Pecos River–New Mexico Basin Case Study
Case Study Contents
Photo by artiste9999 / Getty Images
The Pecos River Basin in New Mexico is an arid basin that begins in the Sangre de Cristo Mountains of northern New Mexico, flows into Texas at New Mexico’s southern border, and eventually joins the Rio Grande along the U.S. border with Mexico (see Figure 1). The Pecos River Basin has significant water supply and management challenges. Overall water supplies are marginal, hydrologic variability in the basin is high, and shortages of water to meet the current uses occur frequently. Increasing temperatures, drought severity, and drought duration seen in recent decades highlight the need to plan for increasing water supply challenges.
The Pecos River–New Mexico Basin Study (PRNMBS) is the most recent Reclamation Basin Study included in this online tool. This case study describes how the PRNMBS study team used Decisionmaking Under Deep Uncertainty (DMDU) methods to evaluate future water management conditions and adaptations, and it demonstrates ways in which Robust Decision Making (RDM) techniques could augment the original analysis. It concludes with a discussion of possible next steps. The case study team used available data analysis undertaken for the basin study, mapped this work to the appropriate steps of RDM, and, in some cases, added some RDM analysis.
For this basin study, the study area was the portion of the Pecos Basin within the state of New Mexico. Three irrigation districts (IDs) are the primary water users in this study area: Fort Sumner Irrigation District (FSID) and Carlsbad Irrigation District (CID) are dependent on surface water, while the Pecos Valley Artesian Conservancy District (PVACD) primarily uses groundwater from the Roswell Artesian Basin. In addition, the U.S. Fish and Wildlife Service’s (USFWS’s) "Final Biological Opinion [BO] for the Carlsbad Project Water Operations and Water Supply Conservation, 2016–2026" (USFWS, 2017 [PDF]), prescribes water flow targets for the Pecos bluntnose shiner (a member of the minnow family). Finally, New Mexico must comply with water delivery requirements to Texas under the Pecos River Compact. The New Mexico Interstate Stream Commission (NMISC), the nonfederal partner for this study, manages Compact deliveries. All of these demands place increasing stress on the limited Pecos River water supply.
The PRNMBS was a partnership between Reclamation’s Albuquerque Area Office (AAO) and the NMISC, in cooperation with IDs operating in the Pecos Basin in New Mexico. The final PRNMBS report is currently under review, and a study summary [PDF] can be found online. This study was undertaken to project future changes in hydrologic and water management conditions in the basin and then to work with basin stakeholders, including FSID, CID, and PVACD, to develop and model approaches to maintain a vibrant agricultural community in the Pecos Basin in New Mexico, including changes to water deliveries, reservoir operations, or infrastructure. Although there are other water uses in the Pecos River Basin in New Mexico, this study emphasizes projected impacts to irrigation districts and actions that irrigation districts might take to be better prepared for anticipated future conditions.
In particular, the PRNMBS team sought to answer the following questions:
- What are the current and projected hydrological and meteorological changes in the Pecos River Basin in New Mexico?
- How might these changes affect irrigation water supplies and demands?
- What infrastructure or operational changes might the irrigation districts undertake to maintain the viability of irrigated agriculture in this basin?
- How might water management strategies mitigate the projected changes in water supplies?
This case study explores how RDM methods could help address these questions.
As in other basin studies, and consistent with DMDU best practices, the PRNMBS team used a scenario-planning approach rather than a probabilistic assessment to consider uncertainty. In contrast to the Colorado River Basin Water Supply and Demand Study (CRBS), which evaluated thousands of plausible climate futures, the PRNMBS study team preselected a small number of plausible climate futures and evaluated current and additional management approaches against those futures.
This case study highlights how the basin study framed the analytical problem (the decision framing step of RDM), how the study generated a database of results that can be explored interactively (the evaluate strategies across futures step of RDM), and how one can use RDM techniques to present trade-offs among the many metrics of concern (the trade-off analysis step of RDM). This case study did not identify robust strategies (see Figure 2).
