MEMORANDUM TO: CC: MARGARET MEDELLIN, CITY OF ASPEN FROM: GEORGE OAMEK DATE: JULY 3, 2017 SUBJECT: RISK ASSESSMENT OF ASPEN’S WATER SYSTEM Introduction The City of Aspen has contracted with Headwaters Corporation to develop a framework for assessing the risk of water shortages to the City, primarily utilizing existing studies and data, including the recent Water Supply Availability Study.1 This methodological framework has been developed and is currently being used to make preliminary assessments of the City’s ability to both provide water to its residents and meet bypass flow commitments, over a wide range of possible future conditions. The assessments are preliminary because some of the data is still in the development stage and many of the assumptions would benefit from additional discussion, including additional input from City staff, the City Council, and stakeholders. The wide range of future conditions address inherent uncertainties in year-to-year water supply, the potential impact of climate change on available water supply, and impact of alternative future development paths on the demand for water. Water Shortages and Risk The terms shortages, risk, and uncertainty are used repeatedly throughout this memorandum, as well as references to the method of Monte Carlo simulation. These terms and the methods require a brief background discussion. Water Shortages For this effort and in previous studies, a water shortage is defined as periods when flows in Castle and Maroon Creeks are insufficient to meet Aspen’s baseline municipal demand without cutting back supplies for downstream irrigation users or decreed instream flows. Previous studies have found that instream flows bear the brunt of most shortages except for the most severe years. The definition of a water shortage has traditionally been defined as periods when flows in Castle and Maroon Creeks are insufficient to meet Aspen’s baseline municipal demand without reducing minimum instream flows. Changing the definition of a water shortage would change the level and occurrence of shortages. Similar to previous studies, this effort focuses upon the number and severity of shortages during a representative hydrologic period of record. This period of record is discussed in subsequent sections. Although it is uncertain how severe shortages may be allocated between the City and instream usage, this analysis will indicate when a simulated shortage exceeds the instream flow decrees. 1 Wilson Water Group. 2016. City of Aspen Water Supply Available Study, 2016 Update. 405 Urban Street, Suite 401, Lakewood, CO 80228• (720) 524-6115 • Fax: (720) 524-6347 City of Aspen 7/03/17 The possibility of shortage is inherent in any water system, but requires greater focus for systems like Aspen’s which are greatly dependent on highly variable surface water flows. Many water providers use water storage to hedge against short-term disruptions, seasonal supply risks, as well as longer-term drought risks. Currently, Aspen’s raw water storage is very minimal and is more often expressed in hours of available water rather than in volume. Any sort of shortage or disruption to the raw water supply lasting more than about 12 hours will likely create a potable water shortage. Risk and Uncertainty Previous efforts examined risk in the following manner: • Water supply risk was addressed by a historic hydrologic period of record intended to represent the full range of possible good and bad years. • The risk of climate change was addressed in the Wilson Report by examining six alternative water supply scenarios incorporating various climate change impacts. These estimated impacts were considered most likely to occur when the data was developed, approximately 5 to 10 years ago. • Demand risk was assessed considering a number of alternative future growth scenarios. This effort supplements the Wilson Study in several ways intended to more accurately represent the full range of risk for shortages: • • • Water supply risk is initially assessed using the same hydrologic period of record. However, digging deeper into the hydrology indicated that some adjustments to the hydrology data introduced more uncertainty to the water supply than previously reported. This effort has explicitly examined these adjustments and is also using stochastic hydrology methods to potentially extend the data’s usefulness. A wider range of possible climate change impacts are being considered compared to the previous analysis. More contemporary climate change modeling suggests impacts could be significantly more severe than previously considered, especially in the Colorado River basin. Demand risk is focused upon the build-out of existing City water service areas and the likelihood of the City extending service to additional areas. Monte Carlo Simulation An alternative method to those historically used for assessing the range of supply, demand, and climate change possibilities is Monte Carlo simulation. Although the term Monte Carlo suggests a gaming application, this method of simulation has wide application and acceptance, including for water resources planning, financial planning, and energy exploration. The benefit of Monte Carlo simulation is its ability to simultaneously consider a large number of combinations and uncertainties, far more than the handful previously considered. As a greater number of combinations are created through Monte Carlo simulations, a statistical picture begins to develop regarding the probability and severity of shortages over the hydrological period of record. That is, how often does demand exceed supply given these various combinations of uncertain supply, demand, and climate change values. Headwaters Corporation Page 2 of 19 City of Aspen 7/03/17 How these combinations are “matched-up” depends on the assumptions made about the uncertain variables. Input values used in a Monte Carlo analysis are, in technical terms, probability-weighted because the analyst assigns probabilities to their frequency of occurrence. These probabilities describe how the variable might range around its estimated value. Some probabilities can be described with a normal, bell-shaped, distribution, meaning