.4.4. GeoTASO NO2 differential slant column maps for May 22 and June 15, 2017. .......................................................... 59
136
137
Figure 2.4.5. Maps of NO2 DSCs observed by GeoTASO on June 17 and 18, 2017 over the Chicago metropolitan area in the early
morning. ....................................................................................................................................................................... 60
138
Figure 2.5.1. Scientific Aviation flights during LMOS 2017 ............................................................................................................... 61
139
Figure 2.5.2. Three dimensional (latitude, longitude, altitude) view of Scientific Aviation Flight on June 02, 2017....................... 63
140
Figure 2.5.3. Three dimensional (latitude, longitude, altitude) view of Scientific Aviation Flight on June 11, 2017....................... 64
141
Figure 2.6.1. Southern transects driven by the GMAP van. ............................................................................................................. 66
142
Table 2.6.1. GMAP Zion collocation comparisons ............................................................................................................................ 66
143
Table 2.6.2. UW-Eau Claire automobile measurement stops .......................................................................................................... 67
144
145
Figure 2.6.2. Ozone concentration (ppbv) for GMAP Zion Illinois Beach State Park 21st St Transect – West (Transect 1) on June 8,
2017 as a function of time (UTC). ................................................................................................................................ 68
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
5
146
147
Figure 2.6.3. Ozone concentration (ppbv) for GMAP Kenosha O 3 monitor to Lakeshore (Transect 6) on June 12, 2017 as a
function of time (UTC). ................................................................................................................................................. 68
148
149
Figure 2.6.4. Ozone concentration (ppbv) for GMAP Windpoint to Belmar Ave Transect (Transect 9) on June 12, 2017 as a
function of time (UTC). ................................................................................................................................................. 69
150
Figure 2.6.5. UW-Eau Claire automobile platform ozone measurements from June 2, 2017. ........................................................ 70
151
Figure 2.6.6. UW-Eau Claire automobile platform ozone measurements from June 3, 2017. ........................................................ 71
152
Table 2.6.1. Summary of forecast models used during LMOS.......................................................................................................... 72
153
Figure 3.1.1. WRF-Chem domain: A. Chicago, B. Zion, C. Kenosha, D. Kohler Andre....................................................................... 73
154
Table 3.1.1. Summary of WRF-Chem configuration ......................................................................................................................... 74
155
Table 3.1.2. Emission scenarios designed for LMOS forecasting and post-analysis system ............................................................ 75
156
Figure 3.1.2. Example of forecast workflow for three days May 21st, 22nd, and 23rd .................................................................... 76
157
158
Figure 3.1.3. Example of forecast products available for the flight briefings. Cross section plots (A), curtain plots (B), spatial plots
(C), and time series graphs (D). .................................................................................................................................... 76
159
160
Table 3.1.3. Summary of wind speed and statistics at along the shore (Zion, Sheboygan, Manitowoc) and inland (Midway and
O’Hare) stations for the duration of 2017-05-22 to 2017-06-21. ................................................................................ 77
161
Table 3.1.4. Summary of Model performance in capturing O 3 at Chicago and Zion sites. .............................................................. 80
162
163
Figure 3.1.5. Scientific Aviation vertical profiles (black) vs. paired WRF-Chem simulation results (red) for O3 (top row) and NO2
(bottom row) over land and over lake spirals .............................................................................................................. 81
164
165
Figure 3.2.1. Wind Speed (m/s, upper) and Ozone (ppbv, lower) wind roses for 1-minute WDNR observations (left) and hourly
NAM-CMAQ (right) at the Chiwaukee, WI monitor during the period from May 22 through June 22, 2017............. 84
166
167
Figure 3.2.2. Wind Speed (m/s, upper) and Ozone (ppbv, lower) wind roses for 1-minute WDNR observations (left) and hourly
NAM-CMAQ (right) at the Sheboygan KA monitor during the period from May 22 through June 22, 2017. ............. 85
168
169
170
Figure 3.2.3. Statistical comparison between Scientific Aviation (SA, yellow bars/solid line) and NAM-CMAQ (red bars/dotted
line) vertical profiles of wind direction (Degrees, upper left), temperature (K, upper right), ozone (ppbv, lower left),
and NO2 (ppbv, lower right) over Lake Michigan during LMOS. .................................................................................. 87
171
172
173
Figure 3.2.4. Statistical comparison between Scientific Aviation (SA, yellow bars/solid line) and NAM-CMAQ (red bars/dotted
line) vertical profiles of Wind Direction (Degrees, upper left), Temperature (K, upper right), Ozone (ppbv, lower
left), and NO2 (ppbv, lower right) over Lake Michigan on June 02 and June 11, 2017............................................... 88
174
Figure 3.3.1. Modeling domains used in the first round of meteorological parameterization sensitivity studies. ......................... 89
175
176
Figure 3.3.2. Temperature (top), wind direction (middle), and wind speed (bottom) as observed (black lines) and modeled (red
lines) at Chiwaukee. ..................................................................................................................................................... 90
177
Table 3.3.1. Statistical comparison of WRF (EPA baseline configuration) with observed meteorology at near-lakeshore sites. ... 90
178
179
Figure 3.3.3. Average mixed layer height (meters) as observed at Zion (black line) and modeled using the EPA baseline WRF
configuration (red line). ............................................................................................................................................... 91
180
Table 3.3.2. WRF sensitivity experiments relative to the EPA baseline. .......................................................................................... 92
181
182
Table 3.3.3. Statistical comparison (temperature) of WRF (alternate configurations) with observed meteorology at nearlakeshore sites. ............................................................................................................................................................. 92
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
6
183
184
Table 3.3.4. Statistical comparison (wind direction) of WRF (alternate configurations) with observed meteorology at nearlakeshore sites. ............................................................................................................................................................. 92
185
186
Table 3.3.5. Statistical comparison (wind speed) of WRF (alternate configurations) with observed meteorology at nearlakeshore sites. ............................................................................................................................................................. 92
187
188
Figure 3.3.4. SST values output from WRF using EPA baseline configuration (top), and from gridded analysis from the GLSEA
dataset (bottom). ......................................................................................................................................................... 93
189
Figure 3.3.5. Modeling domains used for ongoing meteorological parameterization sensitivity studies. ...................................... 94
190
191
Figure 3.3.6. VIIRS true color image (center) with 1.3-km resolution planetary boundary layer height forecasts for the EPA
Baseline (Pleim-Xiu) LSM (left) and WRF Noah LSM sensitivity experiment (right) at 19Z on May 27, 2017 ............. 95
192
Figure 3.3.7. 1/12th degree GDAS (left) and 1.8km GLSEA (right) Lake Michigan SST (°C) analyses on May 27, 2017. ................. 96
193
194
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
7
195
196
Executive Summary
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
The 2017 Lake Michigan Ozone Study
(LMOS 2017) was a collaborative, multi‐agency
field study of ozone chemistry and meteorology
along the Wisconsin‐Illinois Lake Michigan
shoreline using a combination of aircraft, ground‐
based and ship‐based measurements. The
collection of measurement assets deployed is
conceptually illustrated at right. The Lake
Michigan region is seasonally impacted by
episodes of high ozone concentrations. These
episodes are influenced by complex meteorology
and significant transport of pollutants. The
overall goal of the study was to better
understand ozone formation and transport
around Lake Michigan to assist researchers and
air quality managers who study, predict, and
manage ozone concentrations in the region. Two
particularly challenging issues that LMOS 2017
targeted were understanding the factors that
determine the size, location, timing, and intensity
of ozone-rich air masses, and understanding the
details of the chemistry and meteorology that
create the sharp gradients in ozone
concentrations often observed between the
lakeshore and nearby inland locations.
222
223
224
225
226
227
228
229
Field activities were conducted from May
22‐June 22, 2017 and included two aircraft (one
for remote sensing and one for in situ profiling), two ground based super sites (Sheboygan, WI and Zion,
IL) outfitted with remote sensing and in situ measurements, three mobile sampling platforms measuring
lakeshore‐inland ozone concentration gradients or conditions on the lake surface, and additional
ground‐based remote sensing instruments collocated at several other shoreline monitoring locations.
Air quality and meteorological forecasts provided flight planning guidance and in‐field evaluation of
model prediction skill during the study.
230
231
232
233
Section 1 of this report provides an overview of the LMOS 2017 study including a summary of
objectives and an overview of the meteorology and lakeshore ozone concentrations during the field
observation period. Section 2 provides a summary of the measurements conducted during the LMOS
2017 study, along with preliminary results. Section 3 provides a summary of preliminary comparisons of
Figure 1. Schematic illustration of LMOS 2017 measurement
assets.NASA airborne remote sensing flights (blue raster
pattern) were based out of Madison, WI.
Scientific Aviation airborne in situ flights (red spirals) were
based out of the Sheboygan County Municipal Airport. Remote
and in situ measurements were deployed at Spaceport
Sheboygan, in Sheboygan WI (northern coastal site) and at the
Illinois EPA monitor located within the Illinois Beach State Park
in Zion, IL (southern coastal site). Additional remote sensing
instruments here placed at WDNR and Illinois EPA monitoring
sites along the Wisconsin shoreline coast and Chicago. Mobile
measurements were also made along the Wisconsin and Illinois
shoreline.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
8
234
235
236
237
photochemical model results with measurements highlighting the difficulties in predicting coastal ozone
concentrations along the western shore of Lake Michigan. Section 4 provides initial recommendations
for meteorological and photochemical model configurations to improve the prediction of ozone along
the western shore of Lake Michigan.
238
Key Findings:
239
240
241
242
243
244
Observed and Modeled Ozone. Significant ozone events occurred during LMOS 2017, with exceedances
of the 70 ppbv 8-hr ozone threshold on June 2, June 11-12, and June 14-16. The LMOS 2017 aircraft
observed polluted layers with rapid ozone formation occurring in a shallow layer near the Lake Michigan
surface. Modeling and observations show that this polluted layer over the lake is an important factor in
ozone exceedance events, but that meteorological and photochemical model skill in forecasting these
events needs improvement.
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
Specifically, airborne in situ profiling during selected events showed high ozone and nitrogen dioxide
(NO2) concentrations within the marine boundary layer over Lake Michigan, whose depth as determined
by in situ temperature profiling ranged from ~50 to ~370 m above lake level depending on date, time of
day, and location. Comparisons between model predictions and in situ airborne and surface ozone
measurements showed that both National Weather Service (NWS) Community Multi-scale Air Quality
(CMAQ) and high resolution (4km) Weather Research and Forecasting (WRF) with Chemistry (WRFChem) forecasts underestimated peak ozone concentrations and overestimated NO 2 concentrations
during these events. Both NWS and WRF-Chem models showed persistent underestimates in terrestrial
boundary layer (0-3km) ozone concentrations throughout the LMOS 2017 study. Preliminary emissions
sensitivity studies using the WRF-Chem model suggest that both reductions in anthropogenic nitrogen
oxide (NOx) emissions and increases in anthropogenic volatile organic compounds (VOCs) improve
model results for peak ozone during LMOS 2017; however, evaluation at the process level and
replication in additional models is necessary. WRF physics sensitivity studies suggest that the ability to
capture the inland penetration of the lake breeze circulation is dependent on accurate estimates of the
Lake Michigan water temperatures and the physics options chosen to represent boundary layer mixing
and land surface exchange processes.
261
262
263
264
265
266
267
268
269
270
271
Sheboygan Ground Super-site. At Sheboygan a valuable dataset of continuous wind, temperature, water
vapor, and vertically-resolved aerosol profile was recorded. Meteorological characteristics of lake breeze
penetration were captured in detail, including during the June 2 ozone exceedance event. In situ
chemical measurements at Sheboygan of ozone, NOx, formaldehyde, and NOx plus its reaction products
(NOy) at 1 minute and finer time resolution captured peak events, as well as diel cycles in precursors and
oxidants, and wind-shift related changes. Chemicals (formaldehyde and NO2) that can serve to indicate
the chemical regime for ozone formation (i.e., NOx limited and volatile organic compound (VOC) limited)
were measured at Sheboygan. Substantial variability in the indicators was recorded, suggesting a
complex system with some NO x limited periods, and some VOC limited periods. However, lake breeze
influence within this record is evident, and the chemical indicators at Sheboygan, when combined with
the supporting datasets are promising for ongoing investigation.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
9
272
273
274
275
An experimental network of four lower cost ozone monitors (2B-POM monitors) was spaced out over a 6
km area of Sheboygan to measure differences in concentrations with respect to distance from the lake.
Intermittent data capture from these devices limits the drawing of detailed conclusions regarding spatial
gradients.
276
277
278
279
280
281
282
283
284
285
286
Zion Ground Super-site. At Zion, IL, a dataset of continuous wind, temperature, and water vapor vertical
profile was recorded. These instruments captured thermal and wind characteristics of lake breeze
penetration in excellent detail, including during the June 2 ozone exceedance event. This should provide
a useful dataset for future research and model evaluation. At Zion, a comprehensive suite of chemical
and physical aerosol characterization was captured. This will be used in ongoing source apportionment
analyses and to characterize local influences on the Zion monitoring station. During LMOS 2017, low
levels of fine particulate matter dominated by organic matter were recorded, with PM2.5 enhancements
often coinciding with ozone enhancements. During the largest lake breeze ozone event, elemental
carbon concentrations were enhanced on a relative basis, consistent with a plume including significant
combustion influence. Future organic source apportionment should be able to distinguish relative
contributions from different combustion source types.
287
288
289
290
291
292
293
294
295
Continuous or semi-continuous measurements of ozone, NOx, nitric acid, hydrogen peroxide, and many
volatile organic compounds (VOC) and oxygenated volatile organic compounds (OVOC) were conducted.
Some prominent VOC and OVOC compounds measured included benzene, toluene, isoprene,
monoterpenes, methyl vinyl ketone, and methacrolein. Methyl vinyl ketone and methacrolein are
oxidation products of isoprene, and can be useful indicators of biogenic influence. Preliminary analysis
of these indicates that the field observation period can be divided into two periods – an initial period of
low biogenic influence and higher anthropogenic influence (higher anthropogenic VOCs, low isoprene).
After a transition period of about June 4-June 8, conditions shift to a higher biogenic influence with
lower concentrations of anthropogenic VOCs.
296
297
298
299
300
301
302
Chemicals (nitric acid and hydrogen peroxide) that can serve to indicate the chemical regime for ozone
formation (i.e., NOx limited and VOC limited) were measured at Zion. Background conditions with
westerly winds were mainly NOx limited as expected given regional sources of VOCs exceeding regional
NOx sources. Lake breeze ozone events exhibited higher VOC sensitivity than other periods, and
analyses of these is ongoing. As was the case for Sheboygan, the chemical regime for ozone formation
at Zion had a complex time history; however, within this complexity, the lake breeze influence is clearly
evident and this is expected to be useful in ongoing investigations.
303
304
305
306
307
308
309
Ground-based remote sensing. USEPA deployed remote sensing instruments for mixed layer height,
cloud layer height, column NO2, and column ozone (Vaisala CL51 ceilometers, and UV/visible Pandora
spectrometers). The ceilometers were installed at Grafton, Milwaukee, and Zion. The Pandora
spectrometers were installed at Sheboygan, Grafton, Milwaukee, Zion, and Schiller Park (Chicago). Data
on mixed layer height, ozone column amounts, and NO2 column amounts are being included in ongoing
analyses such as comparison to models, comparison to aircraft in situ profiles, and comparison to other
remote sensing instruments such as GeoTASO and satellites.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
10
310
311
312
313
314
315
316
317
318
Aircraft measurements. Airborne sampling captured critical air pollution and meteorological features
during LMOS 2017. The Geostationary Trace gas and Aerosol Sensor Optimization (GeoTASO) flew 21
research flights on board the NASA LarC UC-12 aircraft. A light aircraft (Scientific Aviation) flew 22
research flights, measuring NO2, ozone, carbon dioxide, methane, and meteorological parameters.
Preliminary analysis of these datasets indicates that they captured their research objectives well,
including strong spatial gradients associated with point and area sources, spatial gradients associated
with oxidative aging of plumes, chemical and thermal stratification over Lake Michigan, and weekdayweekend differences in NOx emissions. A preliminary comparison between Pandora, aircraft, and
GeoTASO NO2 profiles has been completed and indicates satisfactory agreement.
319
320
321
322
323
324
325
Other mobile platforms. Three additional mobile platforms operated during LMOS 2017 at the surface:
(i) the Geospatial Monitoring of Air Pollution (GMAP) mobile vehicle sampled ozone in transects parallel
and perpendicular to the shore, (ii) mobile meteorological and ozone sampling was performed, and (iii)
an instrumented research ship operated out of Sheboygan with daily measurements of a number of
relevant chemical and meteorological parameters. Finally, an aerosol and cloud research-grade remote
sampling system, AirHARP was tested from the NASA LarC UC-12 aircraft. Analysis of these datasets and
integration with the other model- and observation-based analyses is ongoing.
326
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
11
327
Acknowledgments
328
329
330
331
332
333
The LMOS 2017 Science Team acknowledges the NASA Airborne Science Program and the NASA
GEOstationary Coastal and Air Pollution Events (GEO-CAPE) Mission Pre-formulation Science Working
Group for supporting the airborne remote sensing instruments, NSF (award number 1712909) and
NOAA GOES-R Program Office for supporting the measurements at the Zion ground site, EPA and NOAA
GOES-R program office for supporting the measurements at the Sheboygan ground site, and the Electric
Power Research Institute (EPRI) for supporting the Scientific Aviation airborne in situ measurements.
334
335
336
337
338
339
340
341
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the
author(s) and do not necessarily reflect the views of the National Science Foundation nor should they be
construed as an official National Oceanic and Atmospheric Administration or U.S. Government position,
policy, or decision. The United States Environmental Protection Agency’s Office of Research and
Development partially performed and funded the research described within this report. This report
should be considered a draft report and has not undergone a full EPA review. As such the report is not
yet approved for full external publication. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
342
343
344
List of Cooperating Organizations
345
346
347
348
Research institutions: National Oceanic and Atmospheric Administration (NOAA), National Aeronautics
and Space Administration (NASA), U.S. Environmental Protection Agency (EPA), University of WisconsinMadison, University of Iowa, University of Minnesota, University of Northern Iowa, University of
Wisconsin-Eau Claire, University of Maryland Baltimore County, Scientific Aviation.
349
350
351
Air quality management agencies: Lake Michigan Air Directors Consortium (LADCO), Wisconsin
Department of Natural Resources (WDNR), Illinois Environmental Protection Agency (IEPA), and Indiana
Department of Environmental Management.
352
353
Review
354
355
356
357
358
359
This report was released to internal stakeholders on July 19, 2018. Comments were collected by LADCO
and referred to Brad Pierce and Charles Stanier. Comments were submitted by EPA Region V, by staff
from the states of Michigan, Illinois, and Wisconsin, by the Electric Power Research Institute, and by
staff from LADCO. Comments were addressed and the list of comments and their responses (16 pages)
is available upon request from LADCO, or from the corresponding authors Brad Pierce and Charles
Stanier.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
12
360
361
Section 1.
Mission and Study Period Overview
362
1.1. Background: Ozone in the Lake Michigan Region
363
364
365
366
367
368
369
370
Counties around Lake Michigan have a long history of recording ozone concentrations that
exceed the level of the National Ambient Air Quality Standards (NAAQS). Since the promulgation of the
first ozone NAAQS in 1979, lakeshore counties in Wisconsin, Illinois, Indiana and Michigan have been
designated nonattainment with each subsequent standard. While ozone concentrations have decreased
substantially due to implementation of an array of measures controlling emissions of ozone precursors
(WDNR, 2017), two Lake Michigan areas are currently designated nonattainment for the 2008 ozone
NAAQS: the Chicago nonattainment area and Sheboygan County, WI. Additionally, lakeshore areas in all
four states were recently designated nonattainment for the 2015 ozone NAAQS.
371
372
373
374
The persistent high ozone episodes affecting the region have been shown to be influenced by high
precursor emissions from large transportation, urban, and industrial sources in upwind parts of the
airshed coupled with the complex meteorology associated with the land-water interface of the Lake
Michigan shoreline.
375
376
Two types of meteorological patterns are associated with elevated lakeshore ozone concentrations in
the region:
377
378
1)
Synoptic scale meteorology transports high concentrations of ozone and ozone
precursors northward from source regions to the south and southeast, and
379
380
381
2)
Mesoscale meteorology (via land-lake breeze circulation patterns) carries precursors
over the lake at night and in the early morning, where they react to form ozone after
sunrise. Daytime winds then shift to bring the high ozone onshore.
382
383
384
385
386
387
388
389
390
391
392
393
394
395
High pressure systems have been shown to generate meteorological conditions favorable to
elevated ozone as they move through the eastern U.S. from west to east during late May to early
September (Dye et al., 1995; Hanna and Chang, 1995; Haney et al., 1989). These systems are typified by
hazy, sunny skies with generally weak, clockwise-rotating winds and a relatively shallow boundary layer,
which inhibits the vertical mixing and dilution of near-surface pollution concentrations. These
meteorological conditions contribute to the buildup of ozone precursors and facilitate the chemical
production of new ozone. The highest ozone concentrations in the Lake Michigan region result when
polluted air from upwind, high emissions regions is transported northward along the western edge of a
high pressure system, adding to locally produced ozone (Ragland and Samson, 1977). On high ozone
days, the lowest lakeshore ozone concentrations are typically found in the areas with high emissions of
nitrogen oxides (NOx), such as in central Chicago and northwestern Indiana. The highest concentrations
are found downwind of the high NOx sources in rural and suburban coastal areas. During classic Lake
Michigan ozone transport episodes ozone plumes are observed moving northward from the southern
end of the lake over the course of the day, carried by southerly winds.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
13
396
397
398
399
400
401
402
403
Synoptic meteorological conditions often work in combination with lake-induced mesoscale
meteorological features to produce the highest ozone concentrations in this region (Dye et al., 1995).
Other coastal regions in the U.S. share related marine/lake enhanced ozone. Modeling by U.S. EPA
(2016) projected that 9 of the 12 counties in the eastern U.S. that will not attain the 2015 ozone NAAQS
by 2025 are located on coastlines. The coastal high ozone phenomenon has been documented for Lake
Michigan, the Gulf of Mexico/Houston, the Chesapeake Bay/Maryland, and Lake Erie/southwestern
Ontario (Lyons and Cole, 1976, Dye et al. 1995, Foley et al. 2011, Goldberg et al. 2014, Loughner et al
2014, Cleary et al. 2015)
404
405
406
407
408
409
410
411
412
413
414
Historically, the highest ozone concentrations in the Lake Michigan region have been found
along the lakefront (Figure 1.1.2). The sharp temperature gradients between the warm land and cold
lake drives the thermally-driven circulation cell called the “lake breeze” as described above, which runs
approximately perpendicular to the Lake Michigan shoreline (Figure 1.1.3). As this figure shows, the lake
breeze is generally preceded by an early morning land breeze, driven by relatively warm temperatures
over the lake and subsidence over land. The land breeze carries ozone precursors emitted from coastal
urban and industrial areas out over the lake, where they can react in the presence of sunlight to form
ozone. Heating of the land surface in the afternoon leads to a reversal of the flow pattern and the lake
breeze then transports the elevated ozone from over the lake into coastal areas. These mesoscale lake
breeze patterns can overlap with southerly synoptic winds that carry ozone-rich air to impact coastal
sites northward along the lakefront.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
14
415
416
417
418
419
420
Figure 1.1.2. Map of ozone design values and LMOS 2017 ground site locations. Ozone design values (fourth-highest
monitored eight-hour average ozone concentration at that monitor in a three-year period, ppbv) for 2014-2016 for the Lake
Michigan region along with the location of the primary ground sites hosting measurements for the Lake Michigan Ozone
Study 2017 (LMOS 2017).
