Adapting to a Changing Colorado River
In January, 2010 the U.S. Bureau of Reclamation and representatives of the seven states that use Colorado River water began a study of the long-term future of the Colorado River basin system. The Colorado River Basin Study, released in December 2012, was the first collaborative and systematic evaluation of the vulnerabilities of the Colorado River management system and management strategies to reduce them. RAND was asked to join the effort in January 2012 to help develop an innovative decision analytic approach for the study and to assist in its implementation.
In the tabs below, we tell the story of the Colorado River's vulnerabilities and strategies to reduce them. We use interactive graphics developed for the study to showcase key findings. We encourage you to interact with the graphics—while we describe in broad terms what the elements show and how water managers and policymakers might tap them, the interactivity offers much more to explore.
Readers interested in even more detail can view RAND's report, Adapting to a Changing Colorado River: Making Future Water Deliveries More Reliable Through Robust Management Strategies.
What's at Stake? The Mighty Colorado River System
The Colorado River system
- provides water and power for 40 million people in 7 states and 22 tribes
- supports billions of dollars annually in economic activity
- irrigates 15% of U.S. crops (5 million farm acres), and
- is a lifeline for two dozen national parks, wildlife refuges, and recreation areas.
But as the map shows, the Colorado River is far more than just the river itself. It is a “system” that spans seven Basin States. In the Upper Basin are Colorado, Utah, Wyoming, and New Mexico. The Lower Basin includes California, Arizona, and Nevada. The system also includes the river, tributary streams and rivers, and water storage and delivery infrastructure—dams and reservoirs, hydropower facilities, canals, aqueducts, and pumps.
Much of the Colorado Basin’s infrastructure is operated and maintained by the U.S. Bureau of Reclamation (Reclamation), which ensures that major water users reliably receive their water deliveries each year.
How Uncertain Is the Future of the Colorado River System?
Determining what strategy is best over the next 50 years is challenging. In recent decades, federal managers and water users have grown increasingly concerned about the delivery of reliable water supplies from the Colorado River.
In 1922, stakeholders hammered out the Colorado River Compact (the Compact)—the legal document that determines how water is allocated to the Upper and the Lower Basins. Over time, demand for water in the Lower Basin has soared, already exceeding the 7.5 million acre-feet (maf) volume allocated in the Compact. The Compact has continued to work without leading to shortages, but only because the Upper Basin has yet to use its full allocation. But demand continues to grow in the Upper Basin states, and the region also faces increasingly variable and possibly declining supplies. This all means that future imbalances between supply and demand are inevitable.
But just when they will occur, or how large they will be, is deeply uncertain. Reclamation planners created a set of six demand and four supply scenarios that could play out over the next half century and face a sharply different future depending on which scenarios emerge. As it turns out, the uncertainty related to water supply is much higher than uncertainty with demand, and the potential gap between supply and demand varies dramatically in different scenario combinations.
Future Supply and Demand Imbalances
The gap between future demand and supply in the Colorado River basin is deeply uncertain. The graphic below shows historical demand and supply estimates for the Colorado River Basin (1920–2008) on the left and plausible ranges for future demand (red and orange) and supply (blue) on the right, going out to 2055.
The graphic depicts historical water supply as a single blue line on the left and, on the right, the range of supply for one of four future supply scenarios, as selected below. The small horizontal blue dashes indicate the middle projection for each year. One scenario is based on the recent historical record—with sequences of flows from 1906 to 2007 starting in different years. The second and third scenarios are based on streamflow estimates derived from tree ring data and other paleoclimatological proxies. The fourth scenario is derived from projections of future climate conditions. These are based on simulations from 112 global climate models run under different projections of global greenhouse gas emissions.
Instructions: Select the demand and supply scenarios shown by clicking the appropriate button.
The graphic also depicts six future demand scenarios without more programs and incentives for water conservation:
- current projected growth
- slow growth, with an emphasis on economic efficiency
- rapid growth because of economic resurgence
- rapid growth, with current preferences toward human and environmental values
- enhanced environment due to expanded awareness
- enhanced environment due to increased stewardship accompanying economic growth
The graphic shows that, on average, the higher the demand and lower the supply, the larger the imbalance will be. For example, if you click on the Rapid Growth (C1) demand scenario and the Future Climate Growth supply scenario, demand is projected to be higher be than supply across much of the supply range.