that it is equally likely that the value might fall below or above its estimated value. Figure 1, below, is a depiction of a normal distribution for a hypothetical example. As can be seen, the distribution is symmetric around the expected value of 180 in this example. Figure 1. Hypothetical Example of a “Normal” Statistical Distribution Figure 2 illustrates an alternative depiction of this variable as having skewed characteristics. The mean is the same, 180, but there is a higher probability that the value is higher than 180 than below it. Alternatively stated, the distribution in Figure 2 has a long tail, indicating that although the probability is small, a large impact is possible. Figure 2. Hypothetical Example of a Non-Normal Skewed Statistical Distribution Headwaters Corporation Page 3 of 19 City of Aspen 7/03/17 Variability in time series and cross-sectional data is often used to assist in developing these distributions, although informed judgment may also play a role when data is lacking. Many uncertainties in water planning are non-normal, or skewed, in nature because they are influenced by sometimes erratic weather patterns with periodic extreme events. Monte Carlo simulation is the best tool available for incorporating combinations of these skewed characteristics. Given assumptions about the uncertainties affecting Aspen’s water reliability and their statistical characteristics, what sort of output can be expected? Figure 3 is an example of the type of output that Monte Carlo simulation can create. It shows a hypothetical output that summarizes the number of shortages in excess of a given threshold, 1,000 acre-feet per year, over a 25-year period of record. As a result of variables that have non-normal distributions, the graphic shows that most of the time there are only 2 instances when shortages exceed 1,000 acre-feet over the period of record. However, it is much more likely that there will be more shortages of this magnitude rather than fewer. There may be as many as 7 or 8. Again, the example is hypothetical, but this type of data tells decision makers that reliance upon averages and most likely values does not always paint the full picture. Figure 3. Example Monte Carlo Output for Hypothetical Example Organization Although the focus of the analysis is intended to address long-term water supply and demand, this effort is also helpful in addressing shorter-term risks such as natural and man-made disruptions to the source water supply. However, longer-term water supply and demand issues are primarily discussed here. The exception are several operational issues that have been identified during the course of meeting with City staff. These issues are included after the Demand discussion. Current efforts have focused upon: Headwaters Corporation Page 4 of 19 City of Aspen 7/03/17 • • The quality of the data and assumptions used in the analysis Using sensitivity testing to identify the assumptions most significantly impacting the number and severity of possible future water shortages The remainder of this memorandum discusses the data and assumptions, and the results-to-date of the sensitivity analysis. The modeling framework itself is briefly summarized below, with the following discussion focusing upon water supply, climate change, and water demand. The Modeling Framework Similar to previous water supply assessments, the framework is based on a water balance model for each of the two basins providing the City’s water supply, Castle Creek and Maroon Creek. To summarize, a portion of the water flowing at the City’s diversion points in each creek is diverted to meet the City’s instantaneous demands, with the rest bypassed to meet instream flow obligations and downstream demands, prior to the confluence with the Roaring Fork River. However, the differences between the modeling framework developed in this analysis and that used in previous assessments are significant. • • • The current effort incorporates a weekly time step rather than a monthly time step, such as used in the Wilson Report. This finer resolution allows the model to better incorporate periodic peak demands upon the system, including those experienced during hot summer periods, major events, and major holidays. A demand component has been added that examines water demand by types and timing of usage. Critical assumptions affecting water supply and demand have been isolated and examined in a probabilistic framework using Monte Carlo simulation, as described above. This recognizes that significant uncertainties exist about the impact of future events, most notably the impact of climate change on the timing and volume of run-off, and impact of alternative demand scenarios. Also, uncertainties in the hydrologic period of record underlying the previous and current analyses are also closely examined. Water Supply The water supply component of this effort involved assessment of the hydrologic data, and the potential impact of climate change. A benchmarking study was also developed and is discussed in one of the latter sections of this memorandum. Hydrology Data The accuracy and sufficiency of the hydrologic dataset historically used for this analysis is questionable and it is uncertain whether the historic period of record adequately captures the variability of annual streamflow. This statement is based on relatively little raw streamflow data on Castle and Maroon Creeks and inconsistencies in location and timing of measurements, reducing the usefulness of the available data. Headwaters Corporation Page 5 of 19 City of Aspen 7/03/17 The hydrologic dataset is limited to the years 1970 through 1994, the only active years of water measuring gauges on Castle and Maroon Creek, respectively.2 These The quantity and quality of gauges were located high in the basins, a significant distance from the City’s water treatment plant diversion points, with a number of hydrologic data for Maroon and intervening creeks and diversions along the way. As a result, Castle Creek add considerable adjustments had to be made to the upstream gauge data to reflect uncertainty to flow projections. these creeks and diversions and approximate flow at