421
422
423
Figure 1.1.3. Conceptual model of land/lake breeze circulations responsible for enhanced ozone production along the shores
of Lake Michigan (Foley et al., 2011).
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
15
424
1.2. Objectives of LMOS 2017
425
426
427
428
429
430
431
The Lake Michigan Ozone Study 2017 (LMOS 2017) was conducted to address persistent
violations of the ozone NAAQS in the coastal communities along the western shore of Lake Michigan.
LMOS 2017 leveraged joint aircraft, ground-based, ship-based and mobile lab measurements to
significantly expand observations of ozone and its precursors around Lake Michigan during the early
summer of 2017. The study was designed to collect field measurements to address the following
questions related to enhancements in ozone concentrations observed in coastal regions along Lake
Michigan:
432
433
434
1)
What is the relative contribution of inter- and intra-state nitrogen oxides (NOx) and
volatile organic compound (VOC) emissions and emissions sources on ozone production
rates along Lake Michigan?
435
2)
To what extent do lake breeze circulations affect ozone production?
436
3)
How far inland does ozone rich air penetrate during ozone events?
437
438
4)
What is the spatio-temporal distribution of ozone and its precursors over Lake
Michigan?
439
440
5)
How can remote sensing products (e.g., measurements of nitrogen dioxide, NO2, and
formaldehyde, HCHO) be used to constrain ozone predictions?
441
442
6)
How well do regional models capture ozone production chemistry as assessed through
evaluation of critical measurement indicators?
443
1.3. Overview of the LMOS 2017 Mission
444
445
446
447
448
449
450
451
452
453
454
455
456
LMOS 2017 was conducted from May 22 through June 22, 2017, and was designed to collect
measurements to investigate high ozone episodes in coastal communities along the western shore of
Lake Michigan. Aircraft, ship, mobile lab, and ground-based stations were used to build an extensive
dataset for studying ozone, its precursors, and particulate matter. The LMOS 2017 measurement
strategy was designed to collect onshore and offshore measurements during lake-breeze-driven ozone
exceedance events and to characterize the magnitude and variability of urban emissions on hourly,
daily, and weekly timescales. The LMOS 2017 measurements were designed for use in evaluating
chemical transport model skill in simulating critical aspects of coastal and lake meteorology, oxidant
concentrations, VOC and total reactive nitrogen (NOy) speciation, the spatial and vertical distribution of
ozone precursors, and fine particulate matter (PM2.5) sources. Figure 1.4.1. Schematic showing the
types of measurements involved in LMOS 2017, including two types of aircraft measurements, shipbased measurements, ground-based measurements and mobile platforms. Figure 1.4.1 and Table 1.3.1.
LMOS Science Team PIs show the range of measurements made during LMOS 2017.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
16
457
458
459
Figure 1.4.1. Schematic showing the types of measurements involved in LMOS 2017, including two types of aircraft
measurements, ship-based measurements, ground-based measurements and mobile platforms.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
17
460
Table 1.3.1. LMOS Science Team PIs
Name
Affiliation
LMOS 2017 Role
Zac Adelman
Jassim Al-Saadi
Timothy
Bertram
Greg
Carmichael
Patricia Cleary
Michael Conley
Stephen Conley
Alan Czarnetzki
Angela Dickens
LADCO
NASA Langley Research Center
University of Wisconsin
LADCO Director during post-analysis phase
GeoTASO
Zion Site co-Principle Investigator
University of Iowa
Chemical Forecasting and WRF-Chem Modeling
University of Wisconsin Eau Claire
Scientific Aviation
Scientific Aviation
University of Northern Iowa
Wisconsin Department of Natural
Resources
EPA Region V
NASA Langley Research Center
LADCO
LADCO
Mobile Measurements
Co-owner Scientific Aviation
Co-owner Scientific Aviation
Zion Meteorology Measurements
Ozone Episode Analysis
Marta Fuoco
Laura Judd
Rob Kaleel
Donna Kenski
Mike Majewski
Stephanie Shaw
Charles Stanier
Wisconsin Department of Natural
Resources
University of Maryland Baltimore
County (UMBC)
University of Minnesota
NOAA (now at University of Wisconsin
SSEC)
Electric Power Research Institute (EPRI)
University of Iowa
Betsy Stone
James Szykman
Luke Valin
David Williams
Tim Wagner
University of Iowa
USEPA and NASA Langley
USEPA
USEPA
University of Wisconsin SSEC
Vanderlei
Martins
Dylan Millet
Brad Pierce
Mobile Measurements
GeoTASO
LADCO Director during field campaign
Mobile Measurement, Site Logistics, Data Analysis, IEPA
Liaison
Meteorological Forecasting
AirHARP
Zion Site co-Principle Investigator
Science Director, Flight Planning Team Lead, RAQMS Modeling
Representative of EPRI to LMOS 2017
Zion Site co-Principle Investigator, Zion Representative to
Flight Planning Team
Zion Site co-Principle Investigator
EPA Measurements Lead (Sheboygan, Ceilometers, Pandora)
EPA Measurements
Shipborne Measurements
Sheboygan Meteorology Measurements (SPARC)
461
462
463
464
465
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
18
466
Table 1.3.2. Summary of measurements made during LMOS 2017.
Location
Measurement*
Spaceport
Sheboygan
Zion, IL
Ground Sites
Remote sensing of meteorology
In situ measurements of pollutants
Remote sensing of meteorology
Detailed in situ chemical measurements
Various†
Sheboygan
transect
Lakeshore
region
Lake Michigan
Northeast IL and
Southeast WI
Grafton to
Sheboygan
Routine measurements of ozone
Remote sensing of pollutants and boundary layer
height
In situ measurements of ozone at four locations
Airborne Platforms
Airborne remote sensing of NO2 (GeoTASO)
Airborne remote sensing of clouds (AirHARP)
Airborne in situ profiling of pollutants and
meteorology
Shipboard Platform#
In situ measurements of pollutants and meteorology
Remote sensing of pollutants and boundary layer
height
Mobile Platforms
In situ measurements of pollutants (GMAP)
In situ measurements of ozone and meteorology
Research Institution*
UW-Madison
U.S. EPA ORD
Univ. Northern Iowa
Univ. Iowa, UW-Madison,
Univ. Minnesota
Illinois EPA
U.S. EPA ORD
U.S. EPA ORD
NASA
Univ. Maryland, Baltimore
County
Scientific Aviation
U.S. EPA ORD
U.S. EPA ORD
U.S. EPA Region 5
UW-Eau Claire
467
468
469
*NO2 = nitrogen dioxide, GeoTASO = Geostationary Trace gas and Aerosol Sensor Optimization instrument, AirHARP =
Airborne Hyper Angular Rainbow Polarimeter, GMAP = Geospatial Mapping of Pollutants, UW = University of Wisconsin, EPA
= Environmental Protection Agency, ORD = Office of Research and Development.
470
471
† These measurements were made at Spaceport Sheboygan, Zion, two Wisconsin DNR monitoring locations (Grafton and
Milwaukee SER) and one Illinois EPA monitoring location (Schiller Park).
472
#
473
474
475
476
477
478
479
480
481
482
Aircraft measurements included remote measurements of NO2 and potentially HCHO columns
from the Geostationary Trace gas and Aerosol Sensor Optimization (GeoTASO) spectrometer, and cloud
and aerosol measurements from the Airborne Hyper-Angular Rainbow Polarimeter (AirHARP), which
flew on the NASA King Air aircraft. In situ trace gas and meteorological measurements were made using
instruments onboard a Mooney Aircraft operated by Scientific Aviation. Twenty-one GeoTASO and 22
Scientific Aviation flights were conducted during the LMOS 2017, with most of these flights coordinated
between the NASA and Scientific Aviation aircraft. Flight planning meetings were conducted each
morning during the aircraft deployment to determine the flight plan for the day based on
meteorological and chemical forecasts. Primary flight objectives were to 1) conduct regional scale
coastal surveys along the western shore of Lake Michigan during ozone exceedance events, 2) conduct
Shipboard measurements are still being quality controlled and are not discussed in this initial report
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
19
483
484
485
486
487
488
489
local Chicago, IL, and Milwaukee, WI, flights to characterize the urban NO2 emissions, and 3) conduct
local Sheboygan, WI, and Zion, IL, survey flights to characterize coastal gradients in air pollutants and
meteorological fields. The NASA King Air aircraft flew “raster” patterns to obtain maps of NO2 column
abundances. Scientific Aviation flew low level transects to sample the marine boundary layer along with
in situ spirals to measure vertical profiles at selected waypoints. On most of the deployment days, the
two aircraft flew over the same area at the same time to provide the best overlap between the two
types of measurements.
490
491
492
493
494
Ground-based measurements included enhanced in situ and remote sensing measurements,
and two mobile measurement platforms. The enhanced in situ measurements were made at two
primary ground sites along the western shore of Lake Michigan: Spaceport Sheboygan in downtown
Sheboygan, WI, and the Illinois Environmental Protection Agency (IEPA) Zion monitoring site (AQS ID 17097-1007) just south of the Wisconsin-Illinois state line (Figure 1.1.2 and
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
20
495
496
497
498
499
500
501
502
503
504
505
506
Table 1.3.2). These two sites are each located near the monitors that consistently measure the highest
ozone concentrations in the region. Measurements at the Sheboygan site were focused on vertically
resolved coastal winds, aerosols, temperature, water vapor and surface measurements of ozone, ozone
spatial patterns, NOx, NOy, and formaldehyde (a marker of oxidation chemistry). Measurements at the
Zion site focused on vertically resolved coastal winds, temperature, water vapor, aerosol optical depth
and surface measurements of ozone, select ozone precursors, NO x, VOCs, oxygenated VOCs, nitric acid,
hydrogen peroxide, and aerosol characterization. Additional ground based remote sensing instruments
were deployed at the Wisconsin Department of Natural Resources (WDNR) Grafton (AQS ID 55-0890008) and Milwaukee SER (AQS ID 55-079-0026) sites, as well as at IEPA’s Schiller Park monitoring site
(AQS ID 17-031-3103). Mobile ground based measurements of ozone and ozone precursors were made
along the shoreline and along inland transects near Zion, IL and Sheboygan, WI to investigate the extent
that high ozone penetrates inland during ozone episodes.
507
508
509
510
511
512
513
514
515
The LMOS 2017 data are being used in conjunction with a number of different meteorological
and photochemical models (Table 1.3.3). The LMOS 2017 data are being used to evaluate the skill of the
models in an extended manner that is not possible when just traditional ground-based chemical and
meteorological stations are used. The LMOS 2017 dataset contains chemical species and meteorology
fields that are not routinely measured, observed vertical profiles of ozone and NO2, and measurements
of the structures of the lake boundary layer and lake breeze. Furthermore, LMOS 2017 produced a rare
dataset of chemical measurements over the lake, instead of just on land. Models are also being used to
help interpret the measurement dataset, through visualization of air mass trajectories and spatiotemporal patterns of air pollutants, and through emissions sensitivity experiments.
516
517
518
519
520
521
522
The LMOS team is working toward providing best-practices recommendations on applying
models for simulating high ozone concentrations along the shores of Lake Michigan. These
recommendations will include details on the model configurations, temporal and spatial resolutions
needed to simulate the dynamical and chemical drivers of high ozone. The LMOS 2017 team is also
studying sensitivity of ozone model skill to biogenic emissions. The LMOS 2017 project was designed in
anticipation of long-term use of the resulting data products for future model evaluation and model
development projects.
523
Table 1.3.3. Models being evaluated using LMOS 2017 data.
Model
WRF-CHEM
NAM-CMAQ
WRF-Met
CMAQ
Research Institution
Univ. Iowa
NOAA
NOAA
UW-Madison
524
525
1.4. Meteorology during LMOS 2017
526
527
The measurements collected during LMOS 2017 were designed to capture the key features of
the meteorological and chemical conditions during the study period. This section begins with a
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
21
528
529
530
531
532
comparison of the observed monthly mean meteorological conditions to historic climatological trends in
the region. These comparisons are significant because they frame the meteorological conditions of
LMOS 2017 in the context of the historically observed connections between weather and air quality in
the region. The prevailing synoptic patterns that occurred during each week of LMOS 2017 are then
presented to lend insight into the conduciveness of the meteorological conditions for ozone formation.
533
534
535
536
537
538
539
540
541
542
543
544
The NOAA National Centers for Environmental Information (NCEI) Synoptic Discussion for May
2017 (https://www.ncdc.noaa.gov/sotc/synoptic/201705) states that the weather over the contiguous
United States (CONUS) was dominated by a highly variable jet stream circulation resulting in abovenormal precipitation across much of the eastern half of the country and cooler than normal
temperatures across much of the central U.S. (Figure 1.4.1). Precipitation was above average over
Wisconsin and maximum surface temperatures were below average over Wisconsin and NE Illinois
during May 2017. During June 2017 (https://www.ncdc.noaa.gov/sotc/synoptic/201706) an upper-level
ridge dominated the western half of the U.S. resulting in warmer- and drier-than-normal weather in the
west. An upper-level trough over the eastern half of the country resulted in cooler- and wetter-thannormal weather in the eastern U.S. (Figure 1.4.2). Precipitation was above average over Wisconsin and
below average over Illinois during June 2017. Temperatures were near average in northern Wisconsin
and above average in southern Wisconsin and Northern Illinois during June 2017.
545
546
547
548
549
550
551
552
Weekly (Sunday to Saturday) composite 850 millibar (mb) heights, 2 meter (m) temperatures, 2
m relative humidity, and 2 m temperature anomaly maps from the North American Regional Reanalysis
(NARR) (Mesinger, et al, 2006) data are presented here to summarize the synoptic conditions of the
LMOS 2017 study period. These images were generated using the web tools provided by the
NOAA/ESRL Physical Science Division web site (http://www.esrl.noaa.gov/psd/). The 2 m temperature
anomalies are relative to the 1979 to 2017 mean. Figures 1.4.3 through 1.4.7 show the composite
synoptic conditions for each of the 5 weeks of the study, beginning on Sunday, May 21, 2017, and
ending on Saturday, June 24, 2017.
553
554
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
22
555
556
557
558
559
Figure 1.4.1. National maps of departure from climatological means of precipitation and temperature for May 2017. Total
Precipitation (left) and Maximum Temperature (right) Percentiles relative to the 1895-2017 ranking period from the NOAA
National Centers for Environmental Information Synoptic Discussion for May 2017
(https://www.ncdc.noaa.gov/sotc/synoptic/201705).
560
561
Figure 1.4.2. Same as Figure 1.4.1 for June 2017 (https://www.ncdc.noaa.gov/sotc/synoptic/201706)
562
563
564
565
566
567
568
569
Figure 1.4.3 shows that a large amplitude trough in the 850 mb heights over the Northern Plains
resulted in cooler than normal temperatures over the Central Plains during the first week of LMOS 2017
(May 21-27, 2017). Mean 2 m temperatures ranged from 291K (64oF) in northern Illinois to 288K (59oF)
in southern and central Wisconsin. The mean 2 m relative humidity ranged from 80-85% over
Wisconsin, Michigan and the Great Lakes with slightly lower (70-80%) relative humidity over Illinois and
Indiana. The 2 m temperature anomaly map shows that Wisconsin, Illinois, and Indiana had near normal
to slightly below normal temperatures, while temperatures over Lake Michigan were up to 2K (4oF)
below normal.
570
571
572
573
574
575
576
Figure 1.4.4 shows that during the second week of the study (May 28-June 03, 2017) the trough
over the Northern Plains moved to the NE resulting in near normal 2 m temperatures and somewhat
dryer air over most of the Midwest. Mean 2 m relative humidity ranged from 55-65% over Illinois to 7075% over Wisconsin with somewhat drier air over southern Lake Michigan. Mean 2 m temperatures
rose to 294K (69oF) in central Illinois but remained near 288K (59oF) in central Wisconsin, resulting in
sharper temperature gradients and increased (3-5 m/s) mean westerly 2 m wind speeds (not shown)
over much of Wisconsin.
577
578
579
580
581
582
583
As shown in Figure 1.4.5, a weak ridge dominated the Central Plains during the third week of
LMOS 2017 (June 04-10, 2017) resulting in above average temperatures (294K or 69oF) over southern
and central Wisconsin. Figure 1.4.6 shows that during the fourth week of the study (June 11-17, 2017) a
large amplitude ridge formed over the SE U.S. leading to a broad region of above average temperatures
over the central and eastern U.S. This ridge resulted in significantly (4-5K or 8-10oF) above average
temperatures over Iowa, Wisconsin, Michigan, and northern Illinois. Mean 2 m temperatures reached
300K (80oF) in central and northern Illinois and 297K (75oF) in southern Wisconsin. The warmer
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
23
584
585
temperatures and higher (80%) 2 m relative humidity in NE Wisconsin resulted in frequent afternoon
cloudiness and occasional convective precipitation over eastern Wisconsin during this period.
586
587
588
589
590
591
Figure 1.4.7 shows that the final week of the campaign (June 18-24, 2017) returned to an 850
mb geopotential height pattern that was similar to the second week of the study, resulting in below
normal temperatures over Minnesota and NW Wisconsin. These cooler temperatures north of the study
region, coupled to near normal temperatures in Illinois, produced strong surface temperature gradients
and somewhat stronger (2-4m/s) westerly 2 m winds over eastern Wisconsin and east-central Illinois
(not shown).
592
593
594
595
Figure 1.4.3. CONUS meteorological overview for the week starting May 21. 850mb geopotential height (upper left, m), 2m
air temperature (upper right, K), 2m relative humidity (lower left, %), and 2m air temperature anomaly (lower right, K)
during May 21-27, 2017 (Week 01 of LMOS 2017)
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
24
596
597
598
599
Figure 1.4.4. CONUS meteorological overview for the week starting May 28.850mb geopotential height (upper left, m), 2m
air temperature (upper right, K), 2m relative humidity (lower left, %), and 2m air temperature anomaly (lower right, K)
during May 28-June 03, 2017 (Week 02 of LMOS 2017).
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
25
600
601
602
603
Figure 1.4.5. CONUS meteorological overview for the week starting June 4. 850mb geopotential height (upper left, m), 2m air
temperature (upper right, K), 2m relative humidity (lower left, %), and 2m air temperature anomaly (lower right, K) during
June 04-10, 2017 (Week 03 of LMOS 2017).
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
26
604
605
606
607
Figure 1.4.6. CONUS meteorological overview for the week starting June 11. 850mb geopotential height (upper left, m), 2m
air temperature (upper right, K), 2m relative humidity (lower left, %), and 2m air temperature anomaly (lower right, K)
during June 11-17, 2017 (Week 04 of LMOS 2017).
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
27
608
609
610
611
Figure 1.4.7. CONUS meteorological overview for the week starting June 18. 850mb geopotential height (upper left, m), 2m
air temperature (upper right, K), 2m relative humidity (lower left, %), and 2m air temperature anomaly (lower right, K)
during June 18-24, 2017 (Week 05 of LMOS 2017).
612
1.5. Lakeshore Ozone During LMOS 2017
613
614
615
616
617
618
619
620
621
622
623
This section briefly summarizes the ozone concentrations observed during LMOS 2017. The
study period began with fairly low lakeshore hourly ozone concentrations (< 60 ppbv) through the end
of May. June was more promising for investigating elevated ozone at the lakeshore with three distinct
high ozone periods occurring before the end of the study on June 22. Figure 1.5.1 shows timeseries of
surface ozone during LMOS 2017 at the three lakeshore monitors (shown in blue and purple) that
typically measure the highest ozone concentrations on the western shore of the lake. The figure also
shows concentrations at nearby, inland monitors (shown in green),which typically have lower ozone
concentrations. Peak lakeshore ozone concentrations remained below 60 ppbv until June 2, 2017, when
all five of these monitors measured hourly ozone levels well above 70 ppbv. As shown in Figure 1.5.2,
measured maximum daily eight hour average (MDA8) ozone exceeded 70 ppbv at most of the lakeshore
monitors from Northern Indiana to NE Wisconsin.
624
625
Figure 1.5.1 shows that on June 4 the ozone monitors near the IL-WI border measured hourly
ozone concentrations above 70 ppbv. A second, multiday period of high lakeshore ozone concentrations
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
28
626
627
628
629
occurred at these monitors between June 10-12. The Zion, IL; Chiwaukee, WI; and Kohler-Andrae, WI,
monitors all observed hourly ozone concentrations above 70 ppbv on each of these days. The
Chiwaukee, WI, monitor exceeded 70 ppbv MDA8 on June 12 and the Kohler-Andrae, WI, monitor
exceeded 70 ppbv MDA8 on June 11 and 12.
630
631
632
633
634
A third, multiday period of high lakeshore ozone concentrations occurred between June 14-16
with most of the sites shown in Figure 1.5.1 measuring hourly ozone above 70 ppbv on each of these
days. On June 15 only the Zion, IL, and Chiwaukee, WI, monitors measured hourly ozone concentrations
above 70 ppbv. The Zion, IL, and Chiwaukee, WI monitors exceeded 70 ppbv MDA8 on June 15 and the
Kohler-Andrae, WI, monitor exceeded 70 ppbv MDA8 on June 16.
635
636
637
638
639
640
Figure 1.5.1. Time series of one-hour surface ozone concentration (ppbv) during May 22 – June 22, 2017. Upper panel shows
time series at Chiwaukee Praire (CP), Kenosha Water Tower (WT), and Zion (Zion). Lower panel shows time series at
Sheboygan Kohler-Andrae (KA) and Haven (Haven) sites. The lakeshore monitors are shown in blue and purple, and the
inland (3.2 to 3.6 miles) monitors are shown in green.
641
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
29
MDA8 Ozone on June 2, 2017
642
643
Figure 1.5.2. Maximum daily eight-hour average (MDA8) ozone mixing ratios (ppbv) on June 2, 2017
644
645
646
647
During some of these ozone episodes (e.g., June 2nd in both areas and June 11th and 16th in
Sheboygan), high ozone concentrations reached the inland monitors, located 3.2 to 3.6 miles from the
lakeshore. On other days (e.g., June 10th and June 12th in Sheboygan and June 15th in Kenosha), only the
lakeshore monitors were impacted by high ozone. On these days, concentrations at the inland monitors
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
30
648
649
remained low. These figures demonstrate that the LMOS 2017 campaign included a variety of types of
ozone episodes.
650
651
652
653
The MDA8 ozone concentrations at Chiwaukee and at Kohler Andrae were fairly typical,
compared to values for May 15-July 15 from other years, as shown in Figure 1.5.3. The highest levels of
ozone (MDA8 exceeding 85 ppb) are rare events, with typical years having either zero or one day in that
time period at over 85 ppb. LMOS did not capture such events at Chiwaukee or at Kohler Andrae.
654
655
Figure 1.5.3. Ozone concentrations during LMOS were representative of climatological values.
656
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
31
657
658
Section 2.
Summary of LMOS Observations and Key Results
659
660
661
662
663
664
665
666
667
This section presents an overview of the observations collected during LMOS 2017 and a
summary of the key results from the initial analysis of the data from the different measurement
platforms. The results presented here are preliminary and are largely based on the field data collected
during the study that are still undergoing quality assurance checks, calibration adjustments, and zero
adjustments. Additional results will become available as the intercomparison of data from the different
measurement platforms and research groups evolves, as the observational datasets from the study are
harmonized, and as model-observation comparisons are completed. Data collected during LMOS is
archived at the NASA Langley Research Center (LaRC) Airborne Science Data for Atmospheric
Composition (https://www-air.larc.nasa.gov/missions/lmos/).
