Rethinking Vulnerability on the Colorado River



On the Colorado River (as elsewhere), severe drought is useful for illuminating sources of water supply vulnerability, focusing attention on deficiencies in water allocation and management. A major drought study in the early 1990s, and experience with real drought a decade later, both have been useful in understanding vulnerability as a function of several factors working in consort with drought, including water allocation, reservoir operations, water demands, and climate change. Over this relatively short time-frame, vulnerability has shifted considerably, and will undoubtedly continue to change further in coming decades. Understanding how vulnerability is shifting is central to identifying future management pathways and reform options for the river.

In October of 1995, Water Resources Bulletin (now Journal of the American Water Resources Association) devoted an entire issue to the findings of the “Severe, Sustained Drought in the Southwestern United States” study (seeYoung 1995). That project (hereafter termed the “SSD study”) examined the alarming but seemingly very remote scenario of modern-day catastrophic drought on the Colorado River, with the aim not merely of assessing system vulnerability to water supply shortages, but also of informing and inspiring the institutional reforms needed to avoid any such future calamity. Included in this focus was a consideration of how shortages might actually be allocated given the existing legal and management framework, a subject that quickly went from theoretical to real as severe drought gripped the region and the basin states were forced into “shortage sharing” negotiations.

By most accounts, the study was innovative and ambitious, especially in its treatment of possible institutional reforms. Specifically, it featured a “gaming” exercise that allowed players (representing states) to mitigate drought impacts through both intrastate and, more significantly, interstate water marketing – a subject that remains largely taboo in policy circles. It also introduced the idea of mitigating and reallocating water shortage risk through coordinated reservoir operations, an idea that did find expression in the eventual basin states agreement in 2007 (U.S. Bureau of Reclamation 2007). However, while the SSD study was a milestone in suggesting that new scientific information could be used to rethink longstanding allocation practices, it did so by characterizing vulnerability on the river in a way that now, just a decade later, seems dated if not erroneous. Water supply vulnerability has changed, and with it, so has the context of all future discussions of allocation and reallocation. The change lies in the complex relationship between drought, water demands, and the prospect of chronic reductions in streamflows from climate change. This paper discusses these risks as a function of social and environmental vulnerability and alterations in physical conditions.

The Severe Sustained Drought (SSD) Study

The SSD study investigated the vulnerability of the modern-day Colorado River to a drought of severity and duration beyond anything theretofore seen in the historic record. The study initially used the paleo drought from 1579 to 1600, as documented in tree rings, followed by a wet 16-year recovery period, as the basis for hydrologic inputs into water models (Young 1995). An event of this magnitude was thought to pose the greatest threat to the two major reservoirs on the system (lakes Powell and Mead), and thereby posed the greatest challenge to the interbasin, interstate and international allocation scheme codified in the Law of the River.1 Water allocation models reflecting modern legal/institutional arrangements, infrastructure, and projected demand patterns were used to track and quantify the impacts of the unfolding drought, while additional models and game theory techniques were subsequently applied to investigate potential coping strategies.

The study showed the system to be remarkably reliable during drought – so much, in fact, that the researchers decided to re-order the drought (i.e., all 22 years in the paleo drought were arranged from wettest to driest), thereby generating a period of 16 consecutive years of below average flows with an annual mean of 9.57 maf (million acre-feet), compared to a long-term term average typically estimated at nearly 15 maf. This re-ordering was estimated to lengthen the return interval range of the event from 400-700 years (in its original form) to 2,000-10,000 years. In the scenario, once the run of drier than average years had begun, Lake Powell took only a dozen years to drop to dead storage (where it remained for 7 years). Lake Mead, however, fared dramatically better, hovering near 20 maf (about two-thirds full) until Powell was empty, only then dropping sharply, but still never falling beyond a low of 7.5 maf (Figures 1 and 2). In the worst year of the SSD scenario for water deliveries, roughly 59 percent of projected Upper Basin consumptive demands went unmet, compared to just 3 percent in the Lower Basin; shortages in the Lower basin were primarily confined to surplus uses. In short, the system was found to be quite effective in serving existing Lower Basin consumptive uses (although instream uses suffered), with most problems limited to Lake Powell and Upper Basin users. This, as of 1995, was our collective assessment of system reliability in a seemingly worst-case scenario.