RDM Steps in This Case Study
- Decision framing: The uncertainties, portfolios of management strategies, models, and performance metrics developed for the PRNMBS are shown in Table 1. These inputs were developed by the Reclamation project team.
- Evaluate strategies across futures: Using the five climate projections developed in the decision framing step to represent plausible future conditions, the PRNMBS team used a detailed simulation model of the system to evaluate how well the current “No Action” management portfolio would perform through 2099 in each of these futures according to 13 different metrics. The PRNMBS team then evaluated an additional seven water management portfolios (i.e., strategies).
- Vulnerability analysis: Because a relatively small number of futures were developed for the PRNMBS, a vulnerability analysis was not done. This case study, however, describes how including more climate conditions and other uncertainties would be a clear next step in a follow-up study, particularly so that data-mining tools could be used to characterize vulnerabilities across the many futures and across the many metrics.
- Trade-off analysis: The PRNMBS team evaluated the performance of the different portfolios across the scenarios, showing that no portfolio performs well on all of the performance metrics. This case study expands on this analysis and describes the key trade-offs between metrics that must be made.
- New futures and strategies: The PRNMBS team developed seven additional adaptation portfolios during the decision framing step and evaluated those portfolios in the evaluate strategies across futures step, prior to any trade-off analysis. Using the expanded vulnerability analysis presented in this case study, another iteration of RDM could suggest modifications to the portfolios that would address vulnerabilities and result in more-robust strategies.
The PRNMBS team’s use of relatively few futures limits the ability to conduct a vulnerability analysis, which typically involves a more comprehensive exploration of the uncertainties. This, in turn, limits the ability to characterize the robustness of strategies. A next step of the analysis could expand the variety and number of climate and other future conditions explored in the study, which would generate the data needed to conduct vulnerability analyses and identify robust portfolios.
RDM Step: Decision Framing
After reviewing the study documents, we can summarize the key elements relevant to the basin study analysis using an XLRM matrix (see Table 1). See the RDM section for more information about the XLRM matrix. The subsections below elaborate on each of the elements in the matrix.
Table 1. XLRM Matrix for Pecos River–New Mexico Basin Case Study
|(X) Uncertainties||(L) Management Options and Strategies|
|(R) Relationships or Systems Model||(M) Performance Metrics|
NOTE: The basin study used three additional metrics to describe upper basin, lower basin, and groundwater health. These were excluded in this case study for simplicity.
The PRNMBS team considered uncertainties related to large-scale future climate conditions, assumptions about how large-scale climate conditions translate to local climate conditions (called bias corrections), and the hydrological response of the basin. The study developed a large number of futures from a combination of these uncertainties and then selected five of them—which they called storyline scenarios—to represent a range of plausible future conditions.
The study team began by considering 93 different climate conditions, each from one of 93 global climate models (GCMs) in the CMIP5 archive. They coupled these conditions with two hydrologic models and five bias-correction approaches (one of which represented the raw data without correction), for a total of 930 plausible candidate futures (see Figure 3).
Figure 3. Determining the Number of Plausible Futures
93Climate conditions multiplied by 2 Hydrologic models multiplied by 4+1 Biascorrection equals 930 Candidate futures
Storyline scenarios were developed through a process of iteratively selecting fewer representative futures from this set of plausible candidate futures. First, the PRNMBS team selected one of two models and one of the five bias-correction techniques, each of which produced more-complete and more-accurate representations of the system.
The number of projections under consideration in this study was reduced by selecting only the projections based on the greenhouse-gas emissions scenarios referred to as representative concentration pathways (RCPs) 8.5 and 4.5. RCP 8.5 represents a high emissions future. In contrast, RCP 4.5 describes what could happen if our global society strongly reduces greenhouse-gas emissions over the coming few decades. This process narrowed the number of hydrologic projections under consideration to 58 (28 RCP 8.5 projections and 30 RCP 4.5 projections).
