How reliable are drought predictions? Study finds flaw in popular tool.

The inappropriate use of a metric for measuring drought has led some studies to overstate a trend over the past several decades toward more frequent, intense droughts, according to a new analysis. The trend has been attributed in no small part to global warming.

The authors of the paper caution that the metric's continued use is likely to overstate future effects as well, making it more difficult for water resource managers and others to get reasonable projections of future drought hazards as the climate warms.

The study does not make those projections. Nor does it focus on assessing recent or current droughts, such as this year's drought in the US – the most extensive and severe in at least 25 years, according to the US Department of Agriculture. Instead, the study is a cautionary tale about using the right tool for the right job.

More broadly, and consistent with a wide range of other research, "our work on 21st-century drought [shows] that there are regions of the world that are going to experience more drought" as global average temperatures continue to rise during this century, says Eric Wood, a professor of civil and environmental engineering at Princeton University in Princeton, N.J., and a co-author of the study, which appears in Thursday's issue of the journal Nature.

But the key questions for resource managers and their political overseers in assessing future drought risk, he continues, center on the extent, intensity, and frequency of drought. Some drought projections, which use the approach the new study challenges, yield results where "you'd think it will never rain again," Dr. Wood says.

At issue is a measure for gauging drought, known as the Palmer Drought Severity Index (PDSI). Its most publicly recognizable form is found in drought maps based on the index – with their yellow, tan, red, and brown splotches signifying the severity of the drought in different parts of the country. These maps have appeared on countless newspaper and web pages over the past two years of widespread drought in the United States.

The problem, according to the research team, led by Princeton's Justin Sheffield, lies in the index's original design. It was developed to track drought conditions from one week to the next, or one season to the next, with a particular focus on croplands in the center of the United States. It was not designed to track trends globally over decades or centuries. Yet some researchers continue to used it for that purpose, Dr. Wood says. The reasons range from the index's relative simplicity to a lack of confidence in the quality of more-recently available data that allow for more-sophisticated calculations.

It's an issue not lost on many in the wider community of researchers focusing on drought and global warming.

"We've known for quite a long time that the PDSI calculation is prone to problems dealing with climate change," says Richard Seager, a researcher who focuses on drought and climate change at Columbia University's Lamont-Doherty Earth Observatory, located in Palisades, N.Y. "Rising temperatures drive it haywire."

One example of the disconnect comes from the last major set of climate reports from the Intergovernmental Panel on Climate Change (IPCC), published in 2007. One study that factored heavily in the drought discussion in the volume that focused on the science of climate change showed widespread drying since 1970, with the extent of very dry areas around the world more than doubling – a conclusion that relied on the Palmer Drought Severity Index.

But in last year's volume on weather extremes, which builds on more-recent research, the IPCC authors noted that "there are still large uncertainties" in discerning global trends in observed droughts. They placed "medium confidence" that some of regions of the world were experiencing longer, stronger droughts. For other other regions however, including central North America, droughts were trending toward less frequent, less intense, or shorter.

In some ways the study represents a window on the sociology of science. Many researchers have concluded that the index has little or no value as a climate-change projection tool. Yet some still cling to it.

"There's an awful lot of inertia" in science, says Dr. Seager, who is not a member of the team that produced the PSDI analysis. "We often sit around and wonder: Why are people still doing this?"

The relatively recent availability of the right kind of data and doubts about its quality are factors. But so is science's general tendency to move slowly, he adds.

Dr. Sheffield and colleagues traced the problem with PDSI to the way it handles evaporation rates. Typically, the PDSI has used precipitation and temperature as the key drivers governing evaporation. But researchers have long known that other factors affect evaporation rates as well. These include relative humidity, the amount of the sun's energy reaching the surface, and wind speed.

When the team calculated the PDSI using the temperature-focused approach for evaporation and used the results to identify drought trends, the results indicated that between 1950 and 2008, dryness increased for 98 percent of the globe's land area. When the team used the more sophisticated approach to calculating evaporation, the picture was more mixed: 52 percent of the land area saw increased drying, while 42 percent saw a decrease.

Overall, both approaches show a drying trend, the team reports, but the increase in areas covered by drought using the more complete treatment of evaporation is seven times smaller than the more-simple approach. Focusing in the last decade, the more-complete calculation suggests that the global area covered by drought has increased slightly, although the increase isn't statistically significant, while soil moisture averaged over the globe essentially has remained unchanged.

While the study shows the difference the treatment of evaporation can make in the outcome, the more complete treatment doesn't necessarily redeem the use of the PDSI.

The Palmer Drought Severity Index "should not be used for climate-change projections," Seager says flatly.

Looking forward to the next set of major IPCC reports on climate, due out over 12 months starting next September, Wood says, "I think you will see work coming out of the community that will use appropriate methods" for projecting future drought patterns.