the City’s diversions. These adjustments have previously been made with either single value scalars or linear regression equations. In terms of magnitude, and based on a limited number of data points, the firm Enartech and later Grand River Consulting estimated that these factors were in the range of 2.3 to 2.65 for Castle Creek and in the range of 1.25 to 1.4 for Maroon Creek, depending on the method used. That is, flows at the City’s Castle Creek and Maroon Creek diversions were 2.3 to 2.65 times the upstream gauged Castle Creek flow, and 1.25 to 1.4 times gauged flows for Maroon Creek. The 1994 Enartech analysis assumed the factors were 2.3 for Castle Creek and 1.25 for Maroon Creek. Later, the Wilson Report incorporated factors that were in the range of 2.65 and 1.40 for Castle Creek and Maroon Creek, respectively.3 The firm AMEC examined the hydrology of Castle and Maroon Creeks as part of a 2011 evaluation of a potential Castle Creek hydropower facility and made the same observations as those above.4 In response to their concerns about the adjustment factors, they re-estimated the factors using two alternative methods: a drainage area approach and the use of resident USGS software that estimates flows on non-gauged waterways. The drainage area method resulted in factors of 2.16 and 1.18 for Castle Creek and Maroon Creek, respectively. The USGS regional model indicated that for Castle Creek, these factors ranged from less than 2.00 to about 2.2, depending on the month considered. For Maroon Creek, these factors ranged from 1.06 to 1.10. Table 1 summarizes the differences in these adjustment factors. 2 In 2012, a new gauge was placed on Castle Creek below the City’s water treatment plant diversion. Unfortunately, due to a lack of funding, this gauge is no longer operating. 3 The adjustment process was severely limited by lack of data and when it was collected. Only eight comparable, paired measurements from Castle Creek and one paired measurement on Maroon Creek were usable in the analysis due to adverse conditions during data collection. In addition, these measurements were made during an above average baseflow period, following a very wet year, which may have tended to bias estimates upward with respect to flow. Stream flow occurring during critical drought periods has not been measured at the City intake facilities on either Castle Creek or Maroon Creek. Enartech fully appreciated these shortcomings and conservatively used minimums of their measurements to reduce the possibility of over-estimating available water, but observed that “if actual dry year discharge is less than projected … the amount of water available for diversion by the City may have been overestimated in this study.” The gravity of their statement reflects that the City’s future water resource decisions could partially depend on the accuracy of a scalar developed from a handful of observations taken during a wetter than average year. 4 AMEC. Preliminary Review of City of Aspen’s Proposed Castle Creek Hydroelectric Project. Prepared for Pitkin County, Colorado January 21, 2011. Headwaters Corporation Page 6 of 19 City of Aspen 7/03/17 Table 1. Range of Flow Adjustment Factors for Castle and Maroon Creeks Source of factor Wilson Report, 2016, citing a linear regression method Enartech, 1994, using limited paired data AMEC, 2011, using a drainage area method AMEC, 2011, using USGS procedures Castle Creek Maroon Creek 2.65 1.4 2.30 1.25 2.16 1.18 Ranges by month between 1.85 and 2.36 Ranges by month between 1.02 and 1.11 The differences in these adjustment factors across methods have significant implications. Using the higher factors, the Wilson Report concluded that under baseline 2064 demands and no climate change, Aspen’s water supply was adequate to meet all potable and instream demands. No shortages were shown during the 1970 to 1994 period of record. Figure 4 shows that if an alternative set of adjustment factors are used, 2.10 and 1.10 for Castle and Maroon Creeks, respectively, shortages appear in over 10 of the 25 years during the 1970 to 1994 period of record. With the possible exception of 1977, these shortages apply to instream flows and not potable water supplies. However, Figure 4 illustrates that equally reasonable assumptions about flow adjustments can significantly affect results. The Wilson Report did not mention the AMEC effort or the impact of the uncertainty surrounding the flow adjustment factors. Figure 4. Shortages to Instream Flows Using Alternative Adjustment Factors for Castle and Maroon Creek Flows. Note: the Wilson Report reported no shortages with all other assumptions remaining the same. City of Aspen Annual Water Supply Shortage, WY 1970-1994 2,000 1,800 Shortage Volume [AF] 1,600 1,400 1,200 1,000 800 600 400 200 0 Year Annual Shortage Volume Headwaters Corporation Average Annual Shortage Volume Page 7 of 19 City of Aspen 7/03/17 The adjustment factors are clearly important and as such the current effort incorporate their uncertainties. In response, the flow adjustment factors have been defined as a random variable in future Monte Carlo simulations and the following preliminary assumptions have been made about how they may vary (Table 2). Table 2. Assumptions about Flow Adjustment Factors Incorporated into the Uncertainty Analysis Parameter Castle Creek Maroon Creek Most likely value 2.30 1.25 Low range 2.16 1.05 High range 2.65 1.40 The assumptions in Table 2 translate to a slightly skewed Triangular shape of statistical distribution, shown for Castle Creek in Figure 5, below. It implies that in most simulations, the flow adjustment factor is most likely to be near 2.3, but it will vary between 2.16 and 2.65. The Triangular distribution was assumed because it requires only three points to be identified, the low, most likely, and high values. However, this assumption may be refined as more flow-related data is reviewed and developed. Figure 5. Assumed Statistical Distribution for the Castle Creek Flow Adjustment Factor Period of Record The years 1970 through 1994 were described by the 1994 Report as containing the most severe years on record, but goes on to observe that severely dry years did