In the early 1990s, the U.S. Geological Survey and several other government agencies funded a little-noticed study of the effect of a major drought on the Colorado River. Researchers were particularly interested in its impacts on Lakes Powell and Mead, which store about four years’ worth of the river’s average annual flow and serve as the system’s backup batteries in drought.

The researchers, working with Boulder, Colo., water-resources engineer Ben Harding, wanted a worst-case scenario. To find it, they turned to the growing record of long-term climate data derived from tree-ring studies (HCN, 1/24/05: Written in the Rings). In the tree rings, they found a drought in the 1500s that lasted for 38 years. Its first years were the worst, followed by progressive improvement. To turn the drought into a real beast, Harding flipped it around so that each subsequent year got worse. Then he built a computer model of the river based on the Bureau of Reclamation’s own model, and programmed in the imaginary "severe sustained drought."

In Harding’s drought disaster scenario, the water level in Lake Powell nose-dived like a Disneyland log-flume ride until, 18 years into the drought, it bottomed out. The reservoir had reached "dead storage," the point at which no more stored water could be released downstream out of Glen Canyon Dam. Lake Powell was, for all intents and purposes, empty — and would stay that way for eight years.

The drought used in the model was without real-world precedent. As Harding says, "We’d come up with this drought that was really perverse. It was just the worst kind of drought you could possibly contemplate." But what’s happening on the river now could be even more severe.

A year ago, curious about how the current drought compared with the monster he’d created in his computer, Harding dusted off his 10-year-old graph and superimposed the current drought on it.

"I was really surprised," he says. The line traced by the current drought plunged downward even faster than that of the simulated drought, thanks not only to hydrologic conditions, but also to intense water demand in Arizona, California and Nevada. "The bottom line is that, going into this drought, we’re in much worse shape, because our reservoirs are lower and our water uses in the Lower Basin are higher (than was modeled)."

River flows during the past five years have been lower than at any comparable period in the past century — and clearly show that there’s far less water in the river than originally thought. When negotiators signed the Colorado River Compact in 1922, they believed that the long-term average, measured at Lee’s Ferry, the dividing line between the Upper and Lower Basins, was 16.2 million acre-feet. That number has proven to be wildly optimistic. Subsequent reconstructions of the river’s flow over the past approximately 500 years, extrapolated primarily from tree-ring data, have shown a long-term average of only 13.5 million acre-feet. The average annual flow at Lee’s Ferry during the past five years has been closer to 9.9 million acre-feet.

The Colorado River’s dams and reservoirs have helped buffer the region against the drought. The Central Arizona Project has estimated that, without Lake Powell, the Interior Department might have declared an official shortage on the river in 2002. The shortage could have become big enough to affect California and Nevada as well as Arizona by 2003. But even with Lake Powell, an official shortage could be just a couple of years away.

Could a wet winter break the cycle? Even if this proves to be what Harding calls "a bang-up year," it hardly means the drought is over. "If you go back and look at the original drought (in the 1500s)," he says, "it’s got a bunch of wet years in it. But it just had this relentless aspect to it that kept it dry for a long, long time."

Harding’s comparison of the current and severe sustained droughts can be viewed online at www.hydrosphere.com/publications/SSDRedux.htm