A detailed comparative analysis was performed on the remaining projection traces to characterize the story that each tells about the projected future climate and hydrology in the basin. The overarching goal was to select storylines that were not only plausible representations of the future but also distinct from one another so that a wide range of plausible futures could be modeled within a small, select group.
The study team characterized and compared snowmelt runoff and timing, monsoon intensity, seasonal and spatial runoff and precipitation patterns, temperatures, evapotranspiration rates, and total streamflow in the mainstem and key tributaries. Reclamation and the NMISC evaluated these projections and selected the following five storylines to be carried forward in the study:
- High emissions—dry climate response: substantially hotter and drier than historical conditions, with less spring runoff and fewer monsoon storms. Around the year 2040, this storyline shows much drier conditions in terms of flows at the Acme Gage,1 reservoir storages, settlement releases, CID allotments, frequency of critical drying criteria as determined in the 2016 BO, and less availability of FSID bypass flows.
- High emissions—high monsoon, low snowpack (HMLS) climate response: substantially hotter, with increasing wet conditions when compared with the 1950–2009 period below Acme and because of increased stream flows resulting from slightly more-frequent monsoon storms that are much greater in intensity. However, spring runoff above Santa Rosa decreases as a result of reduced snowpack in the headwater region.
- High emissions—moderate climate response: hotter than historical conditions, with modest drying; more precipitation as rain during the winter, but drier in the summer. This storyline starts to show mild drying across the whole system in the 2050s.
- Reduced emissions—increased monsoon climate response: minor increases in temperature and slightly increased demands. Precipitation increases from much more-frequent monsoon storms that are slightly greater in intensity. This storyline would be modestly hotter with less snow and more-frequent and more-intense monsoon storms.
- Reduced emissions—median climate response: slight increase in temperature and demand, with slight decreases in water availability, so slightly drier than historical conditions. This storyline would be modestly hotter with slightly more runoff during spring and slightly less precipitation in all other seasons.
The PRNMBS method for developing a small set of scenarios differs in some important ways from a more typical DMDU analysis. First, instead of evaluating the performance of the system and adaptation decisions across a large set of plausible futures, researchers evaluated the system and adaptation decisions only with respect to the five storyline scenarios. Visualization 1 plots trends in temperature and precipitation for the 58 climate projections considered for inclusion in the PRNMBS from 2010 to 2099. The five projections used as the basis for the five storylines are labeled. This view shows that the five selected storyline scenarios lie roughly on the perimeter of the region of temperature and precipitation trends. One advantage of this approach is the significantly lower computational and analysis requirements.
Visualization 1. Temperature and Precipitation Trends for 58 Climate Projections
One can also see how these storyline scenarios compare across other related variables. For example, one can replace the temperature variable with trends in total streamflow in the main stem (inflow), which is a good proxy for available supply for the system. In this view, one can see that the five selected scenarios roughly equally sample the range of flows and precipitation trends (see Visualization 2). There are other characterizations of climate that could be examined in a similar way, including snowmelt runoff, monsoon intensity, and overall moisture. These visualizations focus on annual conditions, but evaluating summaries of different seasonal aggregations, as the PRNMBS did, provides additional insights.
These visualizations show that the selected scenarios all sample different combinations of temperature, precipitation, and inflow trend uncertainty. Because of the small number of scenarios defined, there are some areas of the uncertainty space that are not well explored. For example, there are no scenarios representing conditions in which precipitation trends are small (negative or positive) but there is warming of more than 0.3 degrees Fahrenheit per decade.
Visualization 2. Inflow and Precipitation Trends for 58 Climate Projections
Management Options and Strategies (L)
As with the CRBS (Case Study 1) and the Sacramento and San Joaquin Rivers Basin Study (SSJRBS) (Case Study 2), the PRNMBS began by analyzing a baseline management strategy.