not occur in succession, like they may have Headwaters Corporation Page 8 of 19 City of Aspen 7/03/17 during the 1950’s.5 Nor does the data contain the years since 2000, when Colorado has experienced statistically significant higher average temperatures compared to 1970 through 1994. This increase is shown in Figure 6. Also, since the period 1970 through 1994 mostly pre-dates the establishment of statistical trends showing warming in Colorado, any climate change-based impacts occurring between 1994 and the present are likely not reflected in the data. Uncertainties associated with the limited period of record are considered by varying the range of flow adjustment factors. Extending the hydrological record to include the entire 1950 through present period is desirable in order to better quantify the variability of flows, the frequency of critical years, and the possibility of successive critical years. However, for this analysis, more extreme and frequent hydrological events are being simulated to test the sensitivity of the period of record. Figure 6. Colorado average annual temperatures (F) from 1900 to 2012. Note: annual temperatures are shown as departures from a 1971-2000 reference period. The orange, red, and dark-red lines are 100-year, 50-year, and 30-year trends in temperatures, respectively. All three are statistically significant. (Source: NOAA NCDC) Climate Change An important component of this effort is to assess the possible impacts of climate change. Recall that the Wilson Report examined 5 climate change scenarios with varying levels of impact to flow patterns, concluding that climate change would adversely affect Aspen’s water supply, but not to a level requiring 5 Precipitation data and snowfall data from the Aspen Station indicate that all years but one between 1952 and 1958 had less total precipitation than 1977. The year 1953 had substantially less snowfall than the 1977 water year, but the remaining years had more snowfall than 1977. Headwaters Corporation Page 9 of 19 City of Aspen 7/03/17 additional infrastructure, such as a storage reservoir. Since the development of these 5 climate change scenarios, there have been questions as to whether a wider range of impacts should have been considered, especially those based on the greater resolution provided by more recent research. Regardless of the answer, it should be noted that much of this recent research has not yet been downscaled to a level applicable to Castle or Maroon Creeks, or the Roaring Fork Valley. Although various efforts are underway at the State and major water provider level to adapt this data at a basin level, it is not likely to be available in the immediate future.6 More recent climate change research indicates that impacts to flows in the Colorado River and its tributaries may be much more severe than previously thought. Climate change-induced changes to flow volumes on Castle and Maroon Creeks are shown in the Wilson Report to range from about +9% to 19%. Recent research suggests the following: “Recently published estimates of Colorado River flow sensitivity to temperature combined with a large number of recent climate model-based temperature projections indicate that continued business-asusual warming will drive temperature-induced declines in river flow, conservatively −20% by midcentury and −35% by end-century, with support for losses exceeding −30% at midcentury and −55% at endcentury. Precipitation increases may moderate these declines somewhat, but to date no such increases are evident and there is no model agreement on future precipitation changes. These results, combined with the increasing likelihood of prolonged drought in the river basin, suggest that future climate change impacts on the Colorado River flows will be much more serious than currently assumed, especially if substantial reductions in greenhouse gas emissions do not occur”.7 Climate change and its uncertain impacts will ultimately affect Castle and Maroon Creeks through the timing and quantity of their future flows. Snowmelt run-off will likely occur earlier in the year over time and the total volume of flow may or may not decline over time. These impacts will be reflected in their hydrographs that show flows over the course of a representative year, at a specific point in the basin. It should be noted that the existing hydrographs account for existing water use, or evapotranspiration (ET), based on current upstream land uses. With warming associated with climate change, upstream ET will likely increase and further impact the resulting hydrograph, regardless of the precipitation impacts. This increase in ET may also affect Aspen’s customers through an increase in outdoor water demand. To assess overall possible impacts of climate change to the hydrographs, a utility was embedded in the modeling framework which allows the model user to specify changes to the hydrographs’ timing and shape. By specifying 3 variables related to the timing of peak flows and the impact to total flow volume, a wide range of potential impacts can be considered. Figure 7 illustrates the baseline hydrograph for Maroon Creek and a modified hydrograph based on a single set of alternative assumptions about longterm timing and flow impacts associated with climate change. Sensitivity analysis using Monte Carlo simulation has been used to examine a large number of combinations of timing and flow impacts.8 6 http://onlinelibrary.wiley.com/doi/10.1002/joc.4594/full 7 Udall, et al. http://onlinelibrary.wiley.com/doi/10.1002/2016WR019638/abstract, l Exclusively for purposes of assessing sensitivity between water shortages and possible climate change impacts, the Monte Carlo analysis assumed that peak demands could be experienced anywhere from 0 to 6 weeks earlier over the year, each with equal probability of occurring. It was assumed that annual average flows were most likely to decrease 20% per year, but this range could vary from +10% per year to -40% per year. These ranges are 8 Headwaters Corporation Page 10 of 19 City of Aspen 7/03/17 From this sensitivity analysis, it appears that climate change could indeed have an impact on the Aspen water supply. However, it