How Did Reclamation Develop a Strategy to Address the Challenges it Faces?
Because uncertainty about the future condition of the Colorado is so great, the Study Team used Robust Decision Making (RDM) to help structure an analysis of vulnerabilities and strategies to address them. This study was the first of its kind for the Basin, moving away from planning for the most likely future condition—which may be very different from what actually happens—to planning for when things go wrong.
RDM avoids the downsides of traditional methods by running decision methods “backwards.” Traditional methods try to predict future conditions and then make decisions about what works best for the prediction. But, as shown in the figure below, RDM turns that approach on its head.
Robust Decision Making (RDM) Process
It starts with a proposed strategy—in this case, Reclamation’s current management approach. RDM runs models hundreds to millions of times to identify conditions that best distinguish future scenarios where the strategy does and does not meet its goals. It then identifies steps so the strategy may succeed over a wider range of future scenarios. The graphic shows how RDM works, with emphasis on how it involves stakeholders in the process. Of its four steps, three rely on conversations with policymakers, ensuring they are integral in the process. Only the second step—system evaluation that involves simulation modeling—is entirely analytic and researcher-driven.
In the Basin Study, RDM helped to: structure the analysis of a large set of future scenarios and strategies (step 1); determine outcomes of those strategies and scenarios using simulation modeling (step 2); identify key vulnerabilities, based on the modeling outcomes (step 3); and evaluate key tradeoffs among strategies (step 4). Reclamation and the Basin States are now working with this information to implement near-term options and further refine a long-term strategy for the Basin.
Next: Challenges »
What Are the Colorado River's Vulnerabilities?
Given the many plausible futures for the Colorado River system, the Basin Study Team identified the key vulnerabilities that threaten Reclamation’s Current Management approach—the starting point to the analysis.
Doing this required not only a way to model what could happen over time in the Colorado River system but also metrics to determine how well the Current Management approach would perform in different conditions. To understand the river system, the Basin Study Team used a computer model developed by Reclamation called the Colorado River Simulation System. To understand how well the Current Management approach does, we drew on a wide range of performance metrics. Those metrics focused on such indicators as water deliveries, electrical power resources, and ecological resources at various locations along the river.
While the full study used over 30 different performance measures, we tell the story here through just two: potential shortfalls in the delivery of water from the Upper to the Lower Basin (Lee Ferry deficits) and water storage in the linchpin reservoir of the Lower Basin (Lake Mead pool elevation). These two broadly summarize the reliability of water delivery to the Upper and Lower Basins, respectively. If there is a Lee Ferry deficit, then deliveries to the Upper Basin would need to be reduced to increase flows to the Lower Basin. And the health of the Lower Basin system and deliveries to the Lower Basin states similarly are closely tied to the Lake Mead pool elevation.
When the Basin Study Team used the models, the two key metrics, and the Current Management approach, it found the approach is vulnerable to Lee Ferry deficits and low Lake Mead pool elevations under many future conditions. We also found that uncertainty about future supply and demand leads to significant uncertainty in the severity of these risks.
Upper and Lower Basin Vulnerabilities
Just how vulnerable is the Colorado River Basin, as seen in two key metrics: Lee Ferry deficit and Lake Mead pool elevation? These interactive graphics show what we may face across the supply and demand scenarios.
Lee Ferry Deficit
In the top graphic, you can select among the various supply scenarios and see the percentage of years in which a Lee Ferry deficit occurs under each. The default setting shows that Lee Ferry deficits occur with increasing frequency over time under the Future Climate supply scenario across all demand scenarios. In contrast, no deficits would occur for the Historical supply scenario under any of the demand scenarios. The Paleo and Paleo/Historical scenarios show Lee Ferry deficits but less often than under the Future Climate scenario.
Instructions: Select the supply scenario using the drop-down menu. Click the legend color swatch to isolate results for individual demand scenarios.
Lake Mead Pool Elevation
The elevation of Lake Mead broadly reflects conditions for the Lower Basin, and this graphic lets you see how elevations could change under the supply and demand scenarios.
