Impacts of Climate Change and Land Use on the Southwestern United States
Coping with Severe Sustained Drought in the Southwest

How severe would a "really bad" drought in the Colorado River basin be?

Abstracted from Meko and others (1995) and Tarboton (1995)

The scarcity of water in the southwestern United States is exacerbated by the occurrence of frequent and prolonged droughts. Assessments of the historical and prehistorical occurrences of drought in the Colorado River basin were made using streamflow measurements and estimates for the period 1906-83 and treering reconstructions of flows from the 1500s to 1960s. If a drought is simply a continuous period with flows below the long-term average, then severity of droughts can be described in several ways depending on the resource being affected. For example, from the Colorado River flow series, severe sustained droughts could be identified as the periods with the maximum accumulations of flow deficits (relative to the long-term average flows) or as periods in which a hypothetical level of reservoir development, management, and demand would suffer the greatest reservoir- storage depletions. In the present analysis, results were relatively similar using either approach, and the resulting "worst-case" drought scenarios were used in other studies of the likely impacts on hydrologic, ecological, and social systems of the Southwest.

Cross section of 
tree rings

Figure 3. Cross section of tree rings (Photo courtesy the NOAA Paleoclimatology Program)

Meko and others (1995) described the physical basis and limitations of tree rings as indicators of drought severity in the Colorado River basin and identified severe droughts over the last 400 years using tree-ring series from throughout the region (Figure 4). The longest, most severe drought during the last 400 years appears to have been a sustained drought from about 1579-1598. During this 20-year period, flows at Lees Ferry (Compact Point in Fig. 1) are estimated to have been about 81% of the long-term average; altogether this drought amounted to the "loss" of almost 4 years' worth of flow. This same 20-year period includes the two driest 5-year intervals (1590-1594 and 1583-1587 at only 66% of normal flows) and the driest 10-year interval (1584-1593 at 72% of normal flow), so that it was both long and severe. This drought period is also notable for its widespread impact on the upper Colorado River basin (above Compact Point); the 1579-98 period was in the driest half of all 20-yr periods at 18 tree-ring sites used to study it (scattered throughout the upper basin) and was among the driest 6% of 20-yr periods in at least one tree-ring series from each runoff-producing region. The drought was most severe in the San Juan and uppermost Colorado River basins.

Annual streamflow totals graph

Figure 4. Annual streamflow totals and (heavy) 20-year moving averages of flows at Lees Ferry (Compact Point), as estimated from tree-ring series.[MAF = millions of acre-feet per year.] (Modified from Meko and others, 1995)

More recently, but on a much smaller scale, a comparable sustained drought among historical flow measurements is the 1943-1962 period when flow in the Gila River of southeastern Arizona was about 2/3 of normal. Notably the nearby Salt River did not experience a drought this severe during that time. In the water-producing upper Colorado River basin (above Compact Point), however, the first several decades of this century were overall quite wet relative to the four centuries depicted by the tree-ring streamflow reconstructions studied by Meko and others. On average, the 1910-1940 period yielded more than 16 million acre-feet per year or about 20% more flow than the long-term average. Thus the brief 100 years of historical streamflow measurements of the Colorado River have been wetter than most of the 400 years studied with tree rings.

Tarboton (1995) analyzed similar series to design "worst- case" drought scenarios for the Colorado River. The 1943-1964 period was chosen as the worst drought period in the historic record of flow measurements. The 1579-1600 period was selected as a realistic worst-case severe sustained drought based on "simulated" reservoir-storage deficits of over 50 million acre-feet that would have occurred.

Annual flows 1570-1610 graph

Figure 5A. Annual flows, 1570-1610, illustrating three hydrologic droughts in quick succession between 1579 and 1600. (Modified from Tarboton,1995) [MAF = millions of acre-feet per year.

Artificially severe droupht graph

Figure 5B. Definition of the artificially severe sustained drought for the Colorado River basin. [MAF = millions of acre-feet per year. (Modified from Tarboton,1995)

Finally, to develop a truly worst-case scenario, Tarboton (1995) sorted the annual flows from the 1579-1600 period to create a drought with the same annual flows as during the 1579-1600 period (as shown in Figure 5A) but with each year worse than the one before (Figure 5B). Under these artifically harsh conditions, the lowest flows are clustered at the end of the drought when reservoirs would already have been low or dry; in reality, the driest years occurred around 1585, in the first quarter of the drought. This artificially harsh scenario of 19 progressively drier years, together with a period of equal length of more normal weather (based on flows estimated from tree rings for the period 1601-1620), forms the Severe Sustained Drought to which the
hydrologic, environmental, and economic studies subjected the Colorado River water-resources systems in various simulations.

Using the tree-ring series and a variety of statistical models of the variability of Colorado River flows, droughts like the worst-historical drought (1943-1964) are expected about once every 50-100 years, on average. The realistic worst-case severe sustained drought (1579- 1600) might recur once in 400-700 years while the artificial (sorted) drought scenario might be expected about once in 2,000 to 10,000 years, on average.

It has been conjectured that greenhouse warming might lead to an enhancement of the global hydrologic cycle (e.g., Graham, 1995) and that enhancement, in turn, might result in greater hydroclimatic variability. If so, then the probabilities of droughts such as those identified by Meko and others (1995) and by Tarboton (1995) could change and the expected time between such severe sustained droughts might be much shorter or much longer. Notably, Lamb (1995) describes quite varied long-term changes in precipitation over the western United States during the 1100s to 1300s when general cooling prevailed (leading up to the Little Ice Age); some areas experienced more drought during this cooler period and others were wetter. Thus there is no certainty as to whether climate change would bring more or less drought to the Southwest.

Other References Cited

Graham, N.E., 1995. Simulation of recent global temperature trends. Science, v. 267, p. 666-671.

Lamb, H.H., 1995. Climate, history, and the modern world (2nd ed.). New York, Routledge, 433 p.

Back to Motivation section

What kind of hydrologic, environmental, and economic impacts would result?

What management alternatives might help?

Link to USGS home page

Accessibility FOIA Privacy Policies and Notices

Take Pride in America home page. logo U.S. Department of the Interior | U.S. Geological Survey
Page Maintainer: Randy Schumann
Page Last Modified: Fri 9-Dec-2016 14:28:26 MST