has been interesting to note that the timing of the flows, expressed in terms of how many weeks peak flows are accelerated, has so far had a significantly smaller impact on the number of possible shortages compared to the possible impacts to overall volume of flows. As one might expect, the greater the decrease in overall flow, the greater the number of possible shortages. To best incorporate current climate change knowledge, this effort is working with the City’s climate change staff and their associates to incorporate recent data and plausible ranges of data into the current modeling framework. The potential impacts of climate change will remain highly uncertain, but this effort is attempting to bracket the potential range of impacts and assess their implications on the number and severity of water shortages. Figure 7. Example of Existing and Hypothetical Modified Hydrograph Maroon Creek above Aspen, Average Weekly Flow 1970-1994 350.0 325.0 300.0 250.0 Average Weekly Hydrograph, WY 1970-1994 225.0 Modified Hydrograph Average Weekly Flow [cfs] 275.0 200.0 175.0 150.0 125.0 100.0 75.0 50.0 25.0 0.0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 Week of Water Year (starts October 1) thought to bracket possible impacts. It is important to note that, so far, these assumptions are only for purposes of testing sensitivity. Headwaters Corporation Page 11 of 19 City of Aspen 7/03/17 Combined Uncertainties of Water Supply It should be noted that, so far, five uncertain variables have been identified and incorporated into the water supply uncertainty analysis, two involving flow adjustments Castle Creek and Maroon Creek, respectively, and three involving how the hydrograph may shift in response to climate change. The statistical parameters, or characteristics, associated with these variables, such as their most likely values, standard deviation and skew, are still being investigated. However, even with preliminary estimates of Uncertainties associated with these parameters, meaningful sensitivity analyses are climate change are considered by currently being conducted to test the importance of each varying peak flow and annual variable in determining the frequency and severity of any runoff volume. water shortage. Water Demand In contrast to water supply and climate change, future water demand is an area in which the City has some degree of control. In order to finalize the demand projections, input from the City about how Aspen will plan for growth within its water system is needed. However, prior to acquiring this input, a sensitivity analysis is being conducted to test the response of increasing demand and changing demand usage characteristics to the frequency and severity of possible water shortages. Variables being considered in these preliminary sensitivity tests include residential occupancy rates, single family residential outdoor water usage, and total overall 2064 demand. Previous Water Demand Estimates and Water Production The water demand portions of three previous studies have been evaluated, as well as historical water production at the water treatment plant (WTP). Enartech The 1994 Enartech Report examined 6 alternative demand scenarios, primarily using a land use approach to estimate 2010 development levels and associated water demand for each. The scenarios were developed by City staff and represented increasing larger service areas for the City’s water utility. It appears a significant effort went into developing these scenarios and much of the information regarding the extents of various possible water service areas and their build-outs is still applicable. As illustrated in Figure 8, the 1994 estimates anticipated that total demand, expressed in acre-feet per year, would increase to a range of 4,500 acre-feet to 6,000 acre-feet per year, depending on which scenario is selected. The 4,500 acre-feet demand reflects build-out of Aspen’s incorporated and annexed areas. The 6,000 acre-feet demand reflects complete buildout of the Aspen metropolitan area. It is noted that actual 2010 usage was less than 3,000 acre-feet that was projected. A major reason for the divergence between estimated and actual usage was due to adoption of water conservation by Aspen’s customers as a result of low flow plumbing fixture codes and frequent renovations of existing dwelling units in the Aspen area. The remainder of the difference was likely due to slower than anticipated development. Headwaters Corporation Page 12 of 19 City of Aspen 7/03/17 Aspen Water Efficiency Plan The 2015 Aspen Water Efficiency Plan (WEP) broke-down current demands by usage characteristics, ranging from single family residential, multi-family residential, commercial, public facilities, and others.9 For the baseline analysis, demand in each category was extrapolated to 2035 using a 1.2% growth rate, the same anticipated population growth rate embodied in the 2016 Wilson baseline analysis. This implies the mix of land uses stays constant over time. The WEP also developed passive and active conservation policies, the latter of which includes a focus upon reducing outdoor irrigation usage. A result of these efforts is the Water Efficient Landscape Ordinance, recently passed by the Aspen City Council. The water usage estimates developed in the WEP are shown in Figure 8 for the baseline and active conservation scenarios. The active conservation scenario shows a significant flattening of demand over the next 20 years. Whether that flattening can be maintained past 2035 is not addressed. Wilson Report The 2016 Wilson Report’s baseline analysis extrapolates 2012 treated water demand through 2064 at the anticipated population growth rate of 1.2%. No change to demand patterns for either indoor or outdoor usage is assumed. An alternative demand estimate was developed that assumed that new demands served outside the City would be exclusively for indoor water usage, although this policy is not currently in effect and would be subject to review. These future estimates are shown in Figure 8. Water Production at the Water Treatment Plant Alongside of the future demand estimates made at various times, Figure 8 also shows historical water