The graphic shows that in the Historical supply scenario, the Lake Mead elevation remains higher than the critical 1,000-foot level through 2060 for most of the simulations. In contrast, under the other supply scenarios, the lake would fall below the 1,000-foot level in many of the simulations. In fact, in the Future Climate scenario it falls below the 1,000-foot level by 2050 in about half of the simulations.
Instructions: Select the supply scenario using the drop-down menu. Click the legend color swatch to isolate results for individual demand scenarios.
What Conditions Lead to Vulnerabilities?
Beyond understanding how vulnerable their Current Management approach may be, water managers also must understand what external conditions lead to this and what actions they might take in response.
The Basin Study Team looked for a set of future conditions most likely to adversely affect the two basins. For the Upper Basin, the Lee Ferry deficits are likely to occur when two types of conditions occur together in the future: (1) long-term average streamflow declines beyond what has been observed in the recent historical record (which turns out to be below 13.8 maf per year), and (2) there is at least an eight-year period of consecutive drought years where the average streamflow dips below 11.2 maf per year. These are collectively referred to as Declining Supply vulnerable conditions.
Similarly, the external conditions that lead to the most vulnerability for Lake Mead pool elevations are: (1) the long-term average streamflow at Lees Ferry falls below the historical average of 15 maf; and (2) an eight-year drought with average flows below 13 maf occurs. We call these vulnerable conditions Low Historical Supply, as similar conditions have been experienced in the recent past.
What Leads to Colorado River Vulnerabilities?
What external conditions most adversely impact the Upper and Lower Basins? This interactive graphic shows that for the Lee Ferry metric, the Upper Basin is vulnerable when two future conditions occur—when long-term average streamflow declines beyond what has been observed in the recent historical record (which is 13.8 maf per year); and when there are eight years of consecutive drought where the average flow dips below 11.2 maf per year. These are reflected in the two axes on the graphic. These are collectively referred to as the Declining Supply vulnerable conditions.
The two external conditions that constitute Declining Supply are the two axes on the graphic, while the shaded area on the graph represents the intersection of those two external conditions. Results for a subset of the 23,508 traces—individual time sequences of streamflows—that the model evaluated are shown below. Traces that meet both of these conditions lead to a Lee Ferry deficit 87 percent of the time. That means that 87 percent of the vulnerable traces (the Xs) fall within the shaded area, while the remaining 13 percent of vulnerable traces are outside of it. The Os represent the traces that are not vulnerable. By selecting results corresponding to different supply scenarios in the key, you can clearly see that these vulnerable conditions correspond to conditions consistent with the Future Climate and Paleo supply scenarios.
Instructions: Select the performance metric (Lee Ferry Deficit or Lake Mead Levels) using the drop-down menu. Select a supply scenario at the bottom of the graph to isolate those results.
To see which external conditions most affect the Lake Mead Pool Elevation metric, you may click through various supply scenarios, as you did for the Lee Ferry Deficit metric. Those with greatest effect on Lake Mead occur when the long-term average streamflow at Lees Ferry falls below the historical average of 15 maf; and an eight-year drought with average flows below 13 maf occurs. We call these conditions Low Historical Supply vulnerable conditions, and they describe 86 percent of all instances (traces) leading to unacceptable results, as you can see by clicking through the various scenarios.
Next: Solutions »
How Would Different Management Options Reduce Vulnerabilities?
So how can water managers reduce the vulnerabilities in their Current Management approach? The Basin Study first evaluated an array of options to increase supply or to decrease demand in ways that improve system performance and reduce vulnerabilities.
Ultimately, 80 options were evaluated according to cost, water yield (supply increased or demand reduced), availability, and 16 other criteria, including technical feasibility, permitting risk, legal risk, policy risk, and energy intensity; they are summarized by eight categories:
- agricultural conservation
- efficiency of water use for energy production
- importing water into basin
- local supply
- municipal and industrial (M&I) conservation
- watershed management
But a comprehensive approach to reduce vulnerabilities requires more than a simple set of options. So RAND developed an interactive portfolio development tool for use in the Basin Study. It allowed stakeholders and the study team to develop four portfolios of prioritized options to augment supply and reduce demand. From the 80 options evaluated, the Basin Study Team identified four portfolios of options: Portfolio A (Inclusive), Portfolio B (Reliability Focus), Portfolio C (Environmental Performance Focus), and Portfolio D (Common Options).