668
2.1. Spaceport Sheboygan, WI Measurements
669
2.1.1. Overview
670
671
672
673
674
675
676
677
To investigate the lake breeze dynamics and local production of ozone during LMOS 2017 a suite
of ground-based remote sensing and in situ instruments were deployed at Spaceport Sheboygan
(43°44'46.55"N, 87°42'29.09"W) (Figure 2.1.1) in Sheboygan, WI. The Spaceport Sheboygan ground site
is located approximately 0.25 km from the Lake Michigan shoreline, approximately 3.5 km north of the
Edgewater coal-fired power station, and approximately 8.5 km north of the Wisconsin Department of
Natural Resources (WDNR) Sheboygan Kohler-Andrae (KA) air quality monitor. This site served as the
main LMOS 2017 research station for the UW-Madison Space Science and Engineering (SSEC) and the
U.S. EPA Office of Research and Development research groups.
678
679
680
681
682
683
684
685
686
687
688
689
690
Understanding the temporal evolution of photochemical regimes for ozone production at the local
scale via direct observations can provide important insights for model performance as the MDA8 O3
concentrations are sensitive to the changes in NOx and VOC precursors throughout the day. The EPA
measurement suite at Spaceport Sheboygan focused on a limited set of trace gas measurements (Table
2.1.1). These measurements can be used both to provide insight into the local photochemical
conditions during LMOS 2017 through the use of chemical indicators and to evaluate chemical transport
model simulations of the study period. The ratio of formaldehyde (HCHO) and NO2 measurements can
be used to assess the local photochemical regime driving O3 production. A HCHO/NO2 ratio < 0.8
indicates that the ozone production chemistry is radical limited, a ratio > 1.8 indicates NOx limited
conditions, and ratios between indicate a transition state between the two conditions (Sillman, 1995;
Sillman 1998; Tonnesen and Dennis, 2000At the same time, SSEC collected an extensive suite of
meteorological measurements. The combined set of EPA and SSEC measurements will be used to assess
air mass changes and the associated ozone production conditions within the air masses.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
32
691
2.1.2. Methods
692
693
694
695
696
697
698
699
The SSEC Portable Atmospheric Research Center (SPARC, Wagner et al. 2016) is a movable
meteorological research laboratory that hosts in situ and remote observations of conditions at and
above the surface. SPARC observations include continuous ground-based temperature and water vapor
profiles from the SSEC Atmospheric Emitted Radiance Interferometer (AERI) (Knuteson et al. 2004a,b),
aerosol backscatter measurements from the SSEC High Spectral Resolution LIDAR (HSRL) (Shipley et
al,1983), and Doppler Lidar wind measurements. A data logger and other instrumentation feeds are
synchronized to a GPS receiver in order to coordinate internal clocks to facilitate precise
intercomparisons of the data between instruments.
700
220m
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
Figure 2.1.1. Location of the LMOS ground based measurements in Sheboygan, WI. The SPARC trailer was located on the NE
side of SpacePort Sheboygan and the EPA trailer was located on the SW side of Spaceport Sheboygan .
The U.S. EPA research trailer is a mobile research laboratory designed to host a combination of
in situ and remote sensing observations with a primary emphasis on the measurement of key trace gas
species related to ozone chemistry. Table 2.1.1 provides a summary of the EPA measurement suite
deployed in Sheboygan, including the portable 2B-POM ozone measurements that were deployed
further inland to assess ozone gradients. With the exception of the 2B-POM ozone instruments, all
other in situ instruments were housed within the temperature-controlled EPA mobile research
laboratory for the study. The instruments in the EPA mobile lab were calibrated according to
manufacturer’s operation manuals and in accordance with the Federal Reference Method (FRM; 40 CFR
part 50, Appendix F and D). During LMOS 2017 automated zero and span checks were performed every
other night to monitor the validity of the calibration and control for drifts or variations in the span and
zero response. Details regarding generation of span and zero gases, dilution air, inlet materials and
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
33
716
717
placement, filtration of particles, and data acquisition are available upon request and will be included
with final reports.
718
719
720
Table 2.1.1. Summary of EPA Measurements at Sheboygan, WI, during LMOS 2017
Measurement
Measurement
Principle
Model/
Manufacturer
Time
Resolution
Relevant
Reference
O3
Scrubberless Ultraviolet
Photometric (SL-UV)
2B Technology
M211
10-seconds
EQOA-0514215
Ultraviolet Photometric
2B Personnel Ozone
Monitors
1-minute
EQOA–0815–
227
Cavity attenuated phase shift
spectroscopy (CAPS)
Teledyne T500U
10-seconds
EQNA-0514212
O3 Chemiluminescence with
Teledyne T200U
10-seconds
RFNA-1194099
O3 Chemiluminescence with
Teledyne T200U
10-seconds
NA
HCHO
Quantum Cascade Laster (1765
cm-1)
Aerodyne Research
1- seconds
Herndon et
al., 2007
O3/NO2/HCHO
Sun-sky radiances 280-525nm
NASA Goddard
Pandora
Spectrometer
Total columns
every 80-s
Herman et al.,
2009
*WS, WD, T, RH, BP,
Prec.
Various
Vaisala WXT520
with
ultrasonic wind
sensors
NO/NO2/NOx
Molybdenum converter
NOy
external Molybdenum
converter at 10m
column densities
721
* WS-wind speed, WD-wind direction, T-temperature, RH-relative humidity, BP-barometric pressure, Prec.-precipitation
722
723
724
725
In addition to measurements at Spaceport Sheboygan, the EPA placed three ozone monitors
(2B-POMs) at approximate distances of 0.65 km, 3 km, and 6 km from the lakeshore, as shown in Figure
2.1.2. The purpose of these ozone monitors was to investigate potential concentration gradients over
the domain, particularly during high ozone days and lake breeze-driven ozone events.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
34
726
727
728
Figure 2.1.2. Location of the 2B-Personal Ozone Monitors in relation to the lake shore and the main Spaceport Sheboygan
research site.
729
2.1.3. Results and Discussion
730
731
732
733
734
735
736
737
738
739
Over the study period the hourly ozone concentrations measured at the Spaceport Sheboygan
research site reached over 70 ppbv for sustained periods on June 2, 11, 12, and 16 (Figure 2.1.3). Time
series of NO2 and HCHO are also plotted in Figure 2.1.3. The time series of NO2 and HCHO
concentrations show the relative levels of these precursors are different for the high ozone days. Figure
2.1.4 shows the time series of the HCHO/NO2 ratio and wind direction at the Spaceport Sheboygan
research site. The HCHO/NO2 ratio indicates that ozone production during the June 2 and 16 high ozone
events was radical or VOC limited (HCHO/NO2 <0.8). Ozone production on June 11 and 12 was NOx
limited (HCHO/NO2>1.8). The presence of the two different local photochemical regimes driving ozone
production at Spaceport Sheboygan indicates the potential for different causes of the high observed
ozone.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
35
740
741
742
Figure 2.1.3. Time series of O3, NO2 and HCHO mixing ratios as measured at the Spaceport Sheboygan research station during
LMOS.
743
744
745
746
Figure 2.1.4. Time series of the ratio of HCHO/NO2 mixing ratios and wind direction measured at the Spaceport Sheboygan
research station during LMOS. HCHO/NO2 ratios falling below the dark blue line plotted at 0.8 indicates a radical limited
environment, while ratios above the orange line at 1.8 indicate a NO x limited environment.
747
748
749
750
751
752
753
Figure 2.1.5 and 2.1.6 show data collected by instruments within the SPARC trailer on the June 2
and June 11 high ozone days. In addition to producing high ozone at the Spaceport Sheboygan site,
these days also saw ozone exceedances at the Sheboygan KA monitor. In each figure, the panels show
time height cross sections with the Doppler lidar wind profiles overlaid on top of the AERI Optimal
Estimation (AERIoe) (Turner and Löhnert, 2014) retrievals of potential temperature (top) or HSRL aerosol
backscatter (bottom). Times are in UTC and cover the range from 12 midnight to 12 midnight (CDT).
Sunrise is approximately 10 UTC (5:00am CDT).
754
755
756
757
758
759
The temperature profile in Figure 2.1.5 shows that before sunrise on June 2, a very stable
nocturnal boundary layer exists in the lowest 100m, as indicated by temperatures that are cooler at the
surface than aloft. After sunrise, diurnal heating warms the surface and the boundary layer becomes
unstable, and vertical mixing erodes the stable nocturnal boundary layer beginning at around 13 UTC.
Another indication that the boundary layer is deepening is the increase in the depth of aerosol
backscatter seen in the bottom panel at around the same time. Before 15 UTC (10:00am CDT) the
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
36
760
761
762
763
764
765
766
767
768
inversion is mixed out, and rapid increase in boundary layer depth is observed after that time. Shortly
after 17 UTC (noon CDT), the Doppler lidar indicates a shift in the wind direction from westerly to
easterly and then southerly. Since Sheboygan sits on the southern side of a point, flow from the east or
south originates in the lake. As the air sitting over the lake is substantially cooler than the air over land,
the lake breeze generates a new inversion through advection of a shallow layer of cold air onshore.
Note that the winds and temperature inversion are slightly offset in the vertical. This offset is a result of
the 140 m dead zone where the Doppler lidar cannot perform wind measurements coupled with the fact
that the lake breeze inversion builds from the surface upward. The afternoon southerly winds that
extend up to approximately 600 m persist throughout the rest of the day and into evening.
769
770
771
772
The darker colors in the HSRL lidar profile on June 2 indicate very clean (i.e. free of aerosols) air
prior to 21:30 UTC (4:30pm CDT). After 21:30 UTC, air within the lake breeze inversion that is southerly
in origin shows higher aerosol backscatter than the air immediately above the lake breeze inversion and
may indicate polluted air that has been transported over the lake from the south.
773
774
775
776
777
778
779
780
781
Figure 2.1.5. Visualization of boundary layer evolution (winds, temperature, and aerosol backscatter) for June 2 and 3, 2017
at Sheboygan, WI. AERI potential temperature (K, upper panel) and HSRL backscatter (log 10(m-1 sr-1), lower panel) with
Doppler winds during the period from 5 UTC on June 02 to 5 UTC on June 03, 2017 (12 midnight to 12 midnight CDT).
In contrast, on June 11 the AERI temperature retrievals show no evidence of a nocturnal
boundary layer during the early morning hours, strong westerly winds above 400 m during the evening
of June 10 and early morning of June 11, and associated turbulent mixing, likely inhibited the formation
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
37
782
783
784
785
786
787
788
of the stable surface layer. The arrival of the lake breeze at 14:30 UTC (9:30am CDT) is evident in the
AERI temperature profiles as a shallow inversion layer. The Doppler Lidar shows winds switching from
westerly to southwesterly at 17 UTC (12 noon CDT). By 22 UTC (5:00pm CDT), the maritime inversion is
gone. The HSRL Lidar also shows very different behavior on June 11, 2017, with advection of more
polluted (higher aerosol backscatter) air above 400m from the west during the early morning hours and
advection of less polluted (lower aerosol backscatter) air from the southwest during the daytime and
early evening.
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
Figure 2.1.6. Visualization of boundary layer evolution (winds, temperature, and aerosol backscatter) for June 10 and 11,
2017 at Sheboygan, WI. AERI potential temperature (K, upper panel) and HSRL backscatter (log 10(m-1 sr-1), lower panel) with
Doppler winds during the period from 5 UTC on June 10 to 5 UTC on June 11, 2017 (12 midnight to 12 midnight CDT).
Intermittent data capture issues were experienced with all of the 2B-POM instruments. The data
capture rate for each ozone instrument used in this study for a 5-minute average ozone concentration
was 95.8% at Spaceport Sheboygan, 22.2% at Sheboygan Chamber of Commerce, 27.3% at Sheboygan
Fire Department, and 27.3% at Kohler Public Works. A regression of the 5-minute data from the inland
2B-POMs against the Spaceport Sheboygan data is shown in Figure 2.1.7. While the low data capture
rate will limit the utility of this data to investigate the dynamics of the ozone gradients, the regressions
appear to indicate the furthest site inland from the lakeshore, Kohler Public Works, experiences ozone
values 5-6% lower than the close lakeshore site, Spaceport Sheboygan. This difference reflects losses
due to deposition, ozone titration by nitrous oxide (NO), other chemical ozone destruction reactions, or
dilution due to enhanced vertical mixing. When ozone concentrations at Spaceport Sheboygan are
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
38
804
805
806
greater than 70 ppbv, however, the observed concentrations at Sheboygan Fire Department are even
larger. The sampling limitations from these instruments experienced during LMOS 2017 inhibit drawing
additional conclusions from these data.
807
808
SFD
100
KPW
Linear (SFD)
Linear (KPW)
y = 0.9304x + 3.2751
R² = 0.8565
90
80
70
y = 0.9433x + 0.166
R² = 0.7706
ppbv
60
50
40
30
20
10
0
0
20
40
60
80
Spaceport Sheboygan Ozone, ppb
100
809
810
811
Figure 2.1.7. Comparison of 2B-POM ozone from Sheboygan Fire Department, Kohler Public Works, and Spaceport
Sheboygan sites. Linear regression of 5-minute 2B-POM ozone concentrations measured at Sheboygan Fire Department
(SFD) and Kohler Public Works (KPW) against Spaceport Sheboygan (SPS) 5-minute average ozone (all coincidences).
812
2.2. Zion, IL, Measurements
813
2.2.1. Overview
814
815
816
817
818
819
820
The Zion, IL site was selected to focus on gas phase chemistry of the urban plume from the
Chicago-Naperville-Elgin, IL-IN-WI Metropolitan Statistical Area (MSA). By selecting a site at the
northern end of the MSA, we believed we would intercept ozone formation at its early stages during
episodes characterized by synoptic winds from the south. This was supported by preliminary modeling –
sometimes the site would represent relatively fresh emissions, while at other times air would be
moderately processed through oxidative aging and ozone formation, as indicated by H2O2/HNO3 and
HCHO/NO2 ratios from photochemical grid models and remote sensing. Sampling this variability, and
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
39
821
822
823
824
825
826
testing the extent to which models reproduce this variability, was part of the study design. Gas
chemistry was sampled together with a number of supporting measurements to provide meteorological
context and allow source apportionment (through particulate matter source apportionment using
Positive Matrix Factorization, PMF). Finally, the site was far enough to the northeast of O’Hare Airport
that in situ profiles by light aircraft could be performed above the site without too many air traffic
control concerns.
827
828
829
830
Site access, available space and power, and a long-term ozone record were favorable at the Zion,
IL site. Furthermore, at 0.95 km from the lake with no major sources between the monitoring site and
the lake, the Zion site was favorable for capturing the lake breeze and for sampling air “fresh” off the
lake.
831
832
833
834
Continuous or semi-continuous measurements of ozone, NOx, nitric acid, hydrogen peroxide, and
many volatile organic compounds (VOC) and oxygenated volatile organic compounds (OVOC) were
conducted. Some prominent VOC and OVOC compounds measured included benzene, toluene,
isoprene, monoterpenes, methyl vinyl ketone, and methacrolein.
835
836
837
838
839
840
841
842
Volatile organic compounds (VOCs) play several key roles in the atmosphere: their oxidation leads
to the formation of ozone and organic aerosol; they affect the nitrogen cycle by interacting with
atmospheric NOx; and they modulate the atmosphere’s oxidizing capacity and therefore the lifetimes of
greenhouse gases and other pollutants. VOC measurements were performed at Zion during LMOS 2017
with the aim of understanding the role of different VOC species for ozone formation, determining the
importance of biogenic versus anthropogenic emissions for VOC reactivity in this region, and quantifying
indicator species that can be used to diagnose photochemical age and VOC versus NOx limitation for
ozone formation.
843
844
845
846
847
848
Over the past two decades, a wide array of observation-based approaches has been utilized to
study the impact of changing VOC and NOx emissions on ozone production rates and efficiency. During
LMOS 2017 the experimental approach to investigate this area focused on two areas: 1) comparison of
radical production rates (Q) with radical loss rates to NOx (LN) as an indicator of VOC vs. NOx limited
regimes (2005), and 2) the direct use of hydrogen peroxide (H2O2) and nitric acid (HNO3) as indicators for
VOC and NOx sensitivity in ozone production (1995).
849
850
851
852
853
854
855
856
857
In the first approach, radical production was calculated using the current version of the Master
Chemical Mechanism (http://mcm.leeds.ac.uk/MCM) in a box model framework constrained by
observations collected during LMOS 2017. In coastal Lake Michigan during summer, most of the radical
loss is likely to NOx through the reaction of nitrogen dioxide with the hydroxyl radical (OH + NO2). This
hypothesis is being tested with in situ measurements from the Zion, IL site, including biogenic
hydrocarbons and a variety of N-containing oxidation products. To complement the chemical transport
modeling analyses, the ratio LN/Q (radical loss to NOx versus total radical loss rate) are being examined
to provide a direct observational constraint on VOC and NOx limitations to ozone production. A LN/Q
ratio > 0.5 suggests that ozone production is limited by the availability of VOCs.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
40
858
859
860
861
862
863
864
For the second approach, chemical indicators of ozone production sensitivity (e.g., H2O2/HNO3)
are being investigated. Specifically, Sillman suggested that the following chemical ratios indicate
hydrocarbon-sensitive ozone chemistry: HCHO/NOy < 0.28, H2O2/HNO3 < 0.4, and O3/NOz < 7. The
combination of instruments deployed at Zion, IL, during LMOS 2017 enable the assessment of these
indicator ratios in relation to the radical-based approaches discussed above. The simplicity of the
indicator ratio technique makes it a promising tool for assessing VOC and NOx sensitivity if it is proven to
be reliable in characterizing the ozone chemistry in coastal Lake Michigan air masses.
865
2.2.2. Methods
866
Concentration and Bulk Composition of PM2.5
867
868
869
Fine particulate matter (PM2.5) was collected and is undergoing chemical analysis to facilitate
characterization of air mass characteristics at Zion during polluted and non-polluted conditions, and to
provide input data for receptor-based source apportionment (Positive Matrix Factorization).
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
PM2.5 was collected during LMOS 2017 using two medium-volume air samplers (3000B, URG Corp.,
Chapel Hill, NC) for daytime (8:00AM-8:00PM CDT) and nighttime (8:30PM-7:30AM CDT). PM2.5 mass
was determined gravimetrically using a high-precision balance (Mettler-Toledo XP26). Elemental and
organic carbon (EC and OC) were measured by thermal-optical analysis following the ACE-Asia base case
protocol (Schauer et al., 2003). Water soluble cations and anions were determined by water-extraction
and ion exchange chromatography with suppressed conductivity detection as described by Jayarathne et
al. (2014). Planned future analysis includes quantitation of organic species via Gas Chromatography
Mass Spectrometry (GC-MS), organosulfates via Liquid Chromatography Mass Spectrometry (LC-MS),
and metals using Inductively Coupled Plasma Mass Spectrometry (ICP-MS). Once measured, these
species will be used with positive matrix factorization in order to determine sources contributions to
fine particulate matter in Zion, IL, with an emphasis on understanding the distribution of
primary/secondary and biogenic/anthropogenic sources. PMF analyzes the co-variation of chemical
species over time and identifies factors based on a constrained weighted least-squares matrix. Data
inputs include detailed bulk chemical and speciation measurements and measurement uncertainties and
will include organic molecular markers of primary sources, SOA tracers, inorganic ions, and metals. The
model outputs factors and their relative contributions to ambient measurements that will be
interpreted as aerosol sources using elemental and molecular tracers and regional source
characteristics, as done previously by Stone’s group for urban and rural sites.
888
889
Observation-based approach for determining the sensitivity of O3 production rate and efficiency to
NOx and VOCs.
890
891
892
893
894
895
Hydrogen peroxide (H2O2) and nitric acid (HNO3) mixing ratios were measured by the Bertram
group (U. Wisconsin), using chemical ionization time-of-flight mass spectrometry [Bertram et al., 2011]
utilizing O2- and I- reagent ion chemistry (Figure 2.2.1). H2O2 and HNO3 sensitivities and inlet transmission
were determined in the field through standard additions of isotopologues of H2O2 and HNO3. For LMOS
2017, we mounted the detection axis of the mass spectrometer to the roof of the sampling trailer, to
avoid the use of an extensive inlet and associated sampling losses for both molecules. The primary focus
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
41
896
897
during LMOS 2017 was H2O2 and HNO3, however the TOF mass analyzer measured a broader distribution
of molecules beyond H2O2 and HNO3 (e.g., N2O5, ClNO2, select organic acids).
Figure 2.2.1. Schematic of the chemical ionization source and four
stage (S1-S4) differentially pumped interface coupled to the time-offlight mass spectrometer.
898
899
900
High-resolution proton-transfer time-of-flight mass spectrometry VOC measurements
901
902
903
904
905
906
907
908
VOC measurements by Proton Transfer Reaction Quadrupole Interface Time-of-Flight Mass
Spectrometry (PTR-QiTOF, Ionicon Analytik, GmbH) were made at 1-Hz time resolution from the roof of
the Zion trailer using a 0.5” OD / 0.375” ID PFA Teflon inlet line that was wrapped and heated to ~50°C
to minimize VOC-wall interactions and avoid water condensation inside the air-conditioned laboratory.
All downstream wetted sampling surfaces (including valves) were likewise composed of PFA Teflon. The
sampling line was equipped with a 10 m filter that was replaced every ~2-3 days. A rotary vane pump
(1023 Series, Gast Manufacturing Inc.) was used for sampling, yielding flow rates of approximately 40
L/min with an inlet residence time of approximately 3 s.
909
910
911
912
913
914
915
916
917
Measurement blanks (5 minutes) were performed hourly using zero air generated by passing
ambient air through a Pt-bead catalyst (3.2 mm diameter; Sigma Aldrich) heated to 400°C. Calibration
curves (4-point) were performed daily for a set of 28 VOCs by dynamic dilution of certified,
gravimetrically-prepared, ppm-level compressed standards (Apel-Reimer Environmental Inc.).
Instrumental settings were selected for optimal combination of high sensitivity, high mass resolution,
and low humidity-dependence. The time-of-flight analyzer was operated over an m/z range of 0-337
(extraction period 32.5 s), with the mass axis calibrated continuously via addition of a diiodobenzene
internal standard. Peak fitting and integration was performed with the PTR-MS VIEWER 3.2.8. postprocessing software. All subsequent data processing was performed using a custom set of R routines.
918
919
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
42
920
Meteorological Observations
921
922
923
924
925
926
927
928
929
930
A profiling microwave radiometer and acoustic wind profiler, or sodar, from the University of
Northern Iowa were stationed at the Zion field site. Microwave emissions by atmospheric oxygen at any
altitude are proportional to local temperature and density of oxygen, allowing for retrieval of the
temperature vertical profile. Microwave emissions by water vapor are proportional to vapor density
and temperature, allowing for retrieval of water vapor profiles. The echo from acoustic pulses emitted
into the atmosphere will experience a frequency shift that is directly proportional to the motion of the
air. Radiometer measurements extend upward to 10 km, while the sodar determines wind profiles
every 10 m from 30 to 200 m above ground level (AGL). Temperature and water vapor retrievals from
the radiometer and wind measurements from the sodar were used to characterize the structure of the
atmospheric boundary layer and the development of the lake breeze during LMOS 2017.