production at the WTP for the years 1994 through the present. Several issues become apparent: • Overall water production followed a downward trend through the period 1994 through 2010, corresponding to significant gains in conservation due to water conserving plumbing codes and behavioral changes. • Since about 2013, there has been a sharp upswing in production that has continued to the current period. The City is conducting a billing analysis to determine if billed water usage has followed this same trend, to determine whether it is a bona fide increase in demand or if it might be leaks in the distribution system. Installation of new metering infrastructure at the WTP indicates that water production is being accurately measured there, so the source of the increase is likely elsewhere. • Although the 1994 Enartech study and the 2016 Wilson Reports indicate that flows in Castle and Maroon Creeks are sufficient to meet current demands, even at this high level, the increased production is using existing infrastructure at near its full capacity. • If this increase in usage is demand-based, it calls into question the baseline demand estimates contained in the Wilson Report and the WEP. However, at least for the WEP, current demands were based on billed water usage rather than water production, so it is likely the baseline levels of demand assumed in these two efforts were close to actual demand. 9 Element Water Consulting and Water DM. 2015. Aspen Municipal Water Efficiency Plan. Headwaters Corporation Page 13 of 19 City of Aspen 7/03/17 Figure 8. Previous Water Demand Estimates and Historical Water Production 7,000 6,000 1994 Estimates, Scenarios 1 and 6 Wilson Report, Baseline and Restricted Growth Acre-feet/year 5,000 4,000 Water Efficiency Plan, Baseline and Active Conservation 3,000 Water Production at WTP 2,000 - 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030 2032 2034 2036 2038 2040 2042 2044 2046 2048 2050 2052 2054 2056 2058 2060 2062 2064 1,000 Year Role of Population Estimates Estimated growth rates for Aspen’s future water demands have most recently been population-based. That is, current demands, either in total or by sector, are extrapolated at an anticipated rate of population growth. This is a relatively common method in the water industry because often there is a strong link between resident population and water usage. Aspen differs from the norm because of relatively strong controls on growth, in total and for certain types of land use, high real estate prices, a wide range of lifestyles, and the seasonal and transient nature of a portion of the population. Different types of land use will likely grow at different rates in Aspen and in other communities of similar economic and demographic nature due to actions by local planning entities. As a result, future land uses and land use changes may be a better predictor for future water usage than population estimates. The 1994 Enartech effort recognized this and, with City For Aspen, land use type may be a Planning staff assistance, focused upon development better indicator of demand than patterns for specific types of land use. Although the population. resulting total demand estimates weren’t accurate for 2010 (largely due to conservation efforts as previously discussed), they do yield insight as to the magnitude of the possible increase in demand as the area builds out. Development of a Land Use Demand Estimate For purposes of providing an alternative approach to demand, this analysis is currently working towards developing a land use-based estimate. City staff has provided insights, GIS coverage maps, and other support in cooperation of this effort. Water usage categories considered in the 2015 WEP appear to have a logical connection with the City’s land use categories. As a result, WEP usage information, along Headwaters Corporation Page 14 of 19 City of Aspen 7/03/17 with geographic area information about land use being obtained from the City’s GIS group, is being used to characterize water usage for major land uses. A challenge to using a land use basis for estimating water demand over a long period of time is that water demand estimates require a long-term outlook, say 40 to 50 years, whereas land use plans tend to have a shorter outlook, in the range of 5 to 20 years. A strategy for dealing with these differing time frames is to incorporate long-term land use trends and possibly bracket impacts in the more distant years with a range of plausible values. Build-out of the Aspen Area and Associated Water Usage, 1994 View Build-out implies that all developable land has been developed and future growth will depend on either expanding the water service area and/or intensification of existing developed lands. With the City staff’s assistance, the 1994 Enartech study examined both of these types of growth in detail, estimating buildout in terms of Equivalent Capacity Units (ECU), a standardized measure of water consumption and capacity requirements per unit of development. • • • • The 1994 study estimated there were about 12,400 ECU’s being served at that time. Buildout of the City and existing water service areas through 2010 was estimated in 1994 to be about 17,300 ECU’s. It appears that this was considered full buildout of these areas. The Aspen Water Efficiency Plan estimates that there are currently 17,300 ECU’s in the service area. It is a coincidence that this equals the build-out estimate for 2010, made in 1994. Full buildout of the system including the current service areas, service to adjacent fully developed areas, and proposed extensions into new water service areas was estimated in 1994 to total 19,835 ECU’s. Appendix A contains a summary of the 1994 study’s assumptions about the number of future ECU’s to be potentially served. This summary is for Scenario 6, which documents the source of the 19,835 ECU estimate. Applicability of these 1994 estimates to the current analysis is somewhat problematic given the actual development patterns and significant reductions in indoor water usage, but the total estimated number of ECU’s comprising build-out provides a data point for long-term demand. If 19,835 ESU’s represents