Sheffield vs. Dai on Drought Changes

A new paper in Nature by Sheffield et al. (2012) has been making waves, because it argues that there has been little change in global drought over the past 60 years, and that the Palmer Drought Severity Index (PDSI), used for example by Dai (2010), overestimates future drought.  Climate Progress has an interesting response, including comments by Trenberth and Dai.  Paleoclimatologist Kevin Anchukaitis also has a very informative post on the subject at Strange Weather, as does John Nielsen-Gammon.  Below is a repost of another very good post on the subject, by Carbon Brief.

Major droughts in the United States and Russia this summer highlighted the risks climate extremes pose to everyday life, and prompted discussion about links between drought and climate change.

Past research supports a link between climate change and drought - suggesting that globally, the area affected by drought has increased in recent decades as global temperatures have risen.

But a new study in Nature challenges this link, suggesting that the relationship might not be that simple, and that older models may have overestimated the change in drought over the last 60 years. In fact, newer models which take a more detailed look at the how droughts occur suggest the world area in drought may not have changed much at all.

We take a look at the new research and why overestimating the past might not affect predictions of how drought may respond to climate change in the future.

What's changed?

The Intergovernmental Panel on Climate Change (IPCC) concluded in its Fourth Assessment Report, published in 2007, that "Globally, the area affected by drought has likely increased since the 1970s".

In its 2012 Special Report on Extreme Events (SREX), the IPCC was more cautious about past trends, concluding "there are large uncertainties regarding global-scale trends in drought".

According to the new study in Nature, those uncertainties are partly down to the way simple drought models simulate water evaporation. Earlier drought models worked out how much water was evaporating based solely on temperatures. But a number of other factors affect evaporation in real life, like wind speed, vegetation cover and aerosols affecting sunlight.

Newer models include more of these factors, giving a more accurate picture of drought. Using the newer models, the study found there's been no significant change in the area affected by drought between 1950 and 2008. This led the researchers to conclude that studies using the earlier models overestimated the global area affected by drought.

Justin Sheffield, lead author on the study, told Carbon Brief:

"Our findings confirm what the SREX report recognized: that at the time of its writing (6 months ago) there was uncertainty about global trends in drought and that the previous IPCC AR4 conclusion [...] was likely overestimated."

Many uncertainties still exist, even with the new models - the meteorological data sets used to calculate levels of drought are incomplete, and the method for using these models differs from one researcher to the next. That makes quantifying change hard, and more studies will be needed to add certainty to these findings.

But this isn't the only research finding little change in droughts over past decades.

Is "little change" new?

Sheffield and his colleagues aren't the first to find little change in the global area affected by droughts over past decades. In 2006, researchers from the Climatic Research Unit (CRU), based at the University of East Anglia, found similar trends even when they used their own versions of the older, simpler models. Their research on European droughts between 1901 and 2002 concluded:

"evidence for widespread and unusual drying in European regions over the last few decades is not supported by the [models]"

Similarly, the CRU researchers found no statistically significant trend in drought trends over North America during the same time period. Using newer models to simulate drought, they've also looked at global drought trends. The findings, yet to be published, suggest drought has not become unusually strong or widespread in recent decades, agreeing with their earlier work.

Although they differ on how much old and new models agree, the research from both CRU and from Sheffield et al. suggest that when averaged out over the globe, drought has changed very little in recent decades. But that's not to say it won't in the future.

Does this affect future predictions of drought?

The finding that some models may have overestimated past drought is unlikely to affect predictions about future droughts. That's because working out how climate change might affect drought in the future is done using different - climate - models. These climate models are more sophisticated and operate in a different way than the simple models this study looked at to analyse past drought.

Lead author Justin Sheffield explained to Carbon Brief:

"[Climate models] respond and feedback with the projected changes in precipitation, temperature, radiative forcing etc. in a more realistic manner."

The authors of the new paper specifically warn that trends identified by the models used to look at past drought shouldn't be extrapolated into the future, not least because the assumptions these models make based on temperature could overestimate future levels of drought.

Climate models however, as the authors explain, are much better at recreating the links between temperature, precipitation and evaporation, so offer a better way to try to estimate future changes. 

What this new study does do is inform the way climate projections are interpreted. At a time when links between climate change and extreme events are under discussion, this research is a reminder that drought is affected by a number of factors. And it's the combination of these factors, rather than simply temperature rise, that will determine how drought changes in the future.

Posted by dana1981 on Friday, 23 November, 2012

Creative Commons License The Skeptical Science website by Skeptical Science is licensed under a Creative Commons Attribution 3.0 Unported License.