931
2.2.3. Results and Discussion
932
933
934
935
936
937
An overview of hourly averaged ozone, NOx, and PM2.5 measurements from the Zion, IL, site is
shown in Figure 2.2.2 and in Figure 2.2.3. Ozone exceeded 70 ppbv on June 2, 4, 10, 11, 12, 14, 15 and
16. The onset of the largest ozone episode at Zion (June 2) was correlated with increases in NOx, but the
peak ozone on June 2 appeared depleted in NOx. PM2.5 on June 2 behaved qualitatively similar to NOx –
with an increase at the onset of the ozone increase – but then with lower PM2.5 at the time of the peak
ozone. During the period of June 11-17, PM2.5 and ozone concentrations peak at the same time of day.
938
939
940
Figure 2.2.2. Hourly-averaged O3 and NOx measurements at the Zion site.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
43
1 hr Average
Ozone (ppb)
20
90
18
80
16
70
14
60
12
50
10
40
8
30
6
20
4
10
2
0
5/22 5/24 5/26 5/28 5/30
941
PM2.5
6/1
6/3
6/5
6/7
6/9
Ppbv
Ozone
100
0
6/11 6/13 6/15 6/17 6/19 6/21 6/23
Date (UTC)
942
Figure 2.2.3. Hourly-averaged ozone and estimated PM 2.5 at the Zion site.
943
944
945
946
947
The wind rose of hourly wind measurements collected by IL EPA at the Zion, IL site is shown in Figure
2.2.4, and can be contrasted to the ozone conditional probability function in Figure 2.2.5. Winds from
the SE were much more likely (0.15-0.2 fraction) to have O3 > 70 ppbv. Low ozone (blue) occurred
during weak westerly winds, likely due to weak offshore flow. High ozone corresponded to high wind
speeds from the SE.
948
949
950
951
Figure 2.2.4. Zion wind rose for the duration of the study, based on 1-min averages. The colors show the wind speed and the
distance from the center shows the frequency of winds from that direction. The feature from the southeast is mainly due to
the lake breeze effect at Zion.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
44
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
Figure 2.2.5. Left: Ozone vs. wind direction and speed and Zion, IL. Ozone conditional probability function, showing the
fraction of 1-min ozone measurements above 70 ppbv as a function of wind direction. Right: ozone (color scale in ppm) vs.
wind direction vs. wind speed (radial position in m/s).
At Zion during LMOS 2017, PM2.5 mass averaged 5.1 µg m-3, 12-hour averaged concentrations
ranged from 0.7 to 13 µg m-3 and hourly PM2.5 averaged between ~0.5 and ~16 µg m-3. A summary of
PM2.5 mass concentrations and its chemical constituents are shown in Table 2.2.1. The most significant
contributor to PM2.5 mass was organic matter (OM, calculated as 1.8 x OC), followed by the secondary
ions sulfate, nitrate, and ammonium. Figure 2.2.6 shows that the PM2.5 concentrations were highest
during the June 2 ozone event and were also elevated during the June 10-17 elevated ozone period
compared to non-event periods; however, the relative composition of PM was generally consistent in
terms of ion composition. The average OC:EC ratio was 12.25, suggesting a large contribution from
secondary organic aerosol. Notably, the OC:EC ratio was substantially lower (at 6.9) during the June 2
ozone event. The higher EC measurements on this day indicate a larger influence of combustion
emissions. Future analyses will focus on characterizing organic species, particularly source tracers, to
better define the sources of PM2.5.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
45
Table 2.2.1. Summary of ambient concentrations of PM2.5 organic and elemental carbon (EC and OC) and inorganic
ions.
Component Range (µg m-3)
PM2.5
0.69 – 12.94
EC
<0.05 – 0.78
OC
<0.5 – 4.73
OM
<0.9 – 8.52
Sodium
<0.02 – 0.12
Potassium
<0.004 – 0.10
Magnesium <0.001 – 0.05
Calcium
<0.02 – 0.19
Ammonium <0.003 – 0.83
Fluoride
<0.003 – 0.03
Nitrate
<0.04 – 0.80
Sulfate
<0.03 – 3.05
Mean ± SD (µg m-3)
5.08 ± 2.9
0.22 ± 0.2
1.79 ± 1.0
3.23 ± 1.8
0.04 ± 0.03
0.04 ± 0.03
0.01 ± 0.01
0.06 ± 0.05
0.31 ± 0.18
0.01 ± 0.01
0.19 ± 0.2
0.82 ± 0.5
% PM2.5
2.5
37.1
66.7
0.3
0.6
0.3
0.9
6.6
0.3
3.0
18.3
Concentration ( g m -3)
14
EC
OM (OC x 1.8)
Calcium
Ammonium
Nitrate
Sulfate
PM2.5
12
10
8
6
4
2
0
0.5
O3 Event
06/02 AM –
06/03 AM
969
970
971
972
973
1.0
O3 Period
06/10 AM –
06/17 AM
1.5
Non O3
Events
Figure 2.2.6. The concentration of major bulk components of PM2.5 during the high ozone event, high ozone period, and
background. Na+, K+, Mg2+, and F- were all below 0.1 μg/m 3.
Figure 2.2.7 shows the time series of H2O2 and HNO3 mixing ratios measured at Zion. Typical
midday HNO3 mixing ratios ranged from 1-8 ppbv exhibiting a strong diel profile. Typical midday H2O2
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
46
974
975
976
977
978
979
980
981
982
983
984
985
986
987
mixing ratios ranged from 0.5 – 8 ppbv and were strongly correlated with isoprene mixing ratios, where
the largest concentrations were observed following leaf-out in early June. In contrast to HNO3 diel
profiles, the H2O2 diel profiles were much more muted, perhaps indicating lower dry deposition rates.
Preliminary analysis of HNO3 and H2O2 mixing ratios highlight a complex sampling environment, where
the lake breeze during high O3 events is associated with low H2O2/HNO3 ratios, indicating VOC-limited O3
production (i.e., below ratio-based indicators such as Sillman’s hydrocarbon sensitive region of
H2O2/HNO3 < 0.4). In contrast, the regional background is characterized by elevated H2O2/HNO3 ratios,
suggestive of NOx-limited O3 production. Ozone plumes arriving at Zion in day-time lake breezes were
generally VOC-limited according to the indicator ratios. This was the case even late in June when
biogenic VOCs were present in higher concentrations, and thus contributing a source to H2O2. This
observation is consistent with plumes enriched in NOx and O3, depleted of hydrocarbons on a relative
basis, coming ashore from the lake during some high O3 hours. Ongoing work will focus on the role of
atmospheric mixing at the lake/land boundary between VOC-limited (urban plume) and NOx-limited
(regional background) air.
988
989
990
Figure 2.2.7. Time series of HNO3 (top) and H2O2 (bottom) mixing ratios as measured at the Zion ground station during LMOS.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
47
991
992
993
994
995
996
997
998
999
Analysis of the PTR-QiTOF VOC measurements, including application of finalized humidity
corrections and time-resolved blank and calibration factors, is ongoing. Figure 2.2.8 shows a preliminary
time series for observations of isoprene, methyl vinyl ketone (MVK) plus methacrolein (MACR),
monoterpenes, and toluene. MVR and MACR are prominent oxidation products of isoprene. Mixing
ratios of isoprene and its oxidation products MVK+MACR were low during the first half of LMOS 2017,
and significantly higher during the second half. Anthropogenic species such as toluene show the reverse
trend, with more elevated concentrations seen early in the study. Future analyses will focus on
evaluating the ability of chemical transport models to simulate the observed patterns of VOC
abundance, reactivity, and photochemical age for the LMOS 2017 period.
1000
1001
1002
1003
1004
1005
1006
1007
1008
Figure 2.2.8. Measured mixing ratios of isoprene, MVK+MACR, monoterpenes, and toluene at the Zion field site during
LMOS-2017. June 1 is Julian day 152.
By examining Zion time series, particularly those in Figure 2.2.8, we identify a transition
between an initial period of low biogenic influence (isoprene, MVK, and MACR are low initially) and
higher anthropogenic influence (toluene and other anthropogenic VOCs, not shown). A sharp transition
date is not evident, but the first large isoprene peak is on Julian date 155 (June 4) and the last large
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
48
1009
1010
nocturnal toluene peak is on Julian data 159 (June 8). Peroxide and nitric acid concentrations (Figure
2.2.7) generally increase along with the biogenic VOCs.
1011
1012
1013
1014
1015
1016
1017
1018
The highest 8-hour and 1-hour average ozone concentrations at Zion occurred on June 2.
Meteorological observations collected at Zion that day are displayed in Figure 2.2.9. Orange shading
highlights winds with an onshore (easterly) component; green shading indicates offshore (westerly)
winds. Southwesterly winds are abruptly replaced by southeasterly winds at all levels between hours 14
and 15 UTC (9 and 10 a.m. CDT) as a deep lake breeze arrived at Zion. The wind shift was accompanied
by a rapid drop in surface temperature and a sharp decrease of mixed layer depth. The lake breeze
persisted at all levels until sunset at about 1 UTC (8 p.m. CDT), then eroded near the surface while
persisting aloft.
1019
1020
1021
1022
1023
1024
1025
Figure 2.2.10 shows that a shallower and shorter-lived lake breeze was observed at Zion on June
11. This date was also characterized by high ozone levels at Zion (1-hour averages above 70 ppbv). The
lake breeze arrived just before 18 UTC (1 p.m. CDT), accompanied by a drop in surface temperature and
a rapid decrease in mixed layer depth. The lake breeze did not extend as high as that observed on June
2, but depressed both the surface temperature and mixed layer depth for a longer period. Mixed layer
depth and surface temperatures briefly recovered as the lake breeze dissipated around 00 UTC (7 p.m.
CDT) on June 12.
1026
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
49
1027
1028
1029
1030
Figure 2.2.9. Radiometer and sodar measurements at Zion on June 2, 2017. Top: color shading indicates winds (m/s) with a
westerly (green) or easterly (orange) component. Middle: mixed layer depth (m). Bottom: surface temperature (°C). Time
(UTC) extends from midnight CDT (hour 5 on 2017-06-02) to 11:55 p.m. CDT (hour 4 on 2017-06-03) on June 2.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
50
1031
1032
1033
Figure 2.2.10. Radiometer and sodar measurements at Zion on June 11, 2017 (as in Fig. 2.2.9). Time (UTC) extends from
midnight CDT (hour 5 on 2017-06-11) to 11:55 p.m. CDT (hour 4 on 2017-06-12) on June 11.
1034
2.3. Pandora/Ceilometer Measurements
1035
2.3.1. Overview
1036
1037
1038
1039
1040
1041
In addition to several remote sensing profiling instruments deployed by research groups at
Spaceport Sheboygan and Zion, the EPA deployed three ceilometers (Vaisala CL51) and five ground
based UV/visible grating spectrometers (Pandoras). The ceilometers were installed at two WI DNR
monitoring sites, Grafton and Milwaukee SER, and the Zion research site which is also an IEPA
monitoring site. The Pandora spectrometers were installed at Spaceport Sheboygan, Grafton,
Milwaukee SER, Zion, and Schiller Park in Chicago.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
51
1042
1043
1044
1045
1046
1047
1048
The primary objective of the ceilometers was to provide additional characterization of mixing
heights across the study domain to better understand the lake breeze impacts on mixed layer height
(planetary boundary heights) for model evaluation. The primary objective of the Pandora instruments
was to provide column totals for NO2 – which are useful as a contrast to in situ surface measurements;
further they can be compared to GeoTASO NO2 columns, in situ aircraft profiles, to satellite remote
sensing NO2, and to NO2 columns from photochemical models. Secondary objectives of Pandora were
column totals for ozone and HCHO.
1049
2.3.2. Methods
1050
1051
1052
1053
The mixed layer height measurements were obtained with the Vaisala CL51 via singlewavelength (910 nm) laser aerosol backscatter and use of Vaisala BLView 1.0 software package. The BLview software identifies up to three separate aerosol layers based on a gradient method applied to
aerosol backscatter signals, with the lowest layer most often identified as the mixing height.
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
Remote sensing for trace gas column amounts was performed using ground-based Pandora
spectrometers developed at NASA Goddard Space Flight Center (https://acdext.gsfc.nasa.gov/Projects/Pandora/index.html). Pandora spectrometers were placed at Schiller Park (IL
EPA), Zion (IL EPA/LMOS 2017 super site), Milwaukee (WI DNR NCore), Grafton (WI DNR) and Spaceport
Sheboygan (LMOS 2017 super site), in efforts to cover conditions representative of the Lake Michigan
shoreline environment. The Pandora instrument is a UV-VIS spectrometer system designed to measure
direct sun with a 2.1° field of view with a diffuser and 1.6° field of view without the diffuser. For direct
sun measurements the sensor head attached to a tracker follows a specific solar azimuth and zenith
direction that changes from East in the morning to West in the evening to track the sun (Herman et al.,
2009). The spectra measured by the Pandora spectrometers are collected by the local host computer
and then transferred to a central data processing system at 5-minute intervals, provided internet access,
where the data is processed via a set of common algorithms. The standard set of data products is total
column NO2, HCHO and ozone (Herman et al, 2009, Spinei 2018). Both the total column ozone and NO2
algorithms are considered validated algorithms, while the total column HCHO is still in validation stage.
1068
1069
1070
1071
1072
1073
1074
The Scientific Aviation (SA) concentrations sampled by the Mooney light aircraft during
ascending and descending spirals over Spaceport Sheboygan were used to calculate sub-columns of NO2
over the Spaceport Sheboygan research site. The SA sub-columns were calculated from the measured
NO2 between approximately 1km- 3km AGL and the EPA ground-based NO2 measurement, with the NO2
concentration linearly interpolated between the lower aircraft value and the EPA ground measurement.
The SA data are referred to as sub-columns even though they include data from both the surface
measurement and the SA aircraft profile.
1075
2.3.3. Results and Discussion
1076
1077
1078
1079
CL51 measurements indicate that the average mixed layer height during the LMOS 2017 period
was similar at Zion, Milwaukee SER, and Grafton (Figure 2.3.1, black lines), growing on average from less
than 1 km before 6 AM local time to approximately 1.5 km around 6 PM before lowering. The
observations indicate that the mixed layer height at Zion is slightly lower than at the other sites,
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
52
1080
1081
1082
1083
1084
1085
1086
1087
1088
reflecting more prevalent lower-level stratifications (gray lines). The most striking feature of the dataset
is the large range of day-to-day (gray, red and blue lines) and hourly variations, with observed mixing
layer heights under 200 m observed at all three sites. These variations are driven by the competition
between continentally influenced air (deeper mixing due to daytime heating and subsequent
convection) and stratified marine layers with shallow mixed layer depths. Also, in practice, the Vaisala
BLView1.0 algorithm for detecting mixed layer height may not detect shallow stratified layers of aerosol
backscatter that occur within a much deeper continental offshore flow aloft. As noted previously, the
conditions associated with the June 2 and 12 high ozone days, are distinct. The same can be said for the
mixing conditions based on CL51 measurements on those days (red and blue lines).
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
Figure 2.3.1. All LMOS CL51 First Mixed Layer Height measurements (km AGL) between 15 May and 25 June 2017 by time of
day (CST, gray lines). Data from 2 June (red) and 12 June (blue) are highlighted. The average is also shown (black solid line).
A comparison of the Pandora total vertical column and the SA sub-columns shows good
agreement (Figure 2.3.2), when the two outliers are removed from the regression. A review of the two
outliers show that these measurements are from May 27 and June 14, when the Pandora instrument
likely sampled a large NO2 plume during the SA spiral, which primarily sampled outside the plume. The
offset in the regressions reflects the contribution of stratospheric and tropospheric NO2 to the Pandora
observation that is above the aircraft. The agreement between the Pandora data and the aircraft
dataset lends confidence and caution to both datasets: in 9 of 11 spirals directly overhead of Spaceport
Sheboygan both instruments are sampling representative airmasses with high-quality measurements,
but in 2 of 11 samples, the atmosphere is sufficiently stratified such that there is not appropriate
overlap. Not shown here, the quality of the regression substantially degrades when including
Sheboygan inland, near the airport, and Sheboygan offshore samples, indicating again that the NO2
distributions over Sheboygan vary locally, as demonstrated in the GeoTASO data (e.g., Figure 2.4.4)
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
53
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
Figure 2.3.2. Comparison of in situ NO2 columns (x axis) and Pandora total column (y axis). The two outliers are May 27 and
June 14 when Pandora likely sampled a large plume with more weight than did the SA. Profiles are restricted in to within +/0.035 deg of Spaceport Sheboygan research site. This restricts the profiles to the Sheboygan onshore profiles. Furthermore,
only profiles extending 2000 m or greater vertically and without missing data were included. Regression is reduced mean
axis.
1119
1120
1121
1122
1123
1124
1125
Despite the relatively short data record, day-of-week variations exist in the NO2 measurements
(Fig.2.3.3, black solid line versus dashed). Over Schiller Park, Zion and Milwaukee, NO2 columns were
larger on Monday-Friday (black solid line) than on Saturday-Sunday (black dashed line), likely reflecting
weekend decreases of NOx emissions. Over Grafton and Sheboygan, there is little day-of-week variation
of NO2 columns. The sources of NOx in Sheboygan may have a different day-of-week profile than typical
of urban settings because it is a weekend tourist destination and downwind of a base-load coal-fired
power plant. Neither of these emissions influences typically have large day-of-week variations.
The average NO2 column measurements during LMOS 2017 generally indicate (Figure 2.3.3) that
NO2 columns are largest over Schiller Park, which is located just east of O’Hare International Airport, and
are small to moderate elsewhere. The most notable feature of the Pandora dataset is the departure of
NO2 column observations above the average at four sites during one ozone episode (Friday 2 June red)
versus observations that are average or smaller than average during the second ozone episode (Monday
12 June, blue). This finding is consistent with the more general observation than the dynamical and
chemical factors affecting ozone concentrations are different on these two days.
1126
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
54
1127
1128
1129
1130
1131
Figure 2.3.3. All LMOS 2017 Pandora NO2 column measurements between 15 May and 25 June 2017 by time of day (CDT,
gray lines). Data from 2 June (red) and 12 June (blue) are highlighted. Monday-Friday average (black solid line) and
Saturday-Sunday averages (black dashed line) are also shown.
1132
2.4. GeoTASO Measurements
1133
2.4.1. Overview
1134
1135
1136
1137
1138
1139
1140
1141
1142
The Geostationary Trace gas and Aerosol Sensor Optimization (GeoTASO) is an airborne
hyperspectral mapping instrument built by Ball Aerospace (Leitch et al., 2014) and is being used as an
airborne testbed for future high-resolution trace-gas observations from geostationary sensors such as
TEMPO. This instrument has two spectrometers capable of quickly measuring spectra ranging from
~300-400 nm in the UV channel with a spectral resolution of 0.43nm and ~400-700 nm in the Visible
channel with a spectral resolution of 0.88 nm (Nowlan et al., 2016). Spectra within the 435-460 nm
window are used to retrieve integrated atmospheric column amounts of nitrogen dioxide (NO2).
Preliminary results are described below. In the future, retrievals of column HCHO may also assist in the
analysis of photochemistry around Chicago and along the western shore of Lake Michigan.
1143
2.4.2. Methods
1144
1145
1146
1147
1148
1149
1150
During LMOS, GeoTASO was operated on the NASA LaRC UC-12 aircraft. Figure 2.4.1 (right)
shows GeoTASO integrated into the forward port of the aircraft. This aircraft configuration was flown
for 21 research flights based out of Madison, Wisconsin from May 22 -June 21 in support of LMOS 2017
and its objectives. Figure 2.4.1 (left) depicts how light from the sun travels through Earth’s atmosphere
and is scattered/reflected up to the nadir-viewing window aboard the aircraft. The light captured by the
instrument’s telescope is guided through an arrangement of mirrors, gratings, and beam splitters into
each spectrometer. The 45o nadir field-of-view, as depicted by the swath in Figure 2.4.1 (left), provides
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
55
1151
1152
1153
an across-track ground width of over 7 km from a nominal altitude of 28,000 ft. Flight plans were
designed to create raster maps, defined as a series of parallel flight legs spaced 6 km apart to provide
gapless coverage over regions of interest. Each flight had between 1-3 rasters.
1154
1155
1156
1157
1158
1159
Figure 2.4.1. Depiction oof GeoTASO sampling strategy and photo of GeoTASO instrument. GeoTASO’s sampling strategy
(left) including its 45° swath width with a swath sample taken every 250 ms along the aircraft flight line (from Nowlan et al.,
2016) and GeoTASO integrated onto the NASA LaRC UC-12 (right: Photo credit: Matt Kowalewski, USRA/NASA GSFC).
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
Prior to the campaign, a ‘playbook’ of flight plans (shown in Figure 2.4.2) was developed for
accomplishing a range of science objectives depending on the cloud/air quality forecasts during LMOS
2017. These flight plans are grouped into two categories: urban emissions sampling and coastal mapping
surveys. Emissions sampling flights were designed to observe time-varying emissions over the areas of
Chicago (8 flights/1-2 repeated rasters per flight) and Milwaukee (1 flight/3 repeated rasters). The
coastal mapping plans, oriented around the two supersites (Sheboygan and Zion), were designed to
observe large local point sources (such as power plants) as well as the regional influence from upwind
sources such as Chicago and Milwaukee. The goal of this mission was to capture how these emissions
are influenced by coastal dynamics, especially in the presence of lake breeze events. The yellow plan
(Figure 2.4.2, left) maps the entire extent of the western shore between the two supersites. The other
two plans focused on each supersite individually and the smaller areas of these plans allowed for
multiple repeated rasters per flight (up to three times over Sheboygan and two times over Zion).
1172
1173
1174
1175
1176
Quick look NO2 retrievals were prepared during LMOS 2017 within 24-48 hours of a flight. Due
to this timely turnaround, the science team was able to get a better grasp on what GeoTASO could
capture and flight plans were sometimes adjusted slightly from the ones shown in Figure 2.4.2. These
adjustments included adding/subtracting flight lines and shifting the flight lines to cover areas
forecasted to best accomplish the LMOS 2017 science objectives.
Flight Planning
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
56
1177
1178
1179
1180
1181
Figure 2.4.2. Flight plans within the LMOS ‘playbook’ separated by coastal mapping surveys of the western shore (left) and
emissions sampling over Chicago and Milwaukee (right). Red targets indicate the location of Pandora Spectrometers used
for validating GeoTASO NO2 retrievals.
1182
NO2 Retrieval
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
NO2 differential slant columns (DSCs) were retrieved from GeoTASO spectra via Differential
Optical Absorption Spectroscopy (DOAS) within the spectral window of 435-460 nm. The DOAS
technique provides a column amount relative to a reference scene, which ideally is unpolluted and
cloud-free. The resulting DSCs of each observation can be thought of as incremental columns relative to
the reference spectrum location and time, primarily showing contributions from local pollution sources.
The DSCs are also sensitive to changes in the stratosphere, however the stratospheric column is
relatively constant over the LMOS domain and the influence on the DSCs is a function of solar zenith
angle. The native spatial resolution for these observations is 250 m x 250 m. From there, co-adding to
larger spatial resolution increases signal-to-noise ratio (SNR). For this report, we’ve selected 750 m x
750 m spatial resolution as the best balance between improving the SNR and conserving fine spatial
resolution. All the results below and in the Appendix are shown at this spatial resolution.