The team further conducted a sensitivity analysis of the baseline management strategy by eliminating each of the ID and Endangered Species Act (ESA) diversions and then running the model to evaluate the impact of each district’s and the ESA’s requirements on the river, groundwater, reservoirs, and available supplies to other IDs. These sensitivity analyses helped the team understand the scale of water use change that might be needed in each system component to mitigate the projected changes in water supply. These analyses were also used to help the team design the remaining water management strategies.
The PRNMBS team defined and evaluated several modifications to the baseline management strategy. The following four alternative strategies focused on reducing consumption:
- 20-percent reduction by all districts: The central district (PVACD) reduces irrigation demands by 20 percent by 2045, and the upper district (FSID) and lower district (CID) reduce irrigation demands by 20 percent by 2055.
- 25-percent reduction by all districts: The central district (PVACD) reduces irrigation demands by 25 percent by 2045, and the upper district (FSID) and lower district (CID) reduce irrigation demands by 25 percent by 2055.
- 25-percent reduction by central district: Only the central district (PVACD) reduces irrigation demands by 25 percent by 2045; other IDs remain unchanged.
- 30-percent reduction by central district: Only the central district (PVACD) reduces irrigation demands by 30 percent by 2045; other IDs remain unchanged.
The team also evaluated a strategy to increase on-farm efficiency from 50 percent in 2010 to 75 percent in 2050.2
Relationships or Systems Model (R)
The study team used the VIC model and PRMS to calculate runoff at specific gage locations. The PROM, a RiverWare model, was used to assess (1) interactions between the various operations in the basin, including IDs, and (2) ESA operations under different strategies. The PRNMBS team also used the RAB Groundwater Model, a MODFLOW model, to estimate the likely effects of such actions on future water supply for irrigation.
Performance Metrics (M)
The PRNMBS team used several metrics to evaluate the performance of the water management system. This case study focuses on results for five of the metrics selected in collaboration with the basin study team. We have renamed these metrics so that they are more intuitive for case study users who are unfamiliar with the region. These metrics are upper basin health, lower basin health, species health, legal compliance, and reservoir health. Together, these metrics provide an overall picture of the health and performance of the basin. Upper basin health shows how irrigators higher up in the Pecos Basin are doing, while lower basin health shows how irrigators in the lower part of the Pecos in New Mexico are doing. Because the management strategies all occur downstream of upper basin users, all strategies perform the same as the baseline strategy for upper basin health for a given period and storyline. Species health provides a picture of ESA compliance and fish health, and legal compliance monitors the legal side of the water flowing through the Pecos and New Mexico’s ability to stay in compliance with water delivery to Texas. Reservoir health offers an indication of the basin’s resilience to variability and provides an overall picture of basin health.
These metrics are described further in Table 2, with the original names of the metrics in the PRNMBS in parentheses. For these metrics, the table also lists the historical value. The second half of the table describes other metrics that were used in the basin study but that are not included in the case study. Historical values are not provided for these metrics.