Options and Strategies to Reduce Colorado River Basin Imbalances
To develop different approaches to reduce the Current Management approach’s vulnerabilities, the Basin Study Team first evaluated an array of options—80 in total—that either increase supply or decrease demand; these also could improve system performance and reduce vulnerabilities. The Basin Study team worked with the policymakers and stakeholders to prioritize the 80 options into four “portfolios.”
Each portfolio corresponds to a different strategy, and this interactive graphic allows you to select a portfolio to see a brief description and a list of the included options. The symbols for the options indicate their type (color and symbol shape), the time frame available for rolling out the option (horizontal axis), and its yield (size of the symbol). Rolling the computer mouse over an option shows its details.
Instructions: Select the portfolio to review using the drop-down menu. Click on the option type to highlight them in the list.
The Basin Study Team next turned to evaluating each portfolio across all the supply and demand scenarios and compared how they affected the vulnerabilities associated with the Current Management approach. While vulnerabilities grow over time without additional options, implementing the different portfolios lessens these vulnerabilities. Each portfolio has different impacts on vulnerabilities (and does so in different degrees), and Portfolio A (which includes all the options) unsurprisingly does the most to reduce vulnerabilities.
Effect of Options and Strategies on Reducing Vulnerabilities
How does each portfolio (or strategy) reduce the vulnerabilities in the Current Management approach (or baseline)?
This interactive graphic shows the percent of traces in which the two key performance metrics—Lee Ferry deficits and Lake Mead pool elevation—are vulnerable. On the left are results under the Current Management approach; on the right are results for one or more portfolios. By clicking on different portfolios, you can see how they reduce the vulnerabilities of the Current Management approach over time, relative to the baseline.
For the baseline, those vulnerabilities grow over time without additional options—they grow from 2 percent to 16 percent of instances (traces) between 2012 and 2060 for the Lee Ferry deficit metric and from 13 percent to 40 percent for the Lake Mead pool elevation metric.
If you leave this graphic set at “all scenarios” and click through the portfolios, as expected, Portfolio A (which includes all options in the other three) most greatly reduces the percentage of traces in which a vulnerability occurs in the Current Management approach; Portfolio D, with fewer options, reduces them the least. Portfolio B is more effective than Portfolio C in reducing challenges to Lake Mead; Portfolio C is more effective than Portfolio B in reducing Lee Ferry deficits.
If you look at the portfolios’ performance under specific scenarios, only under Historical supply do they all eliminate all vulnerabilities. But under the Future Climate scenario, significant issues remain, even if the more inclusive Portfolio A gets put in effect.
Instructions: Select the portfolio to compare to the baseline using the check boxes on the bottom-left. Click on a supply scenario to focus the results.
Next: Tradeoffs »
What Are the Key Tradeoffs Among Portfolios?
Given the effectiveness of Portfolio A, it might seem the obvious choice. But effectiveness is not the only factor policymakers must consider. The portfolios also vary in terms of how much they would cost. This means there is a tradeoff between a portfolio’s effectiveness and its cost. RDM helped the Basin Study Team to combine the cost and vulnerability results to draw out distinctions and tradeoffs among the four portfolios.
That evaluation showed that across the portfolios there is indeed a tradeoff between reducing the vulnerabilities in the Current Management approach and what it would cost to do so. This becomes apparent in looking at the conditions most stressing to the Basin identified earlier—Low Historical Supply and Declining Supply.
By looking at the two performance metrics for the Declining Supply vulnerable conditions, for example, the Basin Study Team found that Portfolio C (Environmental Performance Focus) offers significant Lee Ferry deficits vulnerability reduction with a range of costs less than all portfolios other than Portfolio D (Common Options). In contrast, Portfolio B (Reliability Focus) offers just as much vulnerability reduction as Portfolio A (Inclusive), but at lower costs.
Tradeoffs Among Strategies
Effectiveness is only one factor that policymakers need to consider: How cost-effective the portfolios are relative to one another is also critical.