total buildout of the current and potential future water service area, and there are currently 17,300 ECU’s, a proportionate increase in water demand would be about 15%. Annual water demand at the WTP would increase from its current estimated level of about 3,400 acre-feet to about 3,900 acre-feet at full development. Based on population-based estimates of water demand, the Wilson Report estimated demand to be slightly over 6,000 acre-feet per year in 2064; the Aspen WEP estimated baseline demand to be in the range of 4,000 acre-feet in 2035. A portion of current efforts focus upon determining whether Aspen’s water system is, in fact, this close to actual build-out conditions. Verification involves ongoing discussions with Aspen’s Water and Planning staff to assess the accuracy of the 1994 estimates. In addition, Pitkin County Planning staff has been contacted. Further, verification of the water usage implied by an ECU would be useful to review and determine whether this volume has changed over time. Headwaters Corporation Page 15 of 19 City of Aspen 7/03/17 Trends Affecting Future Demands A range of land use and demographic trends have been identified during discussions with City Planning staff. Similar to the 1994 effort, emphasis should be given to examining geographic areas within the City and the surrounding area. Geographic areas include: • Infill areas roughly within the boundaries of the Roaring Fork River, Castle Creek, and Aspen Mountain. This area has been described as 95% built-out. • Areas within the City boundary • Areas within the Urban Growth Boundary, UGB • Areas of Pitkin County, west of Castle Creek, along the Highway 82 corridor • The current water service area and possible future water service areas Additional input is being sought on future usage trends. However, usage trends identified to date include: • The likelihood of seasonal residents extending their stays. This possible trend was acknowledged but was partially countered by another, offsetting trend, discussed below. • Occupancy rates for Free Market residential properties currently occupied year around will decline over the next 10 to 20 years as these residents retire and/or choose to sell their residences to those who would be only part-time residents. This latter trend effectively lowers average household and indoor water usage. Outdoor water usage and trends in future outdoor usage would likely remain unchanged under either of the above trends. • Overall occupancy rates go down in nearly all housing categories with the possible exception of higher density multi-family units. • In general, areas with Free Market housing have lower occupancy and higher water usage than their non-Free Market counterparts. • Future levels and locations of Affordable Housing are uncertain. • Areas with potential for future growth are along the Highway 82 corridor west of the roundabout, described as a sort of wild card with respect to the type and pace of development here. • The area of Pitkin County west of Castle Creek is similarly uncertain. There is currently no Affordable Housing in this area. The long-term plans for this area and the area west of Maroon Creek are currently being developed. Headwaters Corporation Page 16 of 19 City of Aspen 7/03/17 Table 3 summarizes two scenarios for land use-based demand estimates being developed and the information currently being acquired to examine each. Table 3. Information Needed to Develop a Land Use-Based Demand Estimate for 2 Scenarios Scenario 1: Buildout of City and existing water service areas Scenario 2: Buildout of City and existing water service areas, plus adjacent developed areas and proposed new water service areas Estimates made in 1994 Enartech study 17,300 ECU’s 19,835 ECU’s Updated information being developed to revise the 1994 estimate Remaining buildout by land use category; water usage by land use category or customer class; incorporating usage trends and conservation measures identified in the WEP, including the Water Efficient Landscape Ordinance. Same as above, plus identification of potential developed areas and newly developing areas. City of Aspen Water and Planning, and Pitkin County Planning are sources of information. System Operations Operational conditions within the Aspen water system also affect reliability of the water system. Overall, the Aspen water system has been very reliable. Despite being nearly solely dependent upon the seasonal flows of Castle and Maroon Creeks, supplemental groundwater, and having less than a day’s worth of storage at the Thomas Reservoir, no major shortages or outages were identified by the staff.10 However, recent increases in water production, with associated peaks in demand, combined with infrastructure limitations, are increasing the strains upon treatment infrastructure and staff. With respect to treatment, the City’s two parallel treatment plants provide 11.2 million gallons per day (mgd) and 8.2 mgd capacity, totaling about 19.5 mgd, with possibly slightly less actual capacity. Current peak day demands are near 8 mgd, but instantaneous peaks during peak days are in the range of 10 to 12 mgd. During Currently, there is very little ability peak periods, both plants need to be available to maintain to absorb water system disruptions. reliable service. In addition, known risks in operations during peak run-off periods, such as filters clogging due to periodic turbidity spikes, can rapidly reduce effective capacity and increase risk. Beyond this example, WTP staff is concerned that hydraulic capacity may be limited during peaks, leaving no room for error to ensure peak reliability. Concerns about disruptions to the Castle Creek and Maroon Creek diversions that might occur due to avalanches, forest fires, and flooding are realistic and appropriate concerns. Staff also mentioned 10 A possible exception is an anecdotal memory of 1994, when avalanches took both of the City’s diversion dams, Castle Creek and Maroon Creek, off-line for approximately a day. The City issued an emergency request to the public to eliminate as much water use as possible during this period. Between this request and storage in the Thomas Reservoir and within