Table 2. Metrics Used in the PRNMBS and in This Case Study
|Metric Name||Description||Historical Value/Units|
|Metrics used in this case study|
|Upper basin health
|FSID is the most upstream district, and, therefore, metrics related to FSID provide a picture of the health of the upper basin. The Hope Decree allows FSID to divert up to 100 cfs (2.83 cubic meters per second [cms]) based on the natural flow of the river during irrigation season and measured over two-week intervals.||45,758 acre-feet|
|Lower basin health
|CID is the most downstream district, and, therefore, metrics related to CID provide a picture of the health of the lower basin. The CID allotment is the amount of water legally available to CID, with a maximum of 3.697 acre-feet of water per acre of land. The annual allotment is based on how much water is in the reservoirs within the Pecos River Basin and how much CID irrigators have diverted so far.||2.47 acre-feet per acre of land|
(Acme dry days)
|River flow is a significant determinant of species health. In particular, the number of dry days is defined as the number of days per year in which the average daily discharge at the Acme Gage is less than 5 cfs. When this happens, the river likely becomes discontinuous somewhere downstream of the Acme Gage.||32 days|
|The 2003 Pecos Settlement Agreement seeks to ensure that New Mexico complies with water releases to Texas consistent with the 1948 Pecos River Compact. To this end, the model calculates a yearly amount of water available for the 2003 Pecos Settlement based on reservoir storage and CID releases. Although this water is available, it might not be required, because many factors are involved in determining the actual annual releases for Compact compliance. Thus, it is considered a safety, and the lower the value, the less safety in case of a call on the Compact.||14,105 acre-feet|
|Historically, there is a wide range of annual reservoir storage, and the average storage provides an indication of overall basin health and reservoir health. To examine how potential changes in hydrology would affect the system over time, the study assessed the change in daily average for total reservoir storage (i.e., the difference in the sum of storage in all four system reservoirs).||102,900 acre-feet|
|Metrics in the PRNMBS that were not used in this case study|
|FSID shortage||FSID is the smallest and most upstream of the three IDs, with approximately 6,000 acres of irrigation. It is the only ID that has run-of-the-river rights, with an allotment of up to 100 cfs. Shortages are calculated as the result of the amount that is diverted that does not meet the required modeled crop demand.||Acre-feet|
|CID storage||CID is the second-largest ID, with approximately 20,000 acres of irrigation. Shortages are calculated as the result of the amount that is diverted that does not meet the required modeled crop demand.||Acre-feet|
|Groundwater levels||Groundwater is important for the basin, and different climate conditions and strategies might change groundwater levels.||Feet|
RDM Step: Evaluate Strategies Across Futures
This step of RDM evaluates strategies across futures. In the case of the PRNMBS, the study team evaluated the performance of each of the six strategies in each of the five storylines. The study used static tables, such as the one shown in Table 3, to show how the system would perform with respect to different metrics across the different scenarios.
Because of the many ways in which outcomes could be viewed across the many metrics, strategies, and futures, RDM studies typically use interactive visualizations to help explore the results of large ensembles of data. Visualization 3, which we developed for this case study, shows the considerable amount of data that could be explored.
Visualization 3. Metric Outcomes for All Strategies Across Five Storyline Scenarios
Interactive graphics can help decisionmakers interpret and compare outcomes more easily than they can using a table. Visualization 4, which we developed for this case study, shows a different view of the same data. The plot shows the performance of a selected strategy in a selected period for each of the five key performance metrics in each storyline. Higher points are better for all metrics. (To allow for this, the axis is reversed for species health because fewer dry days are better than more dry days.) For context, the shaded areas show values below the historical levels for these metrics, which serve as intuitive benchmarks for understanding how the future might be better or worse than the past.
Consider the performance of the baseline strategy for the 2040–2069 period. One can see that the outcomes differ significantly by storyline, indicating that future climate conditions will have a significant impact on outcomes in the basin. For all metrics, the baseline strategy performs significantly worse under a high emissions–dry storyline relative to other storylines and results in significantly worse outcomes than in the past. However, under the high emissions–HMLS and reduced emissions–increased monsoon storylines, it performs better than in the past, and sometimes significantly so. In addition, a high emissions–HMLS future might result in better outcomes for legal compliance but worse outcomes for species health; the inverse seems true under reduced emissions–median and reduced emissions–increased monsoon storylines, suggesting that the nuances of precipitation and temperature trends will affect these metrics differently. Use the time period selector to change the period and the strategy selector to change the strategy.
Visualization 4. Relative Performance Across Five Metrics of the Selected Strategy
RDM Step: Identifying Vulnerabilities
Researchers conducting an RDM study typically conduct a vulnerability analysis to identify the uncertain conditions in which a strategy performs poorly. An alternative strategy could be preferred under such conditions. The PRNMBS team described key findings based on the information shown above but did not perform a formal RDM vulnerability analysis. For this case study, we illustrate how a vulnerability analysis could be performed. First, a vulnerability analysis asks, “In which futures is a strategy vulnerable (i.e., in which futures does a strategy fail to meet decisionmakers’ goals)?” Second, the vulnerability analysis asks, “What uncertain conditions describe those futures in which a strategy is vulnerable?”
