The report, "When will Lake Mead go dry?" says the Nevada reservoir and its upstream neighbor, Lake Powell, could drop so low gravity won't be able to move water downstream.

Researchers Tim Barnett, a marine geophysicist, and David Pierce of Scripps' Climate, Atmospheric Science and Physical Oceanography division - which is based at the University of California, San Diego - conclude there is a 50-50 chance of Lake Mead being a "dead pool" in 13 years if the system of water allocation now in place remains unchanged.

The study, which the authors say was conservative in its assumptions, disputes the river flow model the U.S. Bureau of Reclamation used to develop a landmark agreement signed in November. The interim agreement in effect through 2026 outlines how the seven Colorado River Basin states will share the pain of water shortages, and is the first major change to the Colorado River Compact since its 1922 establishment.

"The mitigation strategies [the states] have for water shortages is not going to work. We can see that right now," Barnett said Thursday.

The Bureau in 1999 began the environmental impact study that underlies the 2007 water-sharing agreement. In that time, the river basin has experienced the worst drought in 100 years of recorded history, and Lake Powell and Lake Mead have gone from being nearly full to just more than half-full.

The agreement lays out how the upper basin states - Utah, Colorado, Wyoming and New Mexico, which have most of the snowpack - will respond to demand from California, Arizona and Nevada, the lower basin states, which have many millions more residents and senior water rights. Under the agreement, Lake Powell and Lake Mead water levels will remain roughly equal.

The Bureau's final EIS implicitly acknowledged that the Colorado River Compact is based on estimates from unusually wet years and it assumed ongoing shortages as the drought continues.

The Scripps study states explicitly the region is in trouble.

Barnett said he and Pierce were surprised at their findings, which in essence measured the amount of water going into the reservoirs versus the water going out to the lower basin states.

The dead-pool scenario is likely if current Colorado River management remains static - a big if, Barnett acknowledged.

"None of us believe this is the final answer," he said. "It tells us there is a major problem coming."

Similar studies have emerged during the past 20 years. In August, the U.S. Geological Survey took a broader look than the Scripps study at how climate change has set up a self-perpetuating chain of events that ensure the river will continue to shrink.

The Bureau of Reclamation is aware of the science, and representatives defended the water-sharing EIS.

"As of today, if nothing changes, I could agree with [the Scripps study]. But what we know about this river basin is hydrology is highly variable," Bureau hydrologist Paul Davidson said Wednesday.

What's most important, he said, is the reservoir system is working as planned.

"It's seen us through this drought and now it's half empty," Davidson said. "If the drought held for another eight years, then yeah, we have a problem with the water budget. But we don't expect that to happen."

Hydrologists predict Lake Powell will rise 50 feet this year if snowmelt occurs according to projections showing runoff will be 128 percent to 130 percent of normal, he said.