Figure 1.

Lake Powell storage (maf) under actual and SSD scenario condition.

Figure 2.

Lake Mead storage (maf) under actual and SSD scenario condition.

The Current Drought

Upon publication, the SSD study received only modest attention; in fact, up to 2001, much of the interbasin debate was focused on the allocation of Lower Basin surpluses (U.S. Bureau of Reclamation 2001). This situation did not change until after the new millennium when an ongoing series of dry years lowered natural inflows into the basin: 62 percent of the 30-year average in 2000, 59 percent in 2001, 25 percent in 2002, 51 percent in 2003, 49 percent in 2004, 105 percent in 2005, 73 percent in 2006, 68 percent in 2007, and 102 percent in 2008.2 Coincident with this modern drought was a precipitous decline in storage, shown in Figures 1 and 2, with the year 2000 matched up with year 6 of the SSD scenario to synch the start of reservoir declines in the actual and hypothetical drought scenarios. In both traces, Lake Powell drops quickly (Figure 1), slightly less so in the SSD scenario only because the re-ordered drought starts out with the least severe years at the beginning. The traces for Lake Mead diverge radically (Figure 2), however, suggesting a very different story than envisioned in the SSD.

Changing Vulnerability: Lower Basin

The recent decline of Colorado River storage is generally attributed to drought. Indeed, the decreasing level of Lake Powell closely follows the decreasing trend in inflows expressed as virgin flow at Lee Ferry (Figure 3). Similarly, reduced storage in Lake Mead could be expected to reflect changes in Lake Mead inflows, which are primarily a function of Lake Powell releases. However, these releases are remarkably consistent during dry periods at 8.23 maf/year (at least until Powell is empty), dictated more by legal requirements and hydropower regimes than hydroclimatic conditions.3 Reclamation's provisional data suggests that inflows to Lake Mead have not changed significantly during the current drought compared to early 1990s. Compare Lake Mead in two critical 5-year periods: 1990 to 1994, and 2000 to 2004 (Figure 4). In both periods, inflows to Lake Mead are approximately 9 maf/year, but while this was sufficient to keep storage levels stable in the first period, the second period featured a decline of roughly 9 maf.

Figure 3.

Lake Powell storage and estimated Lee ferry virgin flows (maf).

Figure 4.

Lake Mead storage and estimated inflows (maf).

To understand how the risk of shortage in meeting demands from Lake Mead has changed, a revised analysis must go beyond drought and a traditional supply-side analysis. For Lake Powell, 2000-2004 looks very different than 1990-1994 because of reduced drought inflows, but for Lake Mead, the difference is mostly explained by the growth of Lower Basin demands. Reclamation provisional data suggest that Lower Basin consumptive uses (defined here as the sum of deliveries from the mainstem to Arizona, California, and Nevada, and treaty-required deliveries to Mexico) in 2000-2004 versus 1990-1994 grew by approximately 1 maf/year (from 8.7 to 9.7 maf/year), sufficient to pass the threshold that determines whether or not Lake Mead is stable or declining. Key to this increased consumption was the completion of the Central Arizona Project, which allows Arizona to use its Colorado River entitlement to offset ground water depletions. The project was “substantially completed” in 1994, and very quickly was put into full operation. Mainstem deliveries to Central Arizona Project averaged 1.5 maf/year from 2000 to 2004, roughly 3 times higher than the Central Arizona Project demand projections used in the SSD study (derived from assumptions in the 1991 Annual Operating Plan and economic modeling that assumed Central Arizona Project water would be prohibitively expensive for many users). This erroneous assumption, more than any other single item, explain why the SSD and actual drought traces for Lake Mead differ so significantly.