the distribution system, the City apparently got through with minimal disruption. Headwaters Corporation Page 17 of 19 City of Aspen 7/03/17 concerns about contamination of the WTP, whether that be through natural or nefarious measures. Disruptions that might occur within one basin would likely significantly reduce the ability to meet peak demands and disruptions occurring in both basins simultaneously would result in potable water shortages within about 12 hours. There are concerns that flow can be limiting in the winter due to icing and other infrastructure restrictions, including diversion intake capacity. Staff raised similar worries about peak demand infrastructure limitations during dry summer months. Benchmarking A benchmarking analysis consisting of surveying industry leaders and regional water providers of comparable size and location about their water sources, reliability, system storage, and possible sources of emergency supplies has been conducted. The data was obtained from the providers’ websites, their Source Water Protection Plans, and personal communication with the provider itself. Currently in draft form, the results of this survey are being compiled. Next Steps Ongoing efforts will focus upon completing a land use-based estimate of long-term water demand, refining assumptions about climate change, refining the assumptions about uncertainties for demand, and completing sensitivity tests of critical variables. Issues that would benefit from further discussion with the City staff and City Council are discussed below and summarized in Table 4. • • • Measure of satisfaction with the existing hydrology data. As shown, the data is limited and depends on an adjustment factor that is highly uncertain. Although this issue is addressed in the Monte Carlo analysis, a more accurate assessment of Aspen’s water resources requires measuring the supply. At minimum, this would require installing gauges on the two creeks at the City’s diversion points and at other strategic locations. Recommendations for more gauges and better measurement were made in the 1994 Enartech Study and the Wilson Report. Data from these gauges would not be available in the near term, but the act would demonstrate a commitment to managing the water resource. Whether to be satisfied with the hydrologic period of record. It is relatively short, 25 years, and probably does not contain the worst years on record. Alternatives to improve this data involve extending the period through the 1950’s on the front end and through 1995-2016 on the back. Statistical methods would be needed to develop correlations between the Castle and Maroon Creeks with nearby basins having longer records. Alternatively, statistical hydrology can be used to synthesize new data containing the same statistical characteristics as the period of analysis. Tree ring analysis, consisting of correlating tree ring with stream flow, is another method to identify extreme periods of short and long duration. With completion of a utility to model the possible impacts of climate change and test the sensitivity of the number of water shortages to the severity of climate change, this effort has now turned to developing a statistical distribution representing its most severe probable impacts. This is a challenging effort because data is still lacking for the Roaring Fork Valley and Headwaters Corporation Page 18 of 19 City of Aspen 7/03/17 • • • • Castle and Maroon Creeks. Is there a desire on the City’s part to actively work to help downscale more regional data to the local basin level? Much of the future development in the area will be in Pitkin County, rather than Aspen. With respect to future water demand, what is the City’s view with respect to expanding water service to adjacent areas, beyond the build-out of existing and currently served areas? What is the likelihood this may happen? What is the likelihood of increased residential and commercial density within the City limits, in other words, growing-vertical rather than growing-out? Are there additional demand trends not identified above? Also demand related, what is the perception about possible water shortages in general: o No shortages, ever? o Periodic shortages of hopefully short duration? o More frequent shortages as long as they don’t impact water for indoor use? Table 4. Summary of Uncertainties Item Flow measurements Hydrologic period of record 1970 through 1994 Climate Change Future land use changes Summary of uncertainty Gauge data is limited and the flow adjustment factors vary widely depending on the method used to estimate them. Possibly not the most severe period, considering 1950’s and years since 2000. More contemporary research suggests peak run-offs may be earlier and flow impacts may be more severe than previously thought. Most recent demand analyses have implicitly assumed that land uses grow at the same rate. Future expansion of the water service area The level of future growth attributable to new extra-territorial service. Definition of shortage Currently, a shortage occurs when City demands cannot be met while still meeting instream flow commitments. Existing data indicates that the critically dry period is less than 2 years in duration; existing data is limited. Critical Period Headwaters Corporation Input desired Discussion of the stream data set and inherent uncertainty with the adjustment factors; whether to improve the data moving forward. Potential use of statistical methods to extend the record. Perspectives regarding the range of impacts to consider; potential to work with others to downscale these more contemporary results to Castle and Maroon Creeks Thoughts and rationale regarding build-out and future rates of growth of different land use types, including single family residential, commercial, and affordable housing. Level of consideration and/or policies to provide water to existing and new development outside of the current water service area Input to the balance between meeting diverse municipal demands and instream flows in times of shortage. Thoughts about potentially synthesizing a multi-year critical period based on potential climate change impacts. Page 19 of 19