To answer the first question, decisionmakers must define what it means for a strategy to meet stakeholders’ goals. Typically, they define performance thresholds for each metric. If a strategy performs better than a metric’s threshold, then the strategy meets decisionmakers’ goals for that metric. Conversely, if it performs worse than the threshold, then the strategy fails to meet decisionmakers’ goals for that metric. Decisionmakers must then decide how to aggregate successes and failures across all of the metrics to say that a strategy meets their goals in a particular future. For example, decisionmakers might say that for a strategy to be considered successful in a particular future, it must meet all metric thresholds. Thus, a strategy would be vulnerable in a future if it failed to satisfy one or more metric thresholds.
This case study uses the historical baseline as a threshold for each of five metrics to illustrate this methodology (the thresholds were not used in the PRNMBS itself). Using a historical baseline as the threshold essentially means that a strategy succeeds on a particular metric as long as it is as good as what was observed in the past.
Visualization 5 shows whether each strategy (columns) is vulnerable for each of the five performance metrics (rows) in different storylines (rows) in different periods (which are selectable on the right). A red X indicates that the strategy is vulnerable (i.e., it does not meet the performance threshold for that metric), while a green circle indicates that it is not vulnerable (i.e., it meets the performance threshold).
We can observe a few things. First, in the 2010–2039 period, all of the strategies essentially have the same vulnerabilities (e.g., they are all invulnerable with respect to lower basin health in the high emissions–moderate storyline, and they are all vulnerable in the high emissions–dry storyline). Indeed, every column in the 2010–2039 period is the same, with the exception of the increase efficiency strategy, which can meet lower basin health in the increased monsoon storyline, while other strategies cannot. On the other hand, this strategy cannot meet reservoir health in the high emissions–dry storyline, while all other strategies can. In general, this means that in any given climate, the proposed strategies can do little to affect outcomes in the near term.
Second, in the two later periods, all strategies are vulnerable for all of the metrics under the high emissions–dry storyline. This might be expected because the dry storyline most stresses water resources, and all of the considered strategies could be vulnerable.
Third, the 20-percent reduction by all districts and the 25-percent reduction by all districts strategies are not vulnerable in storylines where other strategies are vulnerable. For instance, these strategies are the only ones that meet lower basin health and reservoir health goals in the median storyline in the middle period (2040–2069). They are also the only ones that can meet the species health goals in the high emissions–HMLS storyline in the last period (2070–2099).
Visualization 5. Vulnerability of Strategies by Performance Metric and Storyline Scenario
Because every strategy has the same upper basin health performance as the baseline strategy, these performance data do not help stakeholders compare the strategies, and we omit that metric from the remainder of the analysis.
To understand a strategy’s vulnerability, we next aggregate and summarize these data. Here, strategies generally work well or poorly for the same set of metrics. So, we aggregate results across metrics, asking, “In a given storyline and period, for how many metrics is a strategy successful?” For example, the baseline strategy meets two metric thresholds in the reduced emissions–median storyline in 2040–2069. Visualization 6 is a collapsed version of Visualization 5 and shows the number of successful metrics for each strategy in each storyline.3
We can then address the following question, which was raised earlier in this section: “In which futures is a strategy vulnerable?” Visualization 6 enables you to set a threshold for the minimum number of metrics for which a strategy must be successful in a particular storyline and period to say that the strategy meets decisionmakers’ goals and is not vulnerable (green circles). Conversely, any strategy that meets fewer than this minimum number of metrics is vulnerable (red Xs).
Visualization 6. Overall Vulnerability of Strategies Based on Counting Vulnerable Metrics
For the remainder of this case study, we say that a strategy is vulnerable if it fails to meet the threshold for any of the four metrics (i.e., the number in the interactive table is less than four).