Changing Vulnerability: Upper Basin

The possibility of drought causing shortages in the Upper Basin remains a valid concern, but assessing vulnerability is again a more complex story. Since the extremely dry year in 2002, analysts in the Upper Basin have speculated about whether or not a drought-inspired “call on the river” is a real possibility, how that legally-complex issue would be determined, and how a call might be administered. A call would be triggered if storage in Powell precluded fulfilling the compact-defined delivery requirement of 75 maf over 10-years (or 82.3 maf including half of the Mexican obligation). Although this is a scenario worth considering, as in the Lower Basin, this focus on extreme drought must be coupled with concern over a more chronic concern: in this case, the threat of decreased mean flows due to climate change. Most climate models suggest future declines this century in Colorado River flows – as described in Appendix U of the 2007 Environmental Impact Statement (U.S. Bureau of Reclamation 2007). A modest decline of 10 percent would mean a roughly 1.5 maf/year reduction in average annual supplies. Under many legal interpretations, this full amount would come out of the Upper Basin apportionment.

Ironically, the Upper Basin already “lost” roughly this amount last century as estimates of the mean flow of the river were reduced. This familiar story involves the error of compact negotiators in the 1920s that unwittingly used the hydrology from an exceedingly wet period to over-allocate the river. Due to this error, the Upper Basin for many years has reluctantly assumed that their reliable annual share of the river may be closer to 6 maf than the promised 7.5 maf. Should impacts of climate change reduce this value further, as in this scenario to 4.5 maf/year, then shortages can quickly go from theoretical to real, as Upper Basin consumption (from 1996 to 2000) averaged 4.4 maf/year (accounting for evaporation losses). Certainly responses to this climate change scenario would not play out without legal and political wrangling and a search for engineering solutions, but the central message is unchanged: extreme droughts are a real concern, but assessing their impact on water supply vulnerability requires considering other factors, namely long-term changes in average demands and inflows.

Looking Forward

The drought has prompted several management innovations aimed at balancing the regional water budget. One of the first and most significant accomplishments has been the enactment and implementation of the “Interim Surplus Guidelines,” which includes California's 4.4 Plan reducing its annual use from over 5.0 maf/year in the 1990s to just 4.4 maf/year since 2003 (several years ahead of schedule), a reform made possible in large part through aggressive on-farm conservation and the subsequent reallocation of saved water to urban areas. Similarly, while urban growth continues at a rapid pace throughout the basin, several major southwestern cities have accomplished this without a corresponding increase in water consumption. Another major accomplishment is the new interim operational guidelines for shortage sharing and the modified reservoir operating regimes (including some anticipated in the SSD research) that better protect storage (U.S. Bureau of Reclamation 2007). These guidelines not only outline a more orderly and phased schedule of reductions when reservoirs start to decline, but establish a framework through 2026 for cooperative learning and planning among Federal and state interests focused on new baseline supply estimates. The new rules also include a program allowing some conserved water (“intentionally created surplus”) in the Lower Basin to be marketed to support new or, potentially, existing demands elsewhere. The program is new, and thus untested, and has caps on the amount of water available, but it does institutionalize an incentive for demand management that heretofore was largely absent in the Law of the River. Efforts to reduce “regulatory wastes” (such as over-deliveries and reservoir spills) are also underway, as is the search for remaining opportunities to augment river flows.

Looking back, it is fair to conclude that the SSD project was a major step forward in thinking about drought on the Colorado River, and in focusing renewed attention on issues of allocation, reallocation, and the future of the river. But to the extent that the study was intended to illuminate the vulnerability of Colorado River water users to shortages, the scenario considered was ultimately too narrow to capture the complex interplay between drought, growing demands, and the then-nascent research about climate change and the assumption of hydrologic stationarity. Today, it no longer takes a significant SSD-quality event to cause significant disruptions to water supplies, and any such disruptions are not confined to Upper Basin users (as the SSD analysis suggested). Largely due to our ongoing experience with real drought, the research community now has a significantly better understanding of the forces that shape vulnerability on the Colorado River, and has the attention of water managers and other public policy officials to an extent not enjoyed by the SSD research team.