1194
1195
1196
The NO2 DSC data are available from the NASA Archive (https://wwwair.larc.nasa.gov/missions/lmos/index.html) and mapped in the Appendix. The first publicly available
GeoTASO NO2 differential slant column data from LMOS was released in August 2018.
1197
2.4.3. Results and Discussion
1198
1199
Figure 2.4.3 shows how the preliminary NO2 DSCs compare to the coincident retrievals of the
network of ground-based Pandora Spectrometers (Herman et al., 2009) installed along the western
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
57
1200
1201
1202
1203
1204
1205
1206
1207
shore of Lake Michigan during LMOS 2017 (shown in Figure 2.4.2). Since Pandora’s measurement
reflects both tropospheric and stratospheric contributions, for a troposphere only comparison, the
Pandora stratospheric contribution is estimated and subtracted. The stratospheric contribution was
estimated using OMI Stratospheric Vertical Column over the Lake Michigan region during the time
period of the study multiplied by Pandora’s AMF. Circles show the average GeoTASO DSC within a 750-m
radius of the Pandora site and the average Pandora data within 5 minutes of GeoTASO’s overpass.
Horizontal bars show standard deviation of GeoTASO DSCs within the 750-m radius. Vertical bars show
the standard deviation of Pandora data within the ± 5-minute coincidence window.
1208
1209
1210
1211
1212
1213
1214
1215
1216
The plots show two different ranges. Figure 2.4.3 (left) includes one large outlier over Schiller
Park on June 1st (>50x1015 molecules cm-2). Both instruments captured this occurrence. All
coincidences, including the one outlier, correlate quite well (r2=0.89). When excluding the outlier,
correlation is still quite high with r2=0.67 and slope near 1 (Figure 2.4.3 right). The points with large
variance bars are also from the Schiller Park site in Chicago, which is located near O’Hare airport and
likely experiences very localized spatiotemporal variability of NO2 due to both ground and air traffic.
The rest of the sites were far less polluted and stayed below 10x1015 molecules cm-2. These results
provide confidence that GeoTASO captured the magnitude of NO2 columns within the LMOS 2017 study
region.
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
Figure 2.4.3. Preliminary comparison of the GeoTASO DSCs from the LMOS flights to the Pandora tropospheric slant column.
Circles show the average GeoTASO DSC within a 750-m radius of the Pandora site and the average Pandora data within 5
minutes of GeoTASO’s overpass. Horizontal bars show standard deviation of GeoTASO DSCs within the 750-m radius.
Vertical bars show the standard deviation of Pandora data within the ± 5-minute coincidence window.
GeoTASO captured consistent spatial patterns from area and point emissions sources during the
LMOS 2017 flights. While examples of all flights are found in the Appendix, Figure 2.4.4 shows a few
zoomed in examples from Chicago (left), Pleasant Prairie Power Plant (lower right), and Sheboygan
(upper right). Along the more rural western shore of the lake, the large power plants were often
observed by GeoTASO as distinct plumes. The areal extent of these plumes varied widely from day to
day. The typical spatial pattern in Chicago showed peaks near downtown and the industrial area in NW
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
58
1229
1230
1231
1232
1233
1234
1235
1236
1237
Indiana. This pattern was consistent through most of the Chicago flights, however the magnitude and
extent of these plumes were dependent on meteorology and day of week. Figure 2.4.5 shows the
difference between consecutive weekend/weekday sampling over the Chicago area on June 18 and 19,
2017. Also shown on this figure are the in situ NO2 vertical profiles from the Scientific Aviation aircraft
offshore and downwind of Chicago. Both days had similar meteorological conditions and spatial
patterns. As Sunday transitioned into the work week on Monday, there was a large increase in the
magnitude of NO2 columns over the Chicago metropolitan area. On Monday June 19 the enhancement
in the NO2 column was confined to the lowest 500-600m of the atmosphere, with one profile measured
by the in situ aircraft approaching 12x1015 molecules cm-2.
1238
1239
1240
1241
1242
1243
Figure 2.4.4. GeoTASO NO2 differential slant column maps for May 22 and June 15, 2017. DSC maps: May 22nd Chicago
Raster during mid-morning (left). June 15th morning raster in Sheboygan (hole in data due to clouds) (upper right). June 7th
morning raster over Pleasant Prairie (lower right). The left color bar corresponds to the Chicago map, and the right color bar
corresponds to both maps on the right.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
59
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
Figure 2.4.5. Maps of NO2 DSCs observed by GeoTASO on June 17 and 18, 2017 over the Chicago metropolitan area in the
early morning. Flight tracks colored by NO2 measured by Scientific Aviation below 800m are shown offshore and downwind
of the city. Vertical profiles from Scientific Aviation were averaged into 50 m altitude bins and plotted in the lower left (June
17th) and lower right (June 18th). Horizontal bars indicate the standard deviation of NO2 observed within each 50m bin.
1258
2.5. Scientific Aviation Measurements
1259
2.5.1. Overview
1260
1261
1262
1263
1264
1265
Scientific Aviation (SA) was contracted by the Electric Power Research Institute (EPRI) to
participate in LMOS 2017 with airborne in situ profiling of O3, NO2, CO2, CH4, altitude, T, RH, winds, and
pressure. The majority of the SA flights were conducted in close coordination with the NASA airborne
remote sensing measurements and provided vertical profiles over and offshore from selected ground
sites (Sheboygan and Zion), offshore profiles east of Milwaukee and Chicago, and onshore background
profiles. Low level flight legs were also conducted within the marine boundary layer.
1266
2.5.2. Methods
1267
1268
In situ NO2 measurements were conducted using an Aerodyne cavity attenuated phase shift
(CAPS) NO2 monitor with a sensitivity (S/N=3) less than 1.0 ppbv for 1 second and 0.1 ppbv for 10 second
Future work will include processing DSCs into vertical columns. This requires the calculation of
the air mass factor (AMF), which is defined as the ratio of the slant column/vertical column. AMFs are
calculated according to Palmer et al. (2001) as the product of the normalized vertical distribution of NO 2
(modeled by University of Iowa’s WRF-Chem simulations for the duration of the study period) and
scattering weights calculated by VLIDORT (Spurr, 2006). Cloud contamination can easily be filtered by
radiance levels observed by GeoTASO. These data can be used for assessing emission inventories, air
quality modeling evaluation, as well as linking how local and transported emissions can influence the air
quality along the coast of Lake Michigan.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
60
1269
1270
1271
1272
1273
1274
1275
averages. O3 was measured with a 2B Model 202 monitor with a precision (1-sigma, rms noise) of 1.5
ppbv for O3 less than 75 ppbv, or 2% for O3 greater than 75 ppbv. A Picarro instrument was used to
measure CO2/CH4 with a precision (1-sigma over 30 seconds) of less than or equal to 200 ppbv for CO2
and less than or equal to 2 ppbv for CH4. Temperature and relative humidity (RH) were measured with a
Vaisala HMP60 probe with a temperature accuracy of +/- 0.5 Celsius over temperature range of 10-30
Celsius, and a relative humidity accuracy of +/- 3% for 0-90%and +/- 5% for 90-100% relative humidity
over a temperature range of 0-40 Celcius.
1276
1277
1278
1279
SA flew a total of 21 flights during LMOS 2017 plus 1 post mission flight on June 24, 2017, near
Holland, MI. Each flight included ascending and descending spirals at predetermined waypoints from
near 60 m AGL over water and from 1000 m AGL over land to 3 km with low level legs in between the
spirals (Figure 2.5.1).
1280
1281
Figure 2.5.1. Scientific Aviation flights during LMOS 2017
1282
2.5.3. Results and Discussion
1283
1284
1285
1286
1287
1288
The two SA flights that occurred on days with ozone exceedances at both the Sheboygan KA
monitor and the Zion monitor will be discussed here. Figure 2.5.2 shows O3, NO2, RH, and temperature
for the SA flight on June 2. The SA aircraft took off from the Sheboygan County Municipal Airport at 19
UTC (noon CDT) and conducted ascending/descending spirals at Spaceport Sheboygan followed by
ascending/descending spirals at the Sheboygan KA offshore waypoint. The SA aircraft then headed south
and conducted low level profiling of the marine boundary layer past the Milwaukee offshore waypoint
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
61
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
to the Zion offshore waypoint. Ascending/descending spirals were conducted at the Zion offshore and
onshore waypoints and then the SA aircraft headed north towards the Oak Creek Power Plant waypoint.
The SA aircraft proceeded northeast to the Milwaukee offshore waypoint and conducted
ascending/descending spirals before heading to the northwest past the Port Washington Power Plant
waypoint. The SA aircraft proceeded northeast and conducted ascending/descending spirals at the
Sheboygan KA waypoint before returning to the Sheboygan County Municipal airport at 23:48 UTC
(6:48pm CDT). Shallow layers of higher RH and colder temperatures at the Sheboygan KA and
Milwaukee offshore waypoints, as well as during the low level leg between the Milwaukee offshore and
Port Washington Power Plant waypoint, occurred during periods where the SA aircraft sampled the
shallow marine boundary layer. Hourly ozone ranged from 90-110 ppbv and NO2 ranged from 4 to over
20 ppbv during these marine boundary layer sampling periods. The highest ozone measurements
occurred near the lake surface during the low level leg between the Milwaukee offshore spiral and the
Oak Creek Power Plant waypoint.
1302
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
62
1303
1304
1305
1306
1307
1308
Figure 2.5.2. Three dimensional (latitude, longitude, altitude) view of Scientific Aviation Flight on June 02, 2017, showing in
situ O3 (upper left, ppbv) NO2 (upper right, log10(ppbv)), relative humidity (RH) (lower left, %) and temperature (lower right,
K). Waypoints are indicated by red “+” symbols, land is indicated by black lines. The Sheboygan Haven WDNR monitor is the
northernmost waypoint and Zion is the southernmost waypoint.
1309
1310
1311
1312
1313
1314
1315
1316
1317
Figure 2.5.3 shows O3, NO2, RH and temperature for the SA flight on June 11. The SA aircraft
departed at approximately 18 UTC (1:00pm CDT) and conducted ascending/descending spirals at the
Sheboygan Haven, Sheboygan and Sheboygan offshore waypoints and then conducted a low level leg
within the marine boundary layer between Sheboygan and Sheboygan KA offshore waypoints.
Ascending/descending profiles were conducted at off-shore and onshore at Sheboygan KA followed by a
low level leg to the Sheboygan background waypoint where a final ascending/descending spiral was
performed prior to returning to Sheboygan County Municipal airport slightly prior to 22 UTC (5:00pm
CDT). Shallow layers of higher RH and colder temperatures were observed within the marine boundary
layer during the low level leg between Sheboygan offshore and at the Sheboygan KA offshore. Ozone
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
63
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
mixing ratios were above 60 ppbv throughout the lowest 3 km except for a layer of slightly lower O3
between 2 and 3 km. Ozone below 1km remained in excess of 70 ppbv except for the low level legs and
spirals at Sheboygan Haven. Highest ozone was encountered following the descending spiral at the
Sheboygan KA offshore waypoint where it reached up to 90 ppbv near the surface. NO2 was generally
less than 1 ppbv during most of the offshore flight legs except for very narrow enhancements at the
beginning and end of the Sheboygan spirals. CO2 mixing ratios in excess of 410 ppmv within these
narrow NO2 enhancements suggest that the aircraft sampled the plume from the Edgewater power
plant, which is located between the Sheboygan and Sheboygan KA waypoints. A broad region of NO2
between 1-2 ppbv was observed below 1 km during the Sheboygan Haven spiral and during the low level
legs during and after the Sheboygan background spiral.
1328
1329
1330
1331
1332
1333
Figure 2.5.3. Three dimensional (latitude, longitude, altitude) view of Scientific Aviation Flight on June 11, 2017 showing in
situ O3 (upper left, ppbv) NO2 (upper right, log10(ppbv)), relative humidity (RH) (lower left, %) and temperature (lower right,
K). Waypoints are indicated by red “+” symbols, land is indicated by black lines. The Sheboygan Haven WDNR monitor is the
northernmost waypoint and Zion is the southernmost waypoint.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
64
1334
2.6. Mobile Platform Measurements
1335
2.6.1. Overview
1336
1337
1338
1339
1340
1341
Two mobile platforms were deployed during LMOS 2017 to investigate the chemical gradients
between on shore stationary monitoring sites. The routes driven were to investigate E-W gradients and
some N-S gradients in ozone concentrations (e.g., Figure 2.6.1). The EPA Region 5 Geospatial Monitoring
of Air Pollution (GMAP) platform focused on the area around the Wisconsin-Illinois stateline, extending
northward into southeastern Wisconsin. The UW-Eau Claire platform studied more northern lakeshore
areas in Sheboygan and Ozaukee counties in Wisconsin.
1342
2.6.2. Methods
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
GMAP was used to measure ozone concentrations on June 6, 8, 12, and 13 to better understand
how ozone is produced and transported over Lake Michigan. A TEI49CPS was used to measure ozone
and streamed to a Campbell Scientific CR850 data logger. Meteorology and GPS coordinates were
measured and logged from the existing GMAP platform. The time stamp from the data logger was
paired with the GMAP computer. The TEI49CPS was placed in the rear seat of the GMAP vehicle to
maintain a temperature-controlled environment and minimize vibration, and a ¼ inch Teflon line was
added to the GMAP mast. The 49CPS was checked against the US EPA Region 5 SRP SN6 before and
after the study. The check followed standard procedures on “Standard Operating Procedures For
Conducting Verifications Of Level 2 Ozone Transfer Standards” (R5-ARD-0005-r1). The criteria for
acceptable results were determined for each point to be within 10.0%. All data points fell within the
acceptable criteria pre- and post-campaign. Stationary collocation comparisons between the Zion ozone
site (17-097-1007) and the GMAP ozone readings are included in Table 2.6.1.
1355
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
65
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
Figure 2.6.1. Southern transects driven by the GMAP van. Transect segments are color coded, named, and numbered. See
arrows for the number and overall direction of each segment. Arrow numbers correspond to the following transect
segments: 1) West from Zion Illinois Beach State Park on 21st St., 2) East toward Zion Illinois Beach State Park on 21st St., 3)
Illinois Beach State Park to Chiwaukee Prairie, 4) Chiwaukee Prairie to Kenosha, 5) Kenosha Lakeshore to Kenosha O3
monitor, 6) Kenosha O3 monitor to Lakeshore, 7) Kenosha Lakeshore to Racine O3 monitor, 8) Racine O3 monitor to
Windpoint, 9) Windpoint to Belmar Ave.
Table 2.6.1. GMAP Zion collocation comparisons
Date
6/6/2017
6/7/2017
6/8/2017
6/12/2017
6/13/2017
Time (CDT)
10:18
9:41
14:00
12:12
8:50
GMAP Reading
32.3
33.0
45.2
57.3
45.5
Zion Reading
31.8
32.0
45.5
57.3
45.1
PPB Diff vs Zion
0.5
1.0
-0.3
0.0
0.4
Percent Diff
1.6%
3.1%
-0.7%
0.0%
0.9%
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
The second mobile platform deployed for LMOS 2017 was the UW-Eau Claire automobile
platform, equipped with a 2-B Technologies Personal Ozone Monitor and a handheld Kestrel weather
sensor. This platform made regular measurement stops between the Grafton WDNR site and the
Sheboygan Spaceport site to investigate N-S and shallow E-W gradients. The ozone measurements were
taken at a height of 3 m for at least 5 minute increments at each location. The 15 stops are described in
Table 2.6.2. On June 2, 3, 12, 13, and 17 the UW-Eau Claire vehicle drove the route of some combination
of the stops (e.g. Stops 1-13 on June 2, Stops 10-15 multiple times on June 3 to correspond to ship and
aircraft measurements made concurrently). Data from the ozone monitor was obtained using 1 minute
averaged signal. The duty cycle of the Personal Ozone Monitor subtracts a background (scrubbed ozone)
UV-signal from a whole air sample, which accounts for the UV absorption of water. However, drastically
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
66
1377
1378
changing humidity can lead to higher standard deviations in this instrument, which is why the vehicle
chose to stop at a single location to measure ozone.
1379
Table 2.6.2. UW-Eau Claire automobile measurement stops
Latitude
Longitude
Site name
Colocation
43.3443
43.3342
43.3814
43.4114
43.4347
43.4993
43.4984
43.5576
43.6224
43.6673
43.6728
43.6735
43.7079
43.7459
43.7508
43.7614
-87.9207
-87.8872
-87.8759
-87.8693
-87.8338
-87.7937
-87.8116
-87.7923
-87.7497
-87.7160
-87.7273
-87.7206
-87.7045
-87.7083
-87.7272
-87.7490
1 Grafton
2 Lions Den Park
3 Western and Division
4 Country Inn
5 High Point Beach Ln
6 Harrington Beach
7 Harrington Beach Park
8 Amsterdam Park
9 Foster Road
10 Kohler-Andrae Dunes
11 Straight out W of KA
12 Corner near KA entrance
13 South of Sheboygan Power Plant
14 Spaceport
15 Kiwanis Park
16 HSHS St. Nicholas Hospital
WI-DNR
WI-DNR
WI-DNR
LMOS Spaceport Supersite
1380
1381
2.6.3. Results and Discussion
1382
1383
1384
1385
1386
1387
1388
Figure 2.6.2 through Figure 2.6.4 are timeseries of the measured ozone concentrations over
individual transects on different days during LMOS 2017. Gradients inland were observed on June 8th
and 12th. On June 8th, when ozone concentrations were 50-70 ppbv, the gradient concentration
increased inland (Figure 2.6.2). On June 12th, when ozone levels were higher (50-80 ppbv), the ozone
gradient decreased inland, particularly in the nearly 6 mile transect from the Kenosha ozone monitor to
the lake shore (Figure 2.6.3, transect 6) and Windpoint to Belmar (Figure 2.6.4, transect 9), with a nearly
30 ppbv gradient towards Lake Michigan.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
67
1389
1390
1391
Figure 2.6.2. Ozone concentration (ppbv) for GMAP Zion Illinois Beach State Park 21st St Transect – West (Transect 1) on June
8, 2017 as a function of time (UTC).
1392
1393
1394
Figure 2.6.3. Ozone concentration (ppbv) for GMAP Kenosha O3 monitor to Lakeshore (Transect 6) on June 12, 2017 as a
function of time (UTC).
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
68
1395
1396
1397
Figure 2.6.4. Ozone concentration (ppbv) for GMAP Windpoint to Belmar Ave Transect (Transect 9) on June 12, 2017 as a
function of time (UTC).
1398
1399
1400
1401
1402
1403
The observations from the UW-Eau Claire mobile monitor reflected increasing ozone
concentrations through the afternoon of June 2 (Figure 2.6.5). The spatial gradient moving north along
the lakeshore observed in Figure 2.6.5 does not necessarily reflect a north-south gradient, but is
interpreted as a growth in ozone abundance over the afternoon in which the measurements were taken,
noting the time when the automobile started at the southern sites and when the automobile arrived in
Sheboygan.
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
On June 3, several trips were conducted between the Sheboygan supersite and Kohler-Andrae
State Park WDNR site because of the concurrent ship and aircraft measurements on that day. Figure
2.6.6 depicts the ozone measurements from the multiple trips with an arrow which demonstrates the
general direction of the automobile. On this day, a very shallow incursion of lake air came on land later
in the afternoon, which was observed as higher ozone (>50 ppbv) and lower ambient temperatures than
sites inland. Panel 4 of Figure 2.6.6 shows that sites inland from the Sheboygan shoreline supersite
measured ozone 20 ppbv less than that observed at the shoreline. One interesting observation from that
day was the onset of the lake breeze incursion. At the Kohler-Andrae shoreline location, observations of
ozone remained higher throughout the afternoon at near 60 ppbv. At 21:20 UTC (4:20pm CDT) the
incursion of lake air on land started to occur, where inland observations of ozone were high on the E-W
highway directly outside of the Kohler-Andrae State Park entrance (measurement of 55 ppbv), but a
measurement taken on a N-S street surrounded by trees remained lower at 42 ppbv measured within
minutes of each other. The temperature readings at these two sites were also different, with colder air
correlating to the higher ozone measurements inland. This observation shows the potential for complex
flow patterns within the lake breeze incursion. There is not enough evidence from this monitoring
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
69
1419
1420
1421
platform to indicate that this was a lake breeze driven event. By 5:30-6:30 pm local time, the ozone
measurements at sites 10-13 were ~60 ppbv with less sharp gradients at those sites, but a 20 ppbv drop
off to sites 14-15 farther inland.
1422
1423
1424
1425
Figure 2.6.5. UW-Eau Claire automobile platform ozone measurements from June 2, 2017. Times are given in local time
(CDT) and each point is a 5 minute average of 1 minute points ± 1 standard deviation.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
70
1426
1427
1428
1429
Figure 2.6.6. UW-Eau Claire automobile platform ozone measurements from June 3, 2017. Times are given in local time
(CDT) and each point is a 5 minute average of 1 minute points ± 1 standard deviation.
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
71
1430
1431
Section 3.
Summary of LMOS Model/Measurement Evaluations
1432
1433
1434
1435
Meteorological and photochemical models have been used for forecasting, flight planning, and
analysis of LMOS 2017 observations. As shown in Table 3.1, four models have been used for either
forecasting / planning, and for post-study interpretation and analysis. Three of the models have results
in this report. The fourth modeling system results will be in future reports.
1436
Table 2.6.1. Summary of forecast models used during LMOS.
Model
Group and
Report
Section
Univ. of Iowa;
section 3.1
Horizontal
Resolution
4 km
Chemistry? Coupled
Air Quality
and Met?
Yes
Yes
NWS, NAQFC
NAM-CMAQ
(operational
CMAQ 5.0.2)
WRF v3.8
NWS/NOAA;
section 3.2
12 km
Yes
No
Univ. of
Wisc.; section
3.3
No
NA
WRF v3.8, CMAQ
v5.2.1
Univ. of Wisc.
4 km and 12 km
in this report;
1.3 km in future
reports
4 km & 1.3 km
(in future
reports)
Yes
No
WRF-Chem
v3.6.1
Purposes
Forecast; PostAnalysis (Sensitivity
to Configuration
and Emissions)
Forecast;
Evaluation of
Operational
Forecast
Sensitivity of
Meteorology to
Configuration
Post-Analysis
1437
1438
1439
1440
1441
1442
1443
1444
The University of Iowa modeling group developed a WRF-Chem forecasting framework at 4 km
horizontal resolution to support the flight planning during LMOS 2017, and has extended that modeling
effort to post-analysis sensitivity studies. WRF-Chem is a widely used regional modeling system that
simultaneously solves meteorological and photochemical equations in a coupled fashion . The coupled
architecture of WRF-Chem makes it well suited for air quality forecasting, reduces memory and storage
requirements relative to uncoupled models, and allows feedbacks between air quality and
meteorological variables, such as aerosol particles influencing radiation.
1445
1446
1447
1448
1449
Operational air quality forecasts from the National Weather Service (NWS) National Air Quality
Forecasting Capability (https://airquality.weather.gov/ and
https://www.weather.gov/sti/stimodeling_airquality_predictions) were also used to support daily flight
planning activities during LMOS 2017 and have been compared to observations in post-analysis. The
operational forecast is based on NAM (meteorology) and CMAQ (photochemical grid model).
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
72
1450
1451
1452
1453
1454
An additional LMOS-related modeling activity is focused on optimization of the WRF
meteorological modeling system for prediction and regulatory scenario modeling of lake-breeze ozone
events in the region. Various configurations and resolutions are being explored in order to provide a
best configuration for future modeling. This can improve future products from WRF-Chem as well
photochemical models driven by WRF meteorology (i.e., WRF + CMAQ, WRF + CAMx, etc.).