We can now answer the second question, “What uncertain conditions describe those futures in which a strategy is vulnerable?” Visualization 7 allows you to explore this question by showing the performance of each portfolio against the temperature and precipitation trends of each of the five storylines. This is shown for a particular period. The color of the mark indicates whether the strategy is vulnerable (i.e., meets fewer than four metric thresholds) or not vulnerable (i.e., meets all four metric thresholds).
Visualization 7. Performance of Specified Portfolio Across the Temperature and Precipitation Trends for Each Climate Scenario
For example, the 20-percent reduction by all districts strategy is vulnerable in two storylines in the 2040–2069 period. We might describe the vulnerability of this strategy by identifying the narrowest set of conditions that captures the two vulnerable futures. With this approach, the strategy can be said to be vulnerable if the temperature trend is above 0.8 degrees Fahrenheit per decade and, simultaneously, the precipitation trend is less than –0.3 inches per decade. This is the narrowest set of conditions that contains the two vulnerable scenarios.
However, these plots are sparse. An alternative interpretation could identify the broadest set of conditions that captures the two vulnerable scenarios. With this broad definition approach, the strategy can be said to be vulnerable if the temperature trend is above 0.4 degrees Fahrenheit per decade and, simultaneously, the precipitation trend is less than 0.2 inches per decade (see Table 4).
Table 4. Example Vulnerable Scenarios for the 20-Percent Reduction by All Districts Strategy
|Strategy: 20-Percent Reduction by All Districts|
|Vulnerable scenario with narrow definition||
|Vulnerable scenario with broad definition||
These two definitions highlight the idea that, with so much of the uncertainty space left unexplored, it is difficult to identify the true conditions under which a portfolio succeeds or fails. A next step in this analysis would be to explore more climate uncertainty (i.e., to run the model over more of the gray climate futures that were not evaluated in a first iteration) to enable a more precise vulnerability analysis. We describe what such an analysis could look like in the section on new futures and strategies.
RDM Step: Trade-Off Analysis
The PRNMBS team described the impact of different strategies on each of the metrics of concern. For this case study, we explore how RDM techniques can articulate trade-offs.
Although a complete RDM-style vulnerability analysis is not yet possible, we can still explore key trade-offs by comparing the futures in which each strategy meets the goals.
The first vulnerability table in this step of RDM (Figure 4) shows that the 20-percent reduction to districts and 25-percent reduction to districts strategies meet decisionmakers’ goals at least as well as all of the other strategies, for any metric, in any storyline, and for any period. As we noted earlier, this suggests that, at least with these data, there is not a trade-off between metrics.
However, decisionmakers are likely to have other concerns, such as cost, technical feasibility, social equity, and social feasibility. These factors were not evaluated directly in the PRNMBS, but we can imagine how they might be included as metrics from the beginning of the analysis and reveal trade-offs for decisionmakers.
For example, we have notionally added grades for cost and social feasibility to each of the strategies in Visualization 8. We might expect the baseline strategies to be the least costly and most feasible compared with the others, followed by the increase efficiency strategy, given that efficiency gains can be more palatable to stakeholders than required reductions. Meanwhile, any reduction to districts strategies could be both costly and difficult.
These notional data would suggest that there is a trade-off between basin health, cost, and feasibility. For example, as Figure 4 shows, the increase efficiency strategy performs almost as well as the 20-percent and 25-percent reduction to districts strategies. An argument can be made that the slightly higher vulnerability of this strategy is worth trading off for better cost and feasibility. Another interpretation might be that the 20-percent and 25-percent reduction to districts strategies perform the same, and, therefore, the added cost and lower feasibility of the 25-percent reduction to all districts strategy is unjustifiable, so that strategy should be eliminated from consideration. With such data, RDM can help stakeholders reason about how their various goals interact with one another and where they might need to consider trade-offs when not all of the goals can be met simultaneously.
RDM Step: New Futures and Strategies
The PRNMBS compared one set of strategies with a small set of climate storylines. Another iteration of RDM could consider more climate futures and other uncertainties, along with other factors (e.g., socioeconomic and land use changes).