End Notes

  • 1

    The basin is legally divided into an Upper and Lower Basin at Lee Ferry, with Lake Powell providing the upstream storage needed to ensure compact-specified water deliveries to serve downstream interests, which then pull water from Lake Mead. The 1922 compact allocates 7.5 MAF (million acre-feet/year) to each Basin, and a treaty later (1944) provided 1.5 maf/year for Mexico, for a total apportionment of 16.5 maf/year. This total is higher than actual annual flows, a problem magnified by compact rules that appear to obligate the Upper Basin to allow almost the full Lower Basin and Mexican obligations to pass downstream before the Upper Basin can take its share.

  • 2

    Statistics in this paper regarding inflows, storage, diversions, and consumptive use have primarily been compiled from records prepared by the U.S. Bureau of Reclamation. In many cases, these figures are later amended and updated, and much of the data regarding diversions and consumption since 2001 is provisional (as noted in the text). Inflow data is provided at:; demands are summarized in the Consumptive Uses and Losses reports ( and the annual Decree Accounting Reports ( Deliveries from the Central Arizona Project are from

  • 3

    Lake Mead also receives inflows from tributaries (many ungaged) between the two reservoirs, estimated here (as is commonly done) at a constant 0.77 maf/year.


Funding for this research was provided by the NOAA Human Dimensions program (now part of the Sectoral Applications Research Program).

Author Bios and Contact Information

Douglas Kenney is Director of the Western Water Policy Program at the Natural Resources Law Center, University of Colorado Law School. He has written extensively on several water-related issues, including river basin and watershed-level planning, western US water law and policy reform, and the design of institutional arrangements. Dr. Kenney has served as a consultant to a variety of local, state, multi-state, and federal agencies, as well as governments in Asia, Africa and Latin America.

Andrea Ray is a Physical Scientist with the NOAA Earth Systems Research Laboratory in Boulder, CO. Her research interests are on the potential use of climate information and forecasts in natural resource management, including the analysis of management decision processes and how these condition the potential usability of forecasts. Dr. Ray's areas of expertise include climate and society interactions, water management and development in the western U.S., adaptive management, policy sciences, and qualitative and context-sensitive research methods.

Ben Harding is a water resources engineer with the Earth and Environmental Division of AMEC, a leading worldwide environmental consulting firm, based in the Boulder, CO office. He has more than 35 years of diverse experience in water resources engineering and software development, including projects entailing the design, development, and use of hydrologic models, decision support systems, hydraulic models, water-quality models, geographic information systems and databases.

Roger Pulwarty is Director of the National Integrated Drought Information System, established in 2006 to ensure collaboration between different government agencies on drought-related issues. His research focuses on the climate-weather nexus, society-environment interactions, and in the design of effective services to address associated risks. His research and applications focus on natural resources policy, development and decision-making in the western U.S., Latin America, and the Caribbean. Dr. Pulwarty was a lead author on the IPCC Fourth Assessment Report Working Group 2.

Brad Udall is Director of Western Water Assessment (Boulder, CO), one of seven RISA (Regional Integrated Sciences and Assessments) programs funded by the Office of Global Programs at NOAA. These programs are designed to develop partnerships with regional stakeholders and tailor NOAA data products to meet their needs. His research is focused on impacts and adaptation to climate change in the western US, with a particular emphasis on Colorado River hydrology and policy.

All the authors are affiliated with the Western Water Assessment (a joint partnership between the National Oceanic and Atmospheric Administration and the University of Colorado). Direct inquires to lead author Kenney at (University of Colorado, UCB 401, Boulder, CO 80309-0401).