1455
1456
The optimized WRF configuration (above) is being used to drive WRF-CMAQ simulations by the
University of Wisconsin-SSEC team, and those results will be presented in future reports.
1457
3.1. WRF-CHEM Model
1458
1459
1460
1461
The University of Iowa modeling group developed a WRF-Chem forecasting framework to
support the flight planning during LMOS 2017 and extended this framework for post-campaign analysis.
This section describes the details of WRF-Chem forecasting and modeling system and provides
preliminary evaluations of model and emission performance.
1462
3.1.1. WRF-Chem domain
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
A single domain WRF-Chem v3.6.1 was used to simulate the atmospheric transport and
chemistry around Lake Michigan during LMOS 2017. Figure 3.1.1 illustrates the domain and the
underlying terrain map. The domain covers Minnesota, Iowa, Missouri, Wisconsin, Illinois, Michigan,
Indiana, and Ohio (37.6-48.8oN and 95.45-81.46oE) with 4 km x 4 km horizontal resolution or 310x260
grids. The domain has 53 vertical levels with domain top at 50hPa (~11 layers within the lowest 1 km).
Meteorological initial and boundary fields were retrieved from the National Center for Environmental
Prediction (NCEP) Global Forecasting System (GFS) (National Centers for Environmental Prediction,
National Weather Service, NOAA 2000). European Centre for Medium-Range Weather Forecasts
Monitoring Atmospheric Composition and Climate reanalysis (MACC) (Inness et al. 2013) was used as
chemical boundary conditions.
1473
1474
Figure 3.1.1. WRF-Chem domain: A. Chicago, B. Zion, C. Kenosha, D. Kohler Andre
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
73
1475
1476
Table 3.1.1 summarizes the major WRF-Chem configurations used for the forecast and postanalysis. These configurations remain constant for all WRF-Chem simulations.
1477
Table 3.1.1. Summary of WRF-Chem configuration
Horizontal resolution
Vertical resolution
PBL Scheme
Surface layer
Land Surface
Microphysics
Shortwave radiation
Longwave radiation
Cumulus Parametrization
Gas-phase chemistry
Biogenic emission
Lightening
Land use
SST
4km
53 layers (11 within the lowest 1km)
MYNN3
MYNN surface layer
5-layer thermal diffusion
Morrison double-moment scheme
Goddard shortwave
RRTMG scheme
Grell-Freitas scheme
RACM-ESRL
MEGAN
OFF
USGS
No SST update
1478
1479
1480
1481
1482
1483
1484
Gas phase chemistry in WRF-Chem was simulated using the Regional Atmospheric Chemistry
Mechanism using Earth System Research Laboratory (RACM-ESRL) (Stockwell et al. 1997) with an
updated reaction table for gas phase chemistry. RACM-ESRL (Kim et al. 2009) is an updated version of
the RACM mechanism and includes 23 photolysis and 221 chemical reactions (Ahmadov et al. 2015). The
Modal Aerosol Dynamics Model/Secondary Organic Aerosol Model (MADE/SORGAM) was used to
simulate aerosol thermodynamics and chemistry.
1485
1486
1487
1488
1489
1490
1491
Anthropogenic emissions for the WRF-Chem simulation were estimated with the 2011 National
Emission Inventory (NEI-2011) with 4 km x 4 km horizontal and hourly temporal resolution regridded to
the LMOS 2017 modeling domain. In the base anthropogenic emission inventory (NEI11-0.72 NOx) NOx
emissions were reduced by 28% to account for the reduction in NOx emission from 2011 to 2017
(https://www.epa.gov/air-emissions-inventories/air-pollutant-emissions-trends-data). Reductions were
applied uniformly (proportional to NEI-2011 emission) across the domain (Fig. 3.1.1). Actual reductions
would have a non-uniform spatial profile.
1492
1493
1494
1495
1496
1497
We used the Model of Emissions of Gases and Aerosols from Nature (MEGAN) version 2.1 to
estimate biogenic emissions (Guenther et al. 2012). The base runs used anthropogenic hydrocarbons
from NEI 2011. Adjustment of NEI 2011 to account for documented reductions in bottom up
anthropogenic hydrocarbons was not performed. Different emissions scenarios (based on NEI-2011)
were tested to capture the sensitivity of ozone concentrations to NOx, hydrocarbon (HC), and isoprene
emissions. Table 3.1.2 summarizes the emissions scenarios designed for this study.
1498
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
74
Table 3.1.2. Emission scenarios designed for LMOS forecasting and post-analysis system
Name
FR_base
Mode
Forecast
Emission
NEI11-0.72NOX
Period and description
Duration of campaign/Daily reinitialization
FR_pert
Forecast
NEI11-0.22NOX
Duration of campaign/Daily reinitialization
0.72NOx (base)
Post-analysis
NEI11-0.72NOX
05-20 to 06-22
Daily initialization of met at 00Z
0.22NOx
Post-analysis
NEI11-0.22NOX
05-30 to 06-04
Daily initialization of met at 00Z
0.72NOx_5HC
Post-analysis
NEI11-0.72NOx-5HC
0.72NOx_10HC
Post-analysis
NEI11-0.72NOx-10HC
05-30 to 06-22
Daily initialization of met at 00Z
05-30 to 06-05
Daily initialization of met at 00Z
0.5NOx_5HC
Post-analysis
NEI11-0.5NOx-5HC
0.5NOx_2HC
Post-analysis
NEI11-0.5NOx-2HC
0.72NOx_2iso
Post-analysis
NEI11-0.72NOx-2isop
0.72NOX_MOZ
Post-analysis
NEI11-0.72NOx
Chem
mech.
Bio.
emis.
NOx and HC emis.
Forecast
1499
05-30 to 06-04
Daily initialization of met at 00Z
05-30 to 06-04
Daily initialization of met at 00Z
05-20 to 06-21
Daily initialization of met at 00Z
05-30 to 06-03
Daily initialization of met at 00Z
MOZART chem mechanism
1500
3.1.2. University of Iowa forecast system
1501
1502
1503
1504
1505
1506
1507
During LMOS 2017 WRF-Chem forecasts were used as aids during the flight planning briefings.
Two separate 72-hr forecasts (simulation start time 6Z) were initialized every evening and the workflow
is depicted in Figure 3.1.2. The NEI11-0.72NOx emissions were used in the base simulation (FR_base).
The NOx emissions were further reduced by 50% (total reduction of 78%) and used in a second forecast
simulation to account for uncertainties of NOx emissions in the NEI-2011 and to test the sensitivity of
ozone concentrations to NOx. MACC reanalysis data were used for the chemical initial and boundary
conditions for each forecast.
1508
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
75
1509
Figure 3.1.2. Example of forecast workflow for three days May 21st, 22nd, and 23rd
1510
1511
1512
1513
1514
1515
1516
1517
Several analysis products were made available during LMOS 2017 and uploaded to the
University of Iowa LMOS website: https://bio.cgrer.uiowa.edu/LMOS/. Spatial plots of 11 chemical
species, 4 columns, and 10 meteorological products were available at four pressure levels and the full
domain or zoomed onto Lake Michigan. The cross-sectional plots with cloud and NO2 contours overlaid
on color contours of 10 chemical species were produced for three west-east and two north-south
transects of the domain. Also, 11 ground sites were chosen to show curtain plots and time series graphs
of nine chemical species, relative humidity, and temperature. Examples of these products are shown in
Figure 3.1.3.
1518
1519
1520
Figure 3.1.3. Example of forecast products available for the flight briefings. Cross section plots (A), curtain plots (B), spatial
plots (C), and time series graphs (D).
1521
3.1.3. Post-analysis
1522
1523
1524
1525
1526
1527
This section summarizes the WRF-Chem simulations. The goals of the analyses presented here
are to evaluate the ability of the model and emissions inventory to reproduce observations and to test
the sensitivity of simulated ozone concentrations to changes in NOx, HC, and isoprene emissions. The
model was evaluated against observations from the Zion supersite, hourly ozone mixing ratios, hourly
PM2.5, and hourly METAR station meteorology downloaded using the EPA Remote Sensing Information
Gateway (RSIG, ozone, PM2.5, and metar fields, RSIG3D version 20170410, www.epa.gov/rsig), and in
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
76
1528
1529
1530
1531
situ observations from Scientific Aviation flights. Table 3.1.2 is a summary of the emission scenarios
tested for the WRF-Chem sensitivity analyses. Meteorological fields in the post-analysis run were reinitialized with NCEP meteorological fields every 24 hours (starting 00Z or 19:00 local time). Each cycle
simulated 30 hours and the first 6 hours were discarded as a spin-up.
1532
Meteorological performance
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
Preliminary evaluation of meteorological performance indicates that WRF-Chem successfully
captured much of the synoptic variability of onshore flow. Performance is shown for the reanalysis
configuration of the model, not for the forecast version used for flight planning. The model generally
reproduced the temperature observations but frequently under-predicted observed maximum
temperatures. Observed wind speed and direction were also generally reproduced by the model. Table
3.1.3 includes a summary of wind speed and direction statistics at five different stations for the period
of 2017-05-22 to 2017-06-21. On average, the model better captured wind speed at inland sites and
over-predicted wind speed at sites closer to the shore such as Zion, Sheboygan, and Manitowoc. The
modeling team is continuing to investigate the sensitivity of meteorological skill to selection of boundary
condition, initial condition, land surface model parameterization, and nudging. The goal is to reduce
transport errors on ozone episode days as much as possible, so that model-observation errors can be
interpreted mainly as a function of non-transport model processes including chemistry and emissions.
Commonly used meteorological performance benchmarks such as those in Emery et al. (2001) need to
be interpreted carefully in the context of coastal ozone modeling performance. Depending on the
temporal and spatial averaging choices and met station selection, models may appear skilled in terms of
traditional benchmarks, yet not be sufficient for ozone modeling skill.
1549
1550
Table 3.1.3. Summary of wind speed and statistics at along the shore (Zion, Sheboygan, Manitowoc) and inland (Midway and
O’Hare) stations for the duration of 2017-05-22 to 2017-06-21.
All Days
Midway
(41.8,-87.9)
O’Hare
(42.0,-87.9)
Zion
(42.4,-88.0)
Sheboygan
(43.8,87.85)
Manitowoc
(44.1,-87.7)
wspd avg
(m/s)
4.22
4.45
4.28
4.28
5.10
3.22
5.15
wspd std
(m/s)
1.82
2.32
1.86
2.47
2.14
2.29
2.26
wdir avg
(deg)
250.71
249.41
255.49
262.50
257.47
228.44
264.24
wdir std
(deg)
78.54
89.95
76.56
89.66
80.51
93.80
74.86
avg U
(m/s)
1.67
1.13
1.81
1.05
2.03
0.58
2.35
avg V
(m/s)
0.58
0.43
0.47
0.14
0.45
0.52
0.24
OBS
3.84
2.50
248.62
88.41
1.16
0.45
Model
OBS
5.09
3.61
2.23
2.32
264.12
243.10
79.36
90.58
2.16
1.07
0.22
0.54
Model
OBS
Model
OBS
Model
OBS
Model
1551
1552
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
77
1553
Ozone and NOx performance
1554
1555
1556
1557
1558
1559
1560
1561
This section summarizes the WRF-Chem performance through comparison with ozone
observation from IL EPA monitors at Zion (42.5N, -87.81W) and Chicago (42.1N, -87.9W) and NOx
observations at the LMOS 2017 Zion supersite. The appendix includes an evaluation of WRF-Chem
performance in capturing ozone, NO2, and relative humidity along the Scientific Aviation flights. WRFChem generally reproduced the diurnal variability in observed ozone, however consistently underpredicted the observed maximum ozone values. NOx was over-predicted in the base simulation. During
the last two weeks of the campaign (from June 10 to June 22) WRF-Chem had high biases for NOx. Figure
3.1.4 shows time series of modeled ozone and NOx vs. observations collected during LMOS 2017.
1562
LMOS 2017 Preliminary Finding Report. Version 4/22/2019.
78
1563
1564
1565
1566
1567
Figure 3.1.4. Samples of WRF-Chem model output and sensitivity. LEFT panels (A, B, & C) Model output (red) vs. observations (black) for ozone in Chicago and Zion, and for
NOx at Zion. MIDDLE panels (D, E, & F) sensitivity of ozone (D, E) and NOx (F) to hydrocarbon emissions. RIGHT panels (G, H, & I) sensitivity of ozone (G, H) and NOx (I) to
changes in NOx emissions.
1568
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
79
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
The sensitivity of ozone to changes in emissions was tested by modifying NOx, HC, and isoprene
emissions. The sensitivity simulations were conducted for a short period (May 30 – June 4) to focus on
the June 2 high ozone event. During this event, WRF-Chem showed an underestimation bias in ozone of
~-50 ppbv at Zion and ~-30 ppbv at the Chicago site. In a reduced NOx sensitivity, NOx emissions (from all
sources) were decreased by 50% (0.22NOx). This sensitivity resulted in ~20 ppbv increase in ozone
concentrations at Zion site and ~10 ppbv reduction in ozone concentrations at the Chicago site. A pair of
HC sensitivities increased the HC emissions across the board by a factor of 5 (5HC) and 10 (10HC).
Increasing the HC emissions by a factor of 5 (0.72NOx_5HC and 0.5NOx_5HC) increased the simulated
ozone by ~50 ppbv at Zion and ~5 ppbv at the Chicago site. Increasing the HC emissions by a factor of 10
(0.72NOx_10HC) produced additional simulated ozone at Zion but resulted in no significant change in
ozone at the Chicago site (Figure 3.1.4). In another sensitivity experiment, biogenic isoprene emissions
were doubled to assess the impact on predicted ozone. This had only a small impact on peak ozone
concentrations. Ozone concentrations during the afternoon and evening of June 2 increased over much
of the southern portion of Lake Michigan by up to about 6 ppbv. Isoprene concentrations, as well as
isoprene oxidation products (formaldehyde, MVK, methacrolein) increased by up to a factor of two in
the isoprene sensitivity experiment. It should be noted that isoprene was only increased within the
domain shown in Figure 3.1.1 and not throughout the entire CONUS domain.
1586
1587
1588
1589
1590
Table 3.1.4 shows hourly and MDA8 ozone statistics to quantify the changes in modeled ozone from
each sensitivity experiment. The 0.72NOx_5HC simulation improved the model bias and correlation (R2)
at Zion for both hourly and MDA8 ozone, but did not improve model performance at the Chicago site.
While not displayed in Table 3.1.4, the isoprene sensitivity produced slight (up to 3 ppbv) increases in
modeled ozone during the June 2nd high ozone event.
1591
Table 3.1.4. Summary of Model performance in capturing O3 at Chicago and Zion sites.
Hourly O3 (ppbv)
Chicago
0.72NOx (base)
0.22NOx
0.5NOx_5HC
0.72NOx_5HC
0.72NOx_10HC
n
R2
RMSE
466
72
72
466
96
0.56
0.76
0.78
0.57
0.64
11.3
15.2
13.6
12.2
15.1
mod
mean
33.8
37.9
40.8
38.2
44.3
obs
mean
34.3
34.7
34.7
34.3
36.5
bias
-0.5
3.2
6.1
3.9
7.8
Hourly O3 (ppbv)
Zion
0.72NOx (base)
0.22NOx
0.5NOx_5HC
0.72NOx_5HC
0.72NOx_10HC
n
R2
RMSE
467
72
75
467
99
0.35
0.76
0.79
0.44
0.72
14.2
15.2
8.7
13.3
11.2
mod
mean
38.1
37.9
53.1
42.2
47.1
obs
mean
41.5
34.7
44.4
41.5
45.6
bias
-3.4
3.2
8.7
0.7
1.5
8hr peak O3 (ppbv)
mod
n
R2
mean
18
0.7
49.2
3
1
49.1
3
0.95 56.1
18
0.72 54.6
3
0.89 63.7
obs
mean
48
54.9
54.9
48
55.4
8hr peak O3 (ppbv)
mod
n
R2
mean
19
0.38
51.3
2
1
49.1
3
1
61.3
19
0.54
58.2
4
0.85
70.3
obs
mean
56.6
54.9
63.8
56.6
64.5
bias
1.2
-5.8
1.2
6.6
8.3
bias
-5.3
-5.8
-2.5
1.6
5.8
1592
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
80
1593
1594
1595
1596
1597
1598
1599
1600
1601
Figure 3.1.5 shows the average vertical profiles from the WRF‐Chem model base run (0.72 NEI 2011 NO x)
vs. Scientific Aviation profiles for over the land and over the lake spirals. In other words, the offshore
and onshore spirals (see Figure 2.5.1) measured during the campaign are separated and the
corresponding model values are extracted. Additional vertical profile plots of model vs. observation can
be found in the appendix. Key results from Figure 3.1.5 are that: (i) model captures the vertical variation
in both O3 and NO2 profiles very well; (ii) ozone is underpredicted within the boundary layer sampled by
Scientific Aviation; (iii) underprediction of ozone is most pronounced at the lowest layers, i.e. in the lake
boundary layer; (iv) NO2 is overpredicted at elevations below about 1 km with higher biases over the
lake.
1602
1603
1604
1605
Figure 3.1.5. Scientific Aviation vertical profiles (black) vs. paired WRF-Chem simulation results (red) for O3 (top row) and NO2
(bottom row) over land and over lake spirals. Elevations are meters above sea level. The lake surface is at 176m above sea
level.
1606
3.1.4. Soil NO Emissions
1607
1608
1609
One internal reviewer mentioned the potential importance of soil NO emissions in the region.
Given the overlap in the early summer ozone period, and some fertilizer application timings, agricultural
soil NO emissions may be an important part of the regional oxidant burden. This is an excellent point,
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
81
1610
1611
but one where the study design, which focused on lake-land breezes and anthropogenic plumes, may
not be able to address.
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
A literature review indicates that (now that regional anthropogenic NOx is decreasing), biogenic
NO, particularly from fertilized soils, may be a significant contribution. For example in Vinken et al. (Fig.
3c) indicates that the soil NO contribution to column NO2 in the upper Midwest in June may be
~0.6x1015 molec/cm2. This compares to NO2 measurements from Pandora during LMOS of 5-15x1015
molec/cm2 at Sheboygan. Vinken et al. (Fig. 7) puts total June soil NO emissions from a large group of
North Central Midwest agricultural states at 0.01 Tg N. Comparing this to Wisconsin emissions of NOx
(247,000 short tons per year in 2014 NEI), the Wisconsin emission, when converted to Tg N per month,
is 0.009 Tg N. So again, we see a possibility of a significant contribution from soil NOx. The spatial
patterns and temporal allocation of the soil NOx from OMI indicate it is mainly from fertilizer application,
which has improved spatial and temporal allocations recently, such as (Balasubramanian et al., 2015)
and follow-on publications from the same group. Similarly, cropland NO is reported as significant and
underestimated in traditional inventories in the San Joaquin Valley of California (Almaraz et al., 2018).
They report that cropland NO may contribute 20-51% of California NO emissions. Hudman et al. report
Midwestern cropland emissions of 3-14 kg N/ha. Multiplying this by Illinois cropland acreage and
assuming 1/6 of the emissions in a single growing season month (such as June) is equivalent to 0.0060.026 Tg N soil NO from Illinois cropland alone (Hudman et al., 2012).
1628
3.1.5. Future directions
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
The higher temperatures that occurred during the last two weeks of LMOS 2017 may have
resulted in increased biogenic emissions. A future modeling experiment will explore this trend and
assess how updating the Leaf Area Index (LAI) in the biogenic emissions model with MODIS LAI will
impact the biogenic emissions estimates and the subsequent impact on modeled ozone. The WRF-Chem
simulations performed to date all used the default WRF-Chem lake surface temperatures. A future
modeling experiment will update the model defaults with lake temperature observations from the Great
Lakes Surface Environmental Analysis (GLSEA) data (Schwab et al, 1999) for Lake Michigan. NOx
emissions from lightning were not considered in any of the WRF-Chem simulations. An additional
simulation will include lightning NOx emissions to assess the impact of these emissions on ozone
formation. Finally, additional LMOS 2017 observational data, such as in-situ observation from Scientific
Aviation aircraft, GeoTASO NO2 columns, and HC observation at the Zion site (WAS and PTR-QiTOF), will
be used to evaluate the skill of the WRF-Chem simulations.
1641
1642
3.2. NAM-CMAQ Model
1643
1644
1645
1646
1647
Operational air quality forecasts from the National Weather Service (NWS) National Air Quality
Forecasting Capability (NAQFC, Stajner et al., 2012 Chai, et al., 2013) were used to support daily flight
planning activities during LMOS 2017. The NAQFC uses the NWS North American Model (NAM)
Nonhydrostatic Multiscale Model with Arakawa B-grid staggering (NMMB) (Janjic and Gall, 2012) and
the Environmental Protection Agency’s (EPA) Community Multiscale Air Quality Model (CMAQ; Appel et
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
82
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
al., 2017) to perform twice daily forecasts of air quality over the continental US
(http://airquality.weather.gov/). The NMMB is run on a 12km grid over North America using the Rapid
Radiative Transfer Model (RRTM) radiation scheme (Mlawer et al., 1997) for both shortwave and
longwave, Ferrier's microphysics scheme (Ferrier et al, 2002), Betts-Miller-Janjic convection (Betts 1986;
Betts and Miller 1986; Janjic ́ 1994), Mellor-Yamada-Janjic PBL and surface layer schemes (Mellor and
Yamada, 1982; Janjic 1990, 1994, 2001), and the NOAH land surface scheme (Ek et al., 2003). NAQFC
uses CMAQ 5.0.2 with CB05 gas phase chemistry (Yarwood et al., 2005) and the 6th generation CMAQ
aerosol chemistry (AERO6). The 2017 NAM-CMAQ point source emissions are based on 2015 Continuous
Emissions Monitoring System (CEM), 2017 DoE Energy Outlook, Canada 2011 Environment Canada
Emission Inventory (ECEI), and Mexico inventory (MI) 2012 version2.2. Area source emissions use
NEI2011 (US), ECEI 2006 (Canada) and MI 2012 (Mexico). Mobile source emissions are based on Cross
State Air Pollution Rule (CSAPR) 2011 Emission Data. Wildfire aerosol emissions use NESDIS Hazard
Mapping System (HMS) fire detection and emissions from the US Forest Service (USFS) BlueSky v3.5.1
framework (O’Neill et al, 2008). Biogenic emissions are estimated inline in CMAQ with the Biogenic
Emissions Inventory System (BEIS3) v3.14 (Pouliot, 2008; Pouliot and Pierce, 2009). NAM-CMAQ 48 hr
forecasts are initialized every 6 hours. The 06Z NAM-CMAQ forecasts were used in the following
comparisons.