This interactive graphic shows the range in total annual costs for putting each portfolio in place, as evaluated across two different vulnerable conditions—Low Historical Supply and Declining Supply. We again show the two key metrics: the Lee Ferry Deficit or Lake Mead Pool Elevation (as selected on the right), with the graphic’s default showing the Lee Ferry Deficit and “All Traces” relative to the baseline (the Current Management approach).
Portfolios with lowest costs (farthest to the left in the graph) and that reduce vulnerabilities the most (lowest on the graph) are preferred. The portfolios are distinguished by color, and the baseline level of vulnerability in this case is shown by the dotted line running across the chart (6 percent for All Traces). The ranges in costs are shown by the percentile markings, from .25 to .75. The range in implementation costs occurs because of the dynamic nature of the portfolios—they are designed to implement options only when conditions warrant them.
Instructions: Select the performance metric and conditions in the upper right. Select the color swatches to highlight results corresponding to different portfolios. Select different percentile symbols to isolate those results.
This figure shows that all the portfolios reduce the percent of years vulnerable substantially, but that a tradeoff exists between that reduction and cost. There are only modest differences among the portfolios in costs and how they reduce the vulnerability of the Current Management approach. This is also true if you click on the other performance metric for Lake Mead Pool Elevation.
But when you click on conditions most stressing to the Basin—Low Historical Supply and Declining Supply—more substantial tradeoffs emerge among the four portfolios. By looking at the two performance metrics for the Declining Supply, for example, you will see that Portfolio C offers significant Lee Ferry reductions at costs less than all portfolios other than Portfolio D. In contrast, Portfolio B offers just as much reduction as Portfolio A, but at lower cost.
Next: Robust Strategy »
How Does the Study Help to Develop a Robust, Adaptive Strategy for the Colorado River Basin?
The Basin Study did not select a single portfolio to implement. Instead, managers and stakeholders are using the analysis of the portfolios to understand which options would provide a strong foundation for a robust, adaptive strategy for the Basin.
Not all options are available at the same time. But the analysis helped determine which options are needed most often once they are available over time. Options that are implemented across many future traces—individual time sequences of streamflows—soon after they become available can provide the foundation of an initial robust strategy.
Specifically, the Basin Study Team was able to show water managers how fast options evaluated in the Basin Study need to be rolled out. For example, if the Basin were to consider all options—as represented by Portfolio A (Inclusive)—we found that there would be an urgent, immediate need for municipal and industrial conservation measures. We also found a pressing demand for agricultural conservation, efforts to transfer more supplies into the Basin, and further investigation of desalination efforts for the Salton Sea.
Which Options Form the Foundation of a Robust Strategy?
Policymakers, of course, must consider how over time they roll out the options of any portfolio, and those they take up first can provide the foundation for what may develop into a robust strategy.
This interactive graphic shows results for different portfolios (selected on the right) for the Low Historical Supply or Declining Supply vulnerable conditions (also selected on the right). Results in the lower-right shaded corner are near-term, high-priority options. As you can see, municipal and industrial conservation are required in more than 90 percent of all instances (traces) examined, with a minimum delay of zero years. There is also urgency for the agricultural conservation, supply transfers, and increases in efforts in desalination-Salton Sea options; thus, these are also near-term, high priority options.
Instructions: Select the portfolio and conditions evaluated.
Next: Conclusion »
A First of a Kind Study
The Basin Study—with its use of RDM to help water managers and stakeholders plan for the future under deep uncertainty—was the first of its kind. It moved away from traditional “planning as usual” for the most likely future condition—which may be very different from what actually happens—to planning for what do when things go wrong, thus ensuring resilience over a wide range of plausible futures.
RDM is not a “black box” approach, where data get entered and analyzed and a recommended strategy is presented to policymakers; rather, RDM actively engages policymakers and key stakeholders from beginning to end. In part, as a reflection of the value of the RDM approach, the Basin Study received a U.S. Department of Interior 2012 “Partners in Conservation” award, which recognizes “organizations that have achieved exemplary conservation results through public-private cooperation and community engagement.” Moreover, planners across the Basin are now using the results of this study to support working group deliberations around which specific investments to take as a first step toward a robust, adaptive strategy for the Colorado River.