Additional strategies could be developed that seek to combine aspects of the top strategies identified earlier. For example, could a combination of increasing efficiency with more-modest reductions in districts better meet performance and cost and feasibility goals? Such comparisons as these can produce new strategies and opportunities for achieving policymakers’ goals.
Here, we specifically show how the addition of new futures offers insights into vulnerabilities. We imagine that, in a next iteration, the study team evaluates the performance of strategies in all of the 58 climate futures. Figure 5 uses artificial data to illustrate how the 25-percent reduction by all districts strategy might perform. It offers more nuance about the precipitation and temperature conditions in which the strategy succeeds or fails than the five storyline points alone.
Inspection of the data might reveal the following vulnerable scenario A outlined in Table 5.
Table 5. Example Vulnerable Scenario A for the 25-Percent Reduction by All Districts Strategy
|Strategy: 25-Percent Reduction by All Districts|
|Vulnerable scenario A||
Furthermore, such dense data can help us compare strategies. Figure 6 offers two examples of what might be revealed about the 20-percent reduction by all districts strategy. The artificial data on the left suggest that this strategy is less vulnerable to precipitation decrease and temperature increase, and we might characterize vulnerable scenario B1 as shown in Table 6.
Table 6. Example Vulnerable Scenario B1 for the 20-Percent Reduction by All Districts Strategy
|Strategy: 20-Percent Reduction by All Districts|
|Vulnerable scenario B1||
Vulnerable Scenarios A and B1 are similar, and the early indications in the trade-off analysis step of RDM that the 20-percent and 25-percent reduction by all districts strategies perform similarly bear out. A decision to choose the former over the latter for cost and feasibility reasons could be confirmed.
However, it is possible that the simulations would reveal a different set of outcomes. The artificial data on the right of Figure 6 would suggest that this strategy is more vulnerable to precipitation decrease and temperature increase. A vulnerable scenario B2 based on these results might look different (see Table 7).
Table 7. Example Vulnerable Scenario B2 for the 20-Percent Reduction by All Districts Strategy
|Strategy: 20-Percent Reduction by All Districts|
|Vulnerable scenario B2||
Such a result would contradict the early hypothesis in the trade-off analysis step of RDM that the 20-percent and 25-percent reduction by all districts strategies perform similarly. In this scenario, stakeholders might not be willing to trade off basin health for cost and feasibility reasons and could reverse their earlier hypothesis.
In this way, denser exploration of the uncertainty space can help decisionmakers understand the specific performance characteristics of their strategies, more-precisely consider their strengths and weaknesses, and weigh the trade-offs among them.
In sum, this case study illustrates how scenario planning can be a useful and sometimes necessary step toward a full DMDU analysis. It is familiar and can be useful in building capacity and cooperation. It also shows that scenario planning–based approaches can have limitations. In the Guidance on How Best to Apply DMDU section, we elaborate on the opportunities DMDU methods offer and discuss how best to harness them.
Case study authors: Nidhi Kalra and David Groves
Acknowledgments: The authors would like to thank Dagmar Llewellyn and Lucas Barrett (Reclamation) for providing background information on and data from the PRNMBS and for their review of this case study and helpful suggestions.
The United States Geological Survey’s Acme Gage is located north of the city of Roswell, New Mexico. One of the first signs that the Pecos River is drying is when discharge at the Acme Gage drops below 5 cubic feet per second (cfs). (Return to text)
A 50-percent on-farm efficiency rating means that the ID needs to divert twice the amount of water that the crops need. A 75-percent on-farm efficiency rating means that the ID needs to divert 25 percent. (Return to text)
This is different from the approach in SSJRBS (Case Study 2), where the results showed several trade-offs across metrics. That is, in SSJRBS, different strategies worked well for different metrics, and the case study aggregated results across many futures, asking, “For a particular metric, in how many futures is a strategy successful?” (Return to text)