1665
Comparison with WI DNR Coastal Ground Monitors
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
Figure 3.2.1 and 3.2.2 show observed and NAM-CMAQ wind speed and ozone wind roses for the
Chiwaukee and Sheboygan KA WDNR coastal monitors during LMOS 2017 (May 22-June 22). NAMCMAQ captured the prevailing southwesterly wind direction at Chiwaukee but underestimated the
frequency of the prevailing southerly winds and overestimated the frequency of westerly winds at
Sheboygan KA. The NAM-CMAQ forecasts consistently overestimated wind speed at both monitors, with
a factor of 2 underestimate in the frequency of winds less than 3 m/s and a factor of 2-10 overestimate
in the frequency of winds greater than 6 m/s. The overestimate in the frequency of winds greater than
6 m/s was most pronounced at the Sheboygan KA monitor where they accounted for less than 2% of the
observations and nearly 24% of the NAM-CMAQ wind speeds. NAM-CMAQ forecasts consistently
underestimated ozone mixing ratios at both monitors, with nearly 68% and 88% of the NAM-CMAQ
ozone forecasts predicting less than 40 ppbv at Chiwaukee and Sheboygan KA, respectively. This result
is significantly higher than the percentages of ozone less than 40 ppbv observed at these two sites,
which are 40% and 45%, respectively. During southerly flow, ozone greater than 60 ppbv was observed
at both monitors for 12% of the LMOS 2017 period. In contrast, NAM-CMAQ forecasts predicted that
only 6% and 1% of the ozone during LMOS 2017 was greater than 60 ppbv at the Chiwaukee and
Sheboygan KA monitors, respectively.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
83
1682
1683
1684
1685
1686
1687
Figure 3.2.1. Wind Speed (m/s, upper) and Ozone (ppbv, lower) wind roses for 1-minute WDNR observations (left) and hourly
NAM-CMAQ (right) at the Chiwaukee, WI monitor during the period from May 22 through June 22, 2017. The overall
percentage of wind directions are indicated by the length of the wind rose in 20o bins. The percentage of wind speeds (in 3
m/s bins) and ozone (in 20 ppbv bins) are indicated by colors for each wind direction. The overall percentage of observed and
modeled wind speed and ozone at Chiwaukee during LMOS are indicated underneath the color bars.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
84
1688
1689
1690
1691
1692
1693
1694
1695
1696
Figure 3.2.2. Wind Speed (m/s, upper) and Ozone (ppbv, lower) wind roses for 1-minute WDNR observations (left) and hourly
NAM-CMAQ (right) at the Sheboygan KA monitor during the period from May 22 through June 22, 2017. The overall
percentage of wind directions are indicated by the length of the wind rose in 20o bins. The percentage of wind speeds (in 3
m/s bins) and ozone (in 20 ppbv bins) are indicated by colors for each wind direction. The overall percentage of observed and
modeled wind speed and ozone at Chiwaukee during LMOS are indicated underneath the color bars.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
85
1697
Comparison with Scientific Aviation profiles
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
Figure 3.2.3 shows statistical comparisons between Scientific Aviation (SA) and NAM-CMAQ
vertical profiles of wind direction, temperature, ozone and NO2 for all flight legs over Lake Michigan
during LMOS 2017. The observed and predicted median profiles are indicated by the solid and dotted
lines, respectively. The observed and predicted middle 50th percentiles (i.e., 25th to 75th percentile
values) of the distributions in 250 m altitude bins are shown by the yellow and red bars, respectively.
The “whiskers” show the middle 90th percentiles of the distributions in each 250 m altitude bin. The
NAM-CMAQ over-water wind direction predictions captured the observed transition from prevailing
southerly winds over the surface of the lake to prevailing westerly winds above 2 km. The model tended
to overestimate the frequency of westerly winds in the lowest 250 m and overestimate the frequency of
southerly winds between 1-1.5 km above the lake. NAM-CMAQ temperature predictions were
systematically about 2 degrees colder than observed but captured the median strength of the low level
inversion associated with the marine boundary layer. The NAM-CMAQ simulation captured the overall
variance in the observed temperatures throughout the profiles. Median ozone mixing ratio predictions
systematically underestimated observed values by 12-15 ppbv throughout the column and the observed
variance was underestimated below 1 km. Median NO2 predictions overestimated observed values by a
factor of 2 in the lowest 250 m and underestimated observed values by a factor of 2-10 above 1 km.
1714
1715
1716
1717
1718
1719
1720
1721
1722
Figure 3.2.4 shows statistical comparisons between SA and NAM-CMAQ vertical profiles of wind
direction, temperature, ozone and NO2 for all flight legs over Lake Michigan on June 2 and 11. The
median predicted wind direction profile showed less directional shear in the lowest 1.5 km and the
median predicted temperature profile was more stable than observed below 500 m. Median ozone
mixing ratio was systematically underestimated by NAM-CMAQ by over 20 ppbv throughout most of the
column. The largest underestimates in median ozone on these high ozone days occurred in the lowest
250 m where the median observed ozone mixing ratio was 90 ppbv and the median predicted ozone was
slightly larger than 65 ppbv. These large underestimates in median ozone were associated with a factor
of 10 overestimate in median NO2 mixing ratio in the lowest 250 m during the high ozone events.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
86
1723
1724
1725
1726
1727
1728
1729
Figure 3.2.3. Statistical comparison between Scientific Aviation (SA, yellow bars/solid line) and NAM-CMAQ (red bars/dotted
line) vertical profiles of wind direction (Degrees, upper left), temperature (K, upper right), ozone (ppbv, lower left), and NO 2
(ppbv, lower right) over Lake Michigan during LMOS. The solid and dotted lines show the median observed and predicted
median profiles, respectively. The yellow and red “bars” show the 50th percentiles of the observed and predicted
distributions within each 250 m layer, respectively. The “whiskers” show the 90th percentiles within each 250 m layer.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
87
1730
1731
1732
1733
1734
1735
1736
1737
1738
Figure 3.2.4. Statistical comparison between Scientific Aviation (SA, yellow bars/solid line) and NAM-CMAQ (red bars/dotted
line) vertical profiles of Wind Direction (Degrees, upper left), Temperature (K, upper right), Ozone (ppbv, lower left), and
NO2 (ppbv, lower right) over Lake Michigan on June 02 and June 11, 2017. The solid and dotted lines show the median
observed and predicted median profiles, respectively. The yellow and red “bars” show the 50th percentiles of the observed
and predicted distributions within each 250 m layer, respectively. The “whiskers” show the 90th percentiles within each
250m layer.
1739
1740
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
88
1741
3.3. WRF-Met Modeling
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
The suite of in situ meteorological observations made during LMOS 2017 are being used to test
and inform future modeling of weather within the region, the results of which will also be used in
modeling air quality. The current EPA meteorological modeling platform uses the Weather Research and
Forecasting (WRF) model (Skamarock and Klemp, 2008), configured with Morrison microphysics
(Morrison et al, 2009), Kain-Fritsch cumulus parameterization (Kain, 2004), ACM2 boundary layer
physics (Pleim, 2007), and Pleim-Xiu surface physics (Xiu and Pleim, 2001). This WRF configuration will
be referred to here as the “EPA baseline”. Beyond the parameterizations used for solving key physics in
the meteorological model, the use of data assimilation for initial conditions, boundary conditions, lake
temperature, soil moisture and other variables has been explored by the LMOS 2017 team and reported
herein.
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
Figure 3.3.1. Modeling domains used in the first round of meteorological parameterization sensitivity studies. The outer
domain has a horizontal resolution of 12 km by 12 km; the inner domain has a horizontal resolution of 4 km by 4 km.
Nested model simulations were performed with WRF (Figure 3.3.1) for the period May 21-June 3
2017. This period was selected for modeling because it included a notable lake breeze and ozone
exceedances on June 2. The outer domain has a horizontal resolution of 12 km by 12 km and covers the
continental U.S., the inner domain has a horizontal resolution of 4 km by 4 km. With one-way nesting,
both 12 km resolution results, and 4-km resolution results are available for in inner domain. This enables
comparison of the effect of finer resolution. Though fairly well correlated with observations of
temperature and wind, the EPA baseline WRF configuration was seen to overestimate daytime high
temperatures, underestimate nighttime low temperatures, and underestimate wind speed. Figure 3.3.2
and Table 3.3.1 show these biases for the Chiwaukee monitor and averaged across several coastal sites,
respectively. The EPA-baseline configuration also does not capture the timing of the lake breeze in
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
89
1766
1767
Chiwaukee (Figure 3.3.2, middle panel), nor mixed layer heights observed at Zion, which were
underestimated at night and overestimated during the daytime (Figure 3.3.3).
1768
1769
1770
Figure 3.3.2. Temperature (top), wind direction (middle), and wind speed (bottom) as observed (black lines) and modeled
(red lines) at Chiwaukee.
1771
1772
1773
1774
1775
Table 3.3.1. Statistical comparison of WRF (EPA baseline configuration) with observed meteorology at near-lakeshore sites.
Average statistics across all near-lakeshore sites (Sheboygan KA, Sheboygan Haven, Newport, Milwaukee SER, Manitowoc,
and Chiwaukee) and Lake Geneva, for the period May 23-June 3 2017. Note: biases are calculated as observation-model, a
low model bias is positive.
Mean Bias
12 km
4 km
temperature
wind speed
wind direction
0.740 K
-1.415 m/s
7.466 º
0.289 K
-1.390 m/s
6.832 º
Correlation
12 km
4 km
0.784
0.546
0.758
0.794
0.571
0.771
1776
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
90
1777
1778
1779
Figure 3.3.3. Average mixed layer height (meters) as observed at Zion (black line) and modeled using the EPA baseline WRF
configuration (red line).
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
Three other WRF configurations were tested, as detailed in Table 3.3.2. The first of the alternate
configurations is referred to by the name MYJ-NOAH.1 The second of the alternate configurations is
referred to as YSU-NOAH.2 By comparing the EPA baseline, MYJ-NOAH, and YSU-NOAH results, the
sensitivity of the modeled meteorology to surface and boundary layer physics can be assessed. A fourth
configuration, referred to as YSU-NOAH-Thompson, was used to isolate the impact of cloud and
precipitation (i.e. microphysics) parameterization from the YSU-NOAH sensitivity test. Overall, these
sensitivity experiments were all well-correlated with observed temperatures, but had a low bias in
temperature. The sensitivity runs were well-correlated for wind direction and speed, but wind speed
had an overall high bias compared to observations. Example summary statistics are provided for all
lakeshore sites in Tables 3.3.3 (temperature), 3.3.4 (wind direction), and 3.3.5 (wind speed). All
sensitivity experiments underestimated mixing layer heights at Zion, and the sensitivity to microphysics
parameterization was small. The sensitivity experiments also showed improved correlation with a higher
horizontal resolution.
1793
MYJ refers to the Mellor-Yamada-Janjic Planetary Boundary Layer (PBL) Physics Scheme, while NOAH refers to the
selection of the surface (land) physics option.
2 YSU refers to the Yonsei University scheme for PBL physics.
1
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
91
1794
Table 3.3.2. WRF sensitivity experiments relative to the EPA baseline.
microphysics
Morrison
cumulus
parameterization
Kain-Fritsch
MYJ-NOAH
Morrison
Kain-Fritsch
YSU-NOAH
YSU-NOAHThompson
Morrison
Kain-Fritsch
boundary
layer
ACM2
MYJ Eta
Operational
YSU
Thompson
Kain-Fritsch
YSU
configuration
EPA baseline
1795
1796
1797
1798
1799
1800
surface layer
physics
Pleim
surface
physics
Pleim-Xu
Monin-Obukov
NOAH
Monin-Obukov
NOAH
Monin-Obukov
NOAH
Table 3.3.3. Statistical comparison (temperature) of WRF (alternate configurations) with observed meteorology at nearlakeshore sites. Statistics for three sensitivity experiments’ temperature averaged across all near-lakeshore sites (Sheboygan
KA, Sheboygan Haven, Newport, Milwaukee SER, Manitowoc, and Chiwaukee) and Lake Geneva, for the period May 23-June
3 2017. Note: biases are calculated as observation-model, a low model bias is positive.
MYJ-NOAH
YSU-NOAH
YSU-NOAHThompson
Mean Bias
12 km
4 km
0.384 K
0.591 K
0.110 K
0.426 K
-0.157 K 0.131 K
Correlation
12 km
4 km
0.833
0.858
0.826
0.834
0.806
0.805
1801
1802
1803
1804
1805
1806
Table 3.3.4. Statistical comparison (wind direction) of WRF (alternate configurations) with observed meteorology at nearlakeshore sites. Statistics for three sensitivity experiments’ wind direction averaged across all near-lakeshore sites
(Sheboygan KA, Sheboygan Haven, Newport, Milwaukee SER, Manitowoc, Lake Geneva, and Chiwaukee) for the period May
23-June 3 2017. Note: biases are calculated as observation-model, a low model bias is positive.
MYJ-NOAH
YSU-NOAH
YSU-NOAHThompson
Mean Bias
12 km
4 km
6.916 º
9.207 º
3.129 º
4.711 º
5.473 º
6.101 º
Correlation
12 km
4 km
0.702
0.732
0.731
0.753
0.736
0.744
1807
1808
1809
1810
1811
1812
Table 3.3.5. Statistical comparison (wind speed) of WRF (alternate configurations) with observed meteorology at nearlakeshore sites. Statistics for three sensitivity experiments’ wind speed averaged across all near-lakeshore sites (Sheboygan
KA, Sheboygan Haven, Newport, Milwaukee SER, Manitowoc, and Chiwaukee) and Lake Geneva, for the period May 23-June
3 2017. Note: biases are calculated as observation-model, a low model bias is positive.
MYJ-NOAH
YSU-NOAH
YSU-NOAHThompson
Mean Bias
12 km
4 km
-2.419 m/s -2.588 m/s
-1.809 m/s -1.874 m/s
-1.862 m/s -1.913 m/s
Correlation
12 km
4 km
0.505
0.527
0.511
0.539
0.486
0.514
1813
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
92
1814
1815
1816
1817
1818
1819
1820
1821
1822
The EPA-baseline and sensitivity experiments all assumed a constant surface temperature for
Lake Michigan, reflecting default sea surface temperature (SST) settings in WRF. As shown in Figure
3.3.4, the assumption of constant SST (top row) in WRF is not consistent with data from the Great Lakes
Surface Environmental Analysis (GLSEA, bottom row). Since SST impacts the timing and extent of lake
breeze circulations in the model, future LMOS 2017 WRF modeling will include the high-resolution
satellite-based SST observations from GLSEA. Future work will also include investigating the impact of
both vertical and horizontal resolution on model performance. Based on discussions with stakeholders
and researchers, all future sensitivity simulations will employ an expanded 4-km domain to cover all
LADCO states and a 1.3 km by 1.3 km domain focused on Lake Michigan, as shown in Figure 3.3.5.
1823
1824
1825
Figure 3.3.4. SST values output from WRF using EPA baseline configuration (top), and from gridded analysis from the GLSEA
dataset (bottom).
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
93
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
Figure 3.3.5. Modeling domains used for ongoing meteorological parameterization sensitivity studies. The outer domain has
a horizontal resolution of 12 km by 12 km; the inner domain covering LADCO partner states has a horizontal resolution of 4
km by 4 km; the domain covering Lake Michigan has a horizontal resolution of 1.3 km by 1.3 km.
Figure 3.3.6 illustrates the sensitivity of the simulated lake breeze circulation to the land surface
model (LSM) formulations for May 27, 2017. This result is from the University of Wisconsin WRF
modeling effort, using the 1.3-km resolution domain. The VIIRS true color image (center panel) shows a
sharp transition from clear to fair weather cumulus along the eastern and western shores of Lake
Michigan, clear evidence of a strong lake breeze circulation with fairly deep inland penetration on both
sides of the lake. The EPA Baseline configuration uses the Pleim-Xiu LSM (see Table 3.3.2). The predicted
boundary layer depth from the EPA baseline (left panel) shows only slight penetration of the shallow
marine boundary layer air onshore while the WRF sensitivity experiment using the Noah LSM shows
much more inland penetration, which is in better agreement with the lake breeze penetration inferred
from the VIIRS image. The University of Iowa WRF-Chem sensitivity experiments have also
independently shown improved meteorological skill with the NOAH land surface model.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
94
1842
1843
1844
Figure 3.3.6. VIIRS true color image (center) with 1.3-km resolution planetary boundary layer height forecasts for the EPA
Baseline (Pleim-Xiu) LSM (left) and WRF Noah LSM sensitivity experiment (right) at 19Z on May 27, 2017
1845
1846
1847
1848
Use of satellite and surface observations to constrain the land surface (soil moisture, soil
temperature, leaf area index) and water surface (sea and lake surface temperatures) is an important
contributor to meteorological modeling skill. Options for improving lake surface temperatures, soil
moisture, and vegetation fraction, are explored below.
1849
1850
1851
High resolution 1.8-km resolution GLESA daily SST analyses
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
Figure 3.3.7 compares the NCEP Global Data Assimilation System (GDAS) 1/12th degree SST
analysis used in the EPA baseline configuration with the high-resolution SST analyses from the NOAA
Great Lakes Environmental Research Laboratory (GLERL) Great Lakes Surface Environmental Analysis
(GLSEA) (Schwab et al, 1999) for 27 May 2017. The GLSEA SST analysis shows much colder lake
temperatures along the northwestern shore of Lake Michigan. Improving the resolution of pockets of
colder lake temperatures will lead to better depictions of atmospheric stability within the Lake Michigan
boundary layer. More stable layers due to cold SSTs would lead to shallower boundary layer depths and
more concentrated ozone precursors. Stronger land/water thermal gradients associated with these
colder lake temperatures would strengthen the lake breeze circulation and could play an important role
in ozone predictions along the lakeshore by leading to more extensive transport of pollutants onshore,
and potentially greater inland lake breeze penetration.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
95
1863
1864
Figure 3.3.7. 1/12th degree GDAS (left) and 1.8km GLSEA (right) Lake Michigan SST (°C) analyses on May 27, 2017.
1865
1866
1867
NASA SPoRT-LIS Soil Moisture, Soil Temperature, and Green Vegetation Fraction/Leaf Area Index
Analyses
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
Accurate constraints on soil moisture are necessary to capture the depth of the planetary
boundary layer and near surface temperatures through the impact on surface sensible and latent heat
fluxes. The frequency of MDA8 values greater than 70 ppbv rises sharply beginning in mid-May and
peaks in late June (LADCO, 2016). The timing of the ozone peaks corresponds to the leaf-out period in
the upper Midwest. Leaf-out is associated with enhanced evapotranspiration and leads to rapid
reductions in the Bowen ratio, which in turn affects the boundary layer temperature and humidity
profiles (Fitzjarrald et al, 2001). Weekly variations in vegetation distributions can play a major role in the
accuracy of predicted isoprene emissions. Huang et al (2017) showed that use of NASA Soil Moisture
Active Passive (SMAP) soil moisture retrievals assimilated into the NASA Short-term Prediction Research
and Transition Center (SPoRT) Land Information System (SPoRT-LIS) analyses reduced biases in nearsurface air temperatures and boundary layer heights over the Missouri Ozarks during the NASA SEAC4RS
campaign, which, along with improved estimates of green vegetation fraction from the SPoRT-LIS
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
96
1880
1881
1882
1883
analyses, had a significant impact on biogenic isoprene emissions in the region. Unique feedbacks
between crops and meteorology may exist due to heavy agricultural intensity in the region, with large
changes in crop leaf area during the early ozone season (Alter et al., 2018). Ability of these data
assimilation products at capturing meteorological effects of crops should be assessed.
1884
1885
1886
1887
WRF simulations in the LADCO region should use the SMAP soil moisture analyses to constrain
soil moisture and MODIS Green Vegetation Fraction (GVF) retrievals to constrain the Leaf Area Index
(LAI) to improve the overall performance of the LADCO meteorological modeling framework in
simulating land/lake-breeze circulations planetary boundary layer depths, and surface fluxes.
1888
1889
Section 4.
Recommendations for Meteorological/Photochemical Modeling
1890
4.1. Photochemical Modeling
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
The existing LMOS 2017 WRF-Chem simulations and NWS NAM-CMAQ operational forecast
underestimated the maximum ozone concentrations during high ozone episodes. One take away
message from the model and airborne in situ ozone comparison done to date using WRF-Chem and the
NWS NAM-CMAQ forecast is that the models underpredict ozone in the marine boundary layer during
high coastal ozone events, and consistently underpredict ozone up to the maximum altitude of the
airborne in situ profiles (~3 km). This is a change from recent reports of ozone overprediction such as
those in Cleary et al. (2015) and citations therein. Cleary’s work was based on measurements over the
lake and at shore monitors during summer 2009. Coastal ozone on days (summer 2011 with NEI 2011
emissions) with MDA8 ozone in excess of 60 ppbv are also biased high in the WRF-CMAQ5.2 modeling
by Georgia Tech, UAH, and EPRI (Qin et al., 2019). Recommended modifications to the model
configuration in Qin et al. (MEGAN biogenics, reduction in NOx by 30%) decrease the bias at coastal
sites, and result in underprediction at inland sites. We conclude that, while the reasons are not fully
understood, the overall evolution of these modeling systems seems to have shifted the typical error
from overprediction to underprediction of coastal ozone.
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
Most sensitivity tests performed to date (RACM-ESRL vs. MOZART-GOCART chemical mechanism;
doubling of biogenic emissions) in the WRF-Chem system have had fairly limited leverage on these peak
ozone levels. One notable exception would be sensitivity testing using a 5 times multiplier for
anthropogenic hydrocarbon emissions shows increased ozone concentrations, especially in the Chicago
plume most influenced by those hydrocarbon emissions. This in turn improves WRF-Chem peak surface
ozone bias substantially for the June 2 episode. This result should not be interpreted as a conclusion that
the NEI 2011 hydrocarbon emissions are low by a factor of 5 (relative to actual emissions from summer
2017). It is likely that part of the observed sensitivity is compensation of other model errors (i.e.
meteorology, mixing, chemistry) through emissions. Much more extensive model-model
intercomparison, model-observation comparison, and model process studies would be needed to justify
broad changes in the emission inventory. An additional caveat is that our emissions have not been
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
97
1916
1917
verified against NEI emissions; while a VOC-specific error in translating NEI emissions to model-ready
emissions is unlikely, we have not explicitly ruled it out.
1918
1919
1920
1921
1922
1923
1924
The high sensitivity does indicate, however, that within the WRF-Chem system tested, a more
reactive plume (with higher HO2 radical concentrations) downwind of Chicago helps the model
performance for ozone. These sensitivity studies are ongoing and specific modeling recommendations
for air quality management agency modelers are not yet available. A general focus on the evaluation of
hydrocarbon emissions (including speciation and spatio-temporal release patterns) by regional air
managers, and by LMOS 2017 scientists, seems warranted based on the information on sensitivity and
bias that we have to date (McDonald et al., 2018).
1925
1926
1927
1928
1929
Additional comparisons to column loadings from remote sensing (GeoTASA, OMI, and Pandora), to
vertically resolved meteorological observations, to ship-borne observations, and to speciated
hydrocarbons, as well as additional sensitivity studies (for meteorological and chemical factors) are
continuing. Using an updated emission inventory (e.g. NEI-2014 and NEI-2017) should also be explored
in future modeling simulations.
1930
4.2. Meteorological Modeling
1931
1932
1933
1934
1935
1936
Simulations at 4 km horizontal resolution using standard WRF physics settings common to many
air quality modeling applications successfully captured general meteorological features and some
aspects of the lake breeze. These were completed with WRF-Chem v3.6.1 (University of Iowa) and WRF
v3.8 (University of Wisconsin). However, some shortcomings in the meteorology produced in the initial
WRF-Chem and WRF simulations can be improved with better settings, improved horizontal resolution,
and use of more accurate lake and land surface conditions.
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
•
•
•
Both the Iowa group and the Wisconsin group found decreased bias in temperature from using
the NOAH LSM rather than the 5-layer thermal diffusion LSM used in the base case runs.
Sensitivity to PBL model (MYJ and YSU were tested by Wisconsin; MYNN3 was used by Iowa)
was relatively small. No single PBL model had across-the-board improvement in metrics of
meteorological skill.
Sensitivity to microphysics was relatively small. The University of Wisconsin compared Morrison
and Thompson microphysics.
Tests are ongoing with respect to the use of improved initial and boundary conditions for
atmospheric and surface variables. These improved initial and boundary conditions achieve improved
accuracy due to extensive data assimilation of remote sensed and surface observations. The University
of Iowa is testing the NOAA High-Resolution Rapid Refresh (HRRR) model as an initial and boundary
condition for meteorology. The University of Wisconsin is testing influence of lake surface temperature
(GLSEA vs. GDAS), and the NASA SPoRT-LIS soil moisture, soil temperature, green vegetation fraction,
and leaf area index product. Improved accuracy of these model inputs is expected to improve
meteorological model performance, but the quantitative assessment of that is ongoing.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
98
1952
1953
Recommendations on vertical resolution are also under investigation and should be reported in
future LMOS reports or publications.
1954
1955
1956
1957
1958
1959
While statistical metrics comparing 4 km and 12 km results for meteorology are mixed, LMOS
stakeholders agree that 1.3 km horizontal resolution is necessary to resolve features of the lake
breeze, and the interaction of fine meteorological features with concentrated sources of ozone
precursors. The 1.3-km resolution nest is needed to resolve the large gradients in surface ozone
concentrations that often occur along the Lake Michigan shoreline. The 1.3km domain is also needed to
capture the inland penetration of the lake breeze circulation.
1960
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
99
1961
References
1962
1963
1964
1965
Ahmadov, Ravan, S. McKeen, M. Trainer, R. Banta, A. Brewer, S. Brown, P. M. Edwards, et al. 2015.
“Supplement of Understanding High Wintertime Ozone Pollution Events in an Oil- and Natural GasProducing Region of the Western US.” Atmospheric Chemistry and Physics 15 (1): 411–29.
doi:10.5194/acp-15-411-2015.
1966
1967
1968
Alter, R.E., Douglas, H.C., Winter, J.M., Eltahir, E.A.B., 2018. Twentieth Century Regional Climate Change
During the Summer in the Central United States Attributed to Agricultural Intensification. Geophysical
Research Letters 45, 1586-1594.
1969
1970
1971
Appel, K. W., et al. "Description and Evaluation of the Community Multiscale Air Quality (CMAQ)
Modeling System Version 5.1." Geoscientific Model Development 10.4 (2017): 1703-32. ProQuest. Web.
20 Dec. 2017.
1972
1973
1974
Bertram, T.H., Kimmel, J.R., Crisp, T.A., Ryder, O.S., Yatavelli, R.L.N., Thornton, J.A., Cubison, M.J., Gonin,
M., Worsnop, D.R. A field-deployable, chemical ionization time-of-flight mass spectrometer, Atmos.
Meas. Tech., vol. 4, no. 7, pp. 1471–1479, Jul. 2011.
1975
1976
Betts, A. K., 1986: A new convective adjustment scheme. Part I: Observational and theoretical
basis. Quart. J. Roy. Meteor. Soc., 112, 677–691.
1977
1978
Betts, A., and M. J. Miller, 1986: A new convective adjustment scheme. Part II: Single column tests using
GATE wave, BOMEX, ATEX and Arctic air-mass data sets. Quart. J. Roy. Meteor. Soc., 112, 693–709.
1979
1980
1981
Chai, T., and Coauthors, 2013: Evaluation of the United States National Air Quality Forecast Capability
experimental real-time predictions in 2010 using Air Quality System ozone and NO2 measurements.
Geosci. Model Dev.,6, 1831–1850, doi:10.5194/gmd-6-1831-2013
1982
1983
1984
1985
Cleary, P.A., Fuhrman, N., Schulz, L., Schafer, J., Fillingham, J., Bootsma, H., McQueen, J.,Tang, Y., Langel,
T., McKeen, S. and Williams, E.J., 2015. Ozone distributions over southern Lake Michigan: comparisons
between ferry-based observations, shoreline-based DOAS observations and model forecasts.
Atmospheric Chemistry and Physics, 15(9), pp.5109-5122.
1986
1987
Dye, T.S., P.T. Roberts, and M.E. Korc, 1995: Observations of transport processes for ozone and ozone
precursors during the 1991 Lake Michigan Ozone Study. J. App. Meteor, 34: 1877-1889.
1988
1989
1990
Ek, M. B., K. E. Mitchell, Y. Lin, E. Rogers, P. Grunmann, V. Koren, G. Gayno, and J. D. Tarpley, 2003:
Implementation of Noah LSM advances in the National Centers for Environmental Prediction operational
mesoscale Eta model. J. Geophys. Res.,108,8851, doi:10.1029/2002JD003296.
1991
1992
1993
Emery, C., E. Tai, and G. Yarwood (2001), Enhanced Meteorological Modeling and Performance
Evaluation for Two Texas Ozone Episodes, Final report submitted to Texas Natural Resources
Conservation Commission.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
100
1994
1995
1996
Ferrier, B. S., Y. Jin, Y. Lin, T. Black, E. Rogers, and G. DiMego, 2002: Implementation of a new grid-scale
cloud and precipitation scheme in the NCEP Eta Model. Preprints, 15th Conf. on Numerical Weather
Prediction, San Antonio, TX, Amer. Meteor. Soc., 280–283.
1997
1998
1999
Fitzjarrald, D.R., O.C. Acevedo, and K.E. Moore, 2001: Climatic Consequences of Leaf Presence in the
Eastern United States. J. Climate, 14, 598–614, https://doi.org/10.1175/15200442(2001)014<0598:CCOLPI>2.0.CO;2
2000
2001
Foley, T., et al., Lake Michigan air quality: The 1994e2003 LADCO Aircraft Project (LAP), Atmospheric
Environment (2011), doi:10.1016/j.atmosenv.2011.02.033
2002
2003
2004
2005
Goldberg, D.L., Loughner, C.P., Tzortziou, M., Stehr, J.W., Pickering, K.E., Marufu, L.T. and Dickerson,
R.R., 2014. Higher surface ozone concentrations over the Chesapeake Bay than over the adjacent land:
Observations and models from the DISCOVER-AQ and CBODAQ campaigns. Atmospheric environment,
84, pp.9-19.
2006
2007
2008
2009
Guenther, A. B., X. Jiang, C. L. Heald, T. Sakulyanontvittaya, T. Duhl, L. K. Emmons, and X. Wang. 2012.
“The Model of Emissions of Gases and Aerosols from Nature Version 2.1 (MEGAN2.1): An Extended and
Updated Framework for Modeling Biogenic Emissions.” Geoscientific Model Development 5 (6): 1471–
92. doi:10.5194/gmd-5-1471-2012.
2010
2011
Hanna, S.R., and J.C. Chang, 1995: Relations between meteorology and ozone in the Lake Michigan
region. J. Applied Meteorology, 34: 670-678.
2012
2013
2014
Haney, J.L., S.G. Douglas, L.R. Chinkin, D.R. Souten, C.S.Burton, and P.T. Roberts, 1989: Ozone Air Quality
Scoping Study for the Lower Lake Michigan Air Quality Region, SAI report #SYSAPP-89/101, prepared for
US EPA, August, 197 pp.
2015
2016
2017
2018
Herman, J., A. Cede, E. Spinei, G. Mount, M. Tzortziou, and N. Abuhassan (2009), NO2column amounts
from ground-based Pandora and MFDOAS spectrometers using the direct-sun DOAS technique:
Intercomparisons and application to OMI validation, J. Geophys. Res., 114, D13307,
doi:10.1029/2009JD011848.
2019
2020
2021
2022
Huang, M., G. R. Carmichael, J. H. Crawford, A. Wisthaler, X. Zhan,C. R. Hain, P. Lee, and A. B. Guenther,
2017, Biogenic isoprene emissions driven by regional weather predictions using different initialization
methods: case studies during the SEAC4RS and DISCOVER-AQ airborne campaigns, Geosci. Model Dev.,
10, 3085–3104, 2017, https://doi.org/10.5194/gmd-10-3085-2017
2023
2024
2025
Inness, A., F. Baier, A. Benedetti, I. Bouarar, S. Chabrillat, H. Clark, C. Clerbaux, et al. 2013. “The MACC
Reanalysis: An 8 Yr Data Set of Atmospheric Composition.” Atmospheric Chemistry and Physics 13 (8):
4073–4109. doi:10.5194/acp-13-4073-2013.
2026
Janjic, Z. I., 1990: The step-mountain coordinate: Physical package. Mon. Wea. Rev., 118, 1429–1443.
2027
2028
Janjic ́, Z. I., 1994: The step-mountain eta coordinate model: Furtherdevelopments of the convection,
viscous sublayer, and turbulence closure schemes. Mon. Wea. Rev., 122, 927–945.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
101
2029
2030
Janjic, Z. I., 2001: Nonsingular implementation of the Mellor-Yamada level 2.5 scheme in the NCEP Meso
model. NOAA/NWS/NCEP Office Note 437, 61 pp.
2031
2032
2033
Janjic, Z. I. and R. Gall, 2012: Scientific documentation of the NCEP Nonhydrostatic Multiscale Model on
the B grid (NMMB). Part 1 Dynamics. NCAR Tech. Note NCAR/TN-4891 STR, 75 pp. [Available online at
https://opensky.ucar.edu/islandora/object/technotes%3A502/datastream/PDF/view]
2034
2035
2036
Jayarathne, T., Stockwell, C.E., Yokelson, R.J., Nakao, S., Stone, E.A., 2014. Emissions of Fine Particle
Fluoride from Biomass Burning. Environmental Science & Technology 48, 12636-12644,
doi:10.1021/es502933j.
2037
2038
Kain, J. S., 2004: The Kain–Fritsch convective parameterization: An update. J. Appl. Meteor., 43, 170–
181.
2039
2040
2041
2042
Kim, S. W., A. Heckel, G. J. Frost, A. Richter, J. Gleason, J. P. Burrows, S. McKeen, E. Y. Hsie, C. Granier,
and M. Trainer. 2009. “NO2 Columns in the Western United States Observed from Space and
Simulated by a Regional Chemistry Model and Their Implications for NO x Emissions.” Journal
of Geophysical Research Atmospheres 114 (11): 1–29. doi:10.1029/2008JD011343
2043
2044
Kleinman, L.I., 2005. The dependence of tropospheric ozone production rate on ozone precursors.
Atmospheric Environment 39, 575-586.
2045
2046
2047
2048
Knuteson, R. O., F. A. Best, N. C. Ciganovich, R. G. Dedecker, T. P. Dirkx, S. Ellington, W. F. Feltz, R. K.
Garcia, R. A. Herbsleb, H. B. Howell, H. E. Revercomb, W. L. Smith, J. F. Short, 2004a: Atmospheric
Emitted Radiance Interferometer (AERI): Part I: Instrument Design, J. Atmos. Oceanic Technol.,21, 17631776
2049
2050
2051
2052
Knuteson, R. O., F. A. Best, N. C. Ciganovich, R. G. Dedecker, T. P. Dirkx, S. Ellington, W. F. Feltz, R. K.
Garcia, R. A. Herbsleb, H. B. Howell, H. E. Revercomb, W. L. Smith, J. F. Short, 2004b: Atmospheric
Emitted Radiance Interferometer (AERI): Part II: Instrument Performance, J. Atmos. Oceanic Technol.,21,
1777-1789
2053
2054
LADCO, 2016,
http://ladco.org/reports/ozone/post08/Great_Lakes_Ozone_Study_White_Paper_Draft_v6.pdf
2055
2056
2057
Leitch, J. W., T. Delker, W. Good, L. Ruppert, F. Murcray, K. Chance, X. Liu, C. Nowlan, S.J. Janz, N.A.
Krotkov, K.E. Pickering, M. Kowalewski, and J. Wang. 2014. “The GeoTASO airborne spectrometer
project.” Proc. SPIE 9218, 92181H-92181H-9.
2058
2059
2060
Loughner, Christopher P., Maria Tzortziou, Melanie Follette-Cook, Kenneth E. Pickering, Daniel Goldberg,
Chinmay Satam, Andrew Weinheimer et al. "Impact of bay-breeze circulations on surface air quality and
boundary layer export." Journal of Applied Meteorology and Climatology 53, no. 7 (2014): 1697-1713.
2061
2062
Lyons, W.A. and Cole, H.S., 1976. Photochemical oxidant transport: mesoscale lake breeze and synopticscale aspects. Journal of Applied Meteorology, 15(7), pp.733-743.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
102
2063
2064
2065
2066
McDonald, B.C., de Gouw, J.A., Gilman, J.B., Jathar, S.H., Akherati, A., Cappa, C.D., Jimenez, J.L., LeeTaylor, J., Hayes, P.L., McKeen, S.A., Cui, Y.Y., Kim, S.W., Gentner, D.R., Isaacman-VanWertz, G.,
Goldstein, A.H., Harley, R.A., Frost, G.J., Roberts, J.M., Ryerson, T.B., Trainer, M., 2018. Volatile chemical
products emerging as largest petrochemical source of urban organic emissions. Science 359, 760-764.
2067
2068
Mellor, G. L., and T. Yamada, 1982: Development of a turbulence closure model for geophysical fluid
problems. Rev. Geophys., 20, 851–875.
2069
2070
2071
Mlawer, E. J., S. J. Taubman, P. D. Brown, M. J. Iacono, and S. A. Clough, 1997: Radiative transfer for
inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys.
Res.,102, 16,663-16,682.
2072
2073
2074
Morrison, H., G. Thompson, V. Tatarskii, 2009: Impact of Cloud Microphysics on the Development of
Trailing Stratiform Precipitation in a Simulated Squall Line: Comparison of One– and Two–Moment
Schemes. Mon. Wea. Rev., 137, 991–1007.
2075
2076
2077
2078
National Centers for Environmental Prediction, National Weather Service, NOAA, U.S. Department of
Commerce. 2000. “NCEP FNL Operational Model Global Tropospheric Analyses, Continuing from July
1999.” Boulder: Research Data Archive at the National Center for Atmospheric Research, Computational
and Information Systems Laboratory. doi:10.5065/D6M043C6.
2079
2080
2081
2082
2083
2084
2085
Nowlan, C. R., Liu, X., Leitch, J. W., Chance, K., González Abad, G., Liu, C., Zoogman, P., Cole, J., Delker, T.,
Good, W., Murcray, F., Ruppert, L., Soo, D., Follette-Cook, M. B., Janz, S. J., Kowalewski, M. G., Loughner,
C. P., Pickering, K. E., Herman, J. R., Beaver, M. R., Long, R. W., Szykman, J. J., Judd, L. M., Kelley, P., Luke,
W. T., Ren, X., and Al-Saadi, J. A.: Nitrogen dioxide observations from the Geostationary Trace gas and
Aerosol Sensor Optimization (GeoTASO) airborne instrument: Retrieval algorithm and measurements
during DISCOVER-AQ Texas 2013, Atmos. Meas. Tech., 9, 2647-2668, doi:10.5194/amt-9-2647-2016,
2016.
2086
2087
2088
Odman, T., Qin, M., Hu, Y., and Russell, T. Model Over-Prediction of Ozone in the Great Lakes Region:
The Possible Role of Emissions and Impacts of Chemistry and Transport. Presentation at the LMOS 2017
Science Meeting, Sept. 2017, Chicago, IL.
2089
2090
2091
2092
2093
2094
O’Neill, S. M., Narasimhan (Sim) K. Larkin, Jeanne Hoadley, Graham Mills, Joseph K. Vaughan, Roland R.
Draxler, Glenn Rolph, Mark Ruminski, Sue A. Ferguson, Chapter 22 Regional Real-Time Smoke Prediction
Systems, Editor(s): Andrzej Bytnerowicz, Michael J. Arbaugh, Allen R. Riebau, Christian Andersen, In
Developments in Environmental Science, Elsevier, Volume 8, 2008, Pages 499-534, ISSN 1474-8177, ISBN
9780080556093, https://doi.org/10.1016/S1474-8177(08)00022-3.
(http://www.sciencedirect.com/science/article/pii/S1474817708000223)
2095
2096
2097
2098
Palmer, P. I., Jacob, D. J., Chance, K., Martin, R. V., Spurr, R. J., Kurosu, T. P., Bey, I., Yantosca, R., Fiore,
A., and Li, Q.: Air mass factor formulation for spectroscopic measurements from satellites: application to
formaldehyde retrievals from the global ozone monitoring experiment, J. Geophys. Res.-Atmos., 106,
14539–14550, doi:10.1029/2000JD900772, 2001.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
103
2099
2100
Pleim, J. E., 2007: A Combined Local and Nonlocal Closure Model for the Atmospheric Boundary Layer.
Part I: Model Description and Testing. J. Appl. Meteor. Climatol., 46, 1383–1395.
2101
2102
2103
Pouliot, G.: A Tale of Two Models: A Comparison of the Biogenic Emission Inventory System (BEIS3.14)
and Model of Emissions of Gases and Aerosols from Nature (MEGAN 2.04), 7th Annual CMAS
Conference, Chapel Hill, NC, USA, 7 October 2008.
2104
2105
2106
Pouliot, G. and Pierce, T. E.: Integration of the Model of Emissions of Gases and Aerosols from Nature
(MEGAN) into the CMAQ Modeling System, 18th International Emission Inventory Conference,
Baltimore, Maryland, 14–17 April 2009.
2107
2108
2109
2110
Qin, M. M., H. F. Yu, Y. T. Hu, A. G. Russell, M. T. Odman, K. Doty, A. Pour-Biazar, R. T. McNider and E.
Knipping, 2019, Improving ozone simulations in the Great Lakes Region: The role of emissions,
chemistry, and dry deposition. Atmospheric Environment, 202: 167-179, doi:
10.1016/j.atmosenv.2019.01.025.
2111
2112
Ragland, K. and P. Samson, 1977: Ozone and visibility reduction in the Midwest: evidence for large-scale
transport. J. Applied Meteorology, 16: 1101–1106.
2113
2114
2115
2116
2117
Schauer, J.J., Mader, B.T., DeMinter, J.T., Heidemann, G., Bae, M.S., Seinfeld, J.H., Flagan, R.C., Cary,
R.A., Smith, D., Huebert, B.J., Bertram, T., Howell, S., Kline, J.T., Quinn, P., Bates, T., Turpin, B., Lim, H.J.,
Yu, J.Z., Yang, H., Keywood, M.D., 2003. ACE-Asia Intercomparison of a Thermal-Optical Method for the
Determination of Particle-Phase Organic and Elemental Carbon. Environmental Science & Technology
37, 993-1001, doi:10.1021/es020622f.
2118
2119
2120
Schwab, D. J., G. A. Leshkevich, G. C. Muhr,1999, Automated Mapping of Surface Water Temperature in
the Great Lakes, In Journal of Great Lakes Research, Volume 25, Issue 3, 1999, Pages 468-481, ISSN
0380-1330, https://doi.org/10.1016/S0380-1330(99)70755-0.
2121
2122
2123
Shipley, K, S.T., D.H. Tracy, E.W. Eloranta, J.T. Trauger, J.T. Sroga, F.L. Roesler and J.A. Weinman, 1983, A
High Spectral Resolution Lidar to measure optical scattering properties of atmospheric aerosols , Part I:
Intrumentation and theory" Applied Optics, 23, 3716-3724, 1983.
2124
2125
Sillman, S., 1995, The Use of NOy, H2O2, And HNO3 As Indicators For Ozone-NOx-Hydrocarbon
Sensitivity In Urban Locations, Journal of Geophysical Research-Atmospheres 100, 14175-14188.
2126
2127
2128
2129
Sillman, S., He, D.Y., Pippin, M.R., Daum, P.H., Imre, D.G., Kleinman, L.I., Lee, J.H., Weinstein-Lloyd, J.,
1998. Model correlations for ozone, reactive nitrogen, and peroxides for Nashville in comparison with
measurements: Implications for O-3-NOx-hydrocarbon chemistry. Journal of Geophysical ResearchAtmospheres 103, 22629-22644.
2130
2131
2132
Skamarock, W. C. and Klemp, J. B., 2008, A time-split nonhydrostatic atmospheric model for
weather research and forecasting applications, J. Comput. Phys., 227(7), 3465-3485,
doi:10.1016/j.jcp.2007.01.037.
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
104
2133
2134
2135
Spinei, E., et al., Whitehill, A., Fried, A., et al., 2018. The First Evaluation of Formaldehyde Column
Observations by Pandora Spectrometers during the KORUS-AQ Field Study, Atmos. Meas. Tech. Discuss.,
https://doi.org/10.5194/amt-2018-57, in review.
2136
2137
2138
Spurr, R. J. D.: VLIDORT: A linearized pseudo-spherical vector discrete ordinate radiative transfer code
for forward model and retrieval studies in multilayer multiple scattering media, J. Quant. Spectrosc. Ra.,
102, 316–342, doi:10.1016/j.jqsrt.2006.05.005, 2006.
2139
2140
2141
2142
Stajner, I., P. Davidson, D. Byun, J. McQueen, R. Draxler, P. Dickerson, and J. Meagher, 2012: US National
Air Quality Forecast Capability: Expanding coverage to include particulate matter. Air Pollution Modeling
and Its Application XXI, D. G. Steyn and S. T. Castelli, Eds., Springer, 379–384, doi:10.1007/978-94-0071359-8_64
2143
2144
2145
Stockwell, William R., Frank Kirchner, Michael Kuhn, and Stephan Seefeld. 1997. “A New Mechanism for
Regional Atmospheric Chemistry Modeling.” Journal of Geophysical Research 102 (D22): 25847.
doi:10.1029/97JD00849.
2146
2147
2148
Tonnesen, G.S., Dennis, R.L., 2000. Analysis of radical propagation efficiency to assess ozone sensitivity
to hydrocarbons and NOx : 1. Local indicators of instantaneous odd oxygen production sensitivity.
Journal of Geophysical Research 105, 9213-9213.
2149
2150
2151
2152
Turner, D.D. and U. Löhnert, 2014: Information Content and Uncertainties in Thermodynamic Profiles
and Liquid Cloud Properties Retrieved from the Ground-Based Atmospheric Emitted Radiance
Interferometer (AERI). J. Appl. Meteor. Climatol., 53, 752–771, https://doi.org/10.1175/JAMC-D-130126.1
2153
2154
USEPA, 2016. 2015 Ozone Standards. Retrieved Dec 31, 2018 from
https://ozoneairqualitystandards.epa.gov/OAR_OAQPS/OzoneSliderApp/index.html#.
2155
2156
2157
2158
Wagner, Timothy J.; Smith, N.; Best, F.; Eloranta, E.; Feltz, W. F. and Olson, E.. SPARC: A mobile facility
for a new generation of atmospheric research. Symposium on Meteorological Observation and
Instrumentation,18th, New Orleans, LA, 10-14 January 2016. Boston, MA, American Meteorological
Society, 2016, abstract only; Abstract 390.
2159
2160
WDNR, 2017, 2017 Wisconsin Air Quality Trends Report AM-553,
https://dnr.wi.gov/files/PDF/pubs/am/AM553.pdf
2161
2162
Xiu, A., and J. E. Pleim, 2001: Development of a land surface model. Part I: Application in a mesoscale
meteorological model. J. Appl. Meteor., 40, 192–209.
2163
2164
Yarwood, G., Roa, S., Yocke, M., Whitten, G., 2005. Updates to the carbon bond chemical mechanism:
CB05. Final report to the US EPA, RT-0400675. Available at: /http://www.camx.com
2165
LMOS 2017 Preliminary Finding Report. Version 4/11/2019.
105