Flawed USGS Study Still Links Southwestern Drying to Increasing Carbon Dioxide Pollution and Climate Change

A new U.S. Geological Survey analysis finds that, as climate scientists have been predicting for decades, the Southwestern U.S. is drying in part because of rising levels of carbon dioxide:

The decrease of floods in the southwestern region is consistent with other research findings that this region has been getting drier and experienced less precipitation as a likely result of climate change.

The study, “Has the magnitude of floods across the USA changed with global CO2 levels?” appearing in Hydrological Sciences Journal, however, relies on dubious and “absurd” assumptions, according to a number of climate scientists I spoke with.  Amazingly, the lead author seems to lack an understanding of core issues germane to his analysis, as we’ll see.

The finding about SW drying that I’ve focused on isn’t the main spin the USGS and media have given the study.  The USGS focused on what they claim is the lack of a “significant relationship between carbon dioxide (CO2) in the atmosphere and the size of floods over the last 100 years” in the three other regions they rather arbitrarily divide the country into — northeast, southeast, and northwest.

Interestingly the NE “stretches from the middle of the Dakotas and Nebraska all the way east to the New York and New England area,” and it “shows a tendency towards increases in flooding over this period.”  But in the USGS analysis, the tendency isn’t statistically significant.  I’ll address in a later post why that isn’t particularly surprising given how the USGS does its analysis.

But it’s worth noting that Kevin Trenberth, senior scientist at the National Center for Atmospheric Research, and a leading expert on the impact of global warming on extreme weather and precipitation, thinks it is “absurd” that USGS was looking for a relationship between global CO2 levels and flooding.  Other climate scientists I spoke to expressed similar reservations about this.  One called it, “simply wrong.”  Why?

CO2 levels have been rising pretty steadily for many decades, but flooding is primarily linked to warming (through the greater water vapor in the atmosphere and things like early snowmelt).

Many studies and all global climate models have made clear that global temperatures don’t rise in lockstep with global mean carbon dioxide concentrations (GMCO2)  — thanks to aerosols, volcanoes, and the El Niño Southern Oscillation (ENSO) and so on.  And, of course, the temperatures relevant to U.S. flooding don’t rise in lockstep with global temperatures thanks to ENSO and the like.  So finding  correlations between flooding and GMCO2 is not terribly dispositive.

Yet even with that big flaw and many others, the study still found one trend that is similar to what the IPCC models have predicted:


One region, the southwest, showed a statistically significant negative relationship between GMCO2 and flood magnitudes….

The similarity is strong in terms of the trend towards drying conditions in the Rocky Mountains and arid southwest.

In ClimateWire (subs. req’d), the lead author, Robert Hirsch, is quoted rather bizarrely on this statistically significant negative correlation in the SW:

But even that shred of correlation could be dismissed. “It has more to do with a general decrease in precipitation that has been observed by many people,” said Hirsch.

That is a head-exploding quote.  There have been several major studies published in which climate scientists have projected a decrease in precipitation in the U.S. Southwest thanks to rising CO2 levels and global warming — one dating back to 1990.  I have a comment piece in Nature on this very subject coming out Wednesday.  You can see some of the literature here: “USGS on Dust-Bowlification: Drier conditions projected to accelerate dust storms in the U.S. Southwest.”

The serious hydrological changes and impacts known to have occurred in both historic and prehistoric times over North America reflect large-scale changes in the climate system that can develop in a matter of years and, in the case of the more severe past megadroughts, persist for decades. Such hydrological changes fit the definition of abrupt change because they occur faster than the time scales needed for human and natural systems to adapt, leading to substantial disruptions in those systems. In the Southwest, for example, the models project a permanent drying by the mid-21st century that reaches the level of aridity seen in historical droughts, and a quarter of the projections may reach this level of aridity much earlier.

Again, the USGS release itself statesThe decrease of floods in the southwestern region is consistent with other research findings that this region has been getting drier and experienced less precipitation as a likely result of climate change.”

Apparently Hirsch doesn’t even read his own agency’s press releases!  The USGS paper doesn’t cite a single one of the studies above on SW drying.  The entire analysis is weak, and Hirsch’s quote suggests a severe lack of understanding of core issues germane to his analysis.

The ClimateWire piece headline was “Flood sizes not rising along with CO2 levels, ‘at odds’ with IPCC — study.”  As I’ll discuss in a later post, that headline may reflect the study, but the study itself is just too flawed to warrant such a strong conclusion.  UPI did better with its headline, “USGS finds complex link to floods and CO2.”

Want updates from Climate Progress? You can follow us on Twitter, like us on Facebook, or subscribe to our RSS Feed.

15 Responses to Flawed USGS Study Still Links Southwestern Drying to Increasing Carbon Dioxide Pollution and Climate Change

  1. Mossy says:

    I’d like to know more about the authors and funding of this study. We’ve often found geologists to be among the denier camp — and I know this is a vast generality — but I think that it stems from the fact that often geolgists make a living by exploring drilling sites.

    “The Hill’s E2 Wire” reports on the study with this lead sentence: “A new study conducted by federal scientists found no evidence that climate change has caused more severe flooding in the United States during the last century.”

    That’s a “put your head in a vise” statement, ripe for the deniers’ plucking.

  2. Hank says:

    Is “Dust-Bowlification” an actual word??

  3. Joe Romm says:

    Is now.

  4. Colorado Bob says:

    I read the USGS press release . I noted the study ends with 2008. Just from my observations it’s been the last 3 years that , water machine has begun to really take off.

    Flood Defenses Are Overrun in Bangkok

  5. Colorado Bob says:

    Thailand’s worst floods in more than a half century may have wiped out as much as 14 percent of paddy fields in the world’s biggest rice exporter, potentially erasing the predicted global glut.

    The Thai export price, a global benchmark, may climb 20 percent to $750 a metric ton by December, according to Sumeth Laomoraphorn, president of C.P. Intertrade Co., the country’s largest seller of packaged rice. Tropical storms inundated 62 of 77 provinces, destroying 1.4 million hectares (3.5 million acres) and as much as 7 million tons of crops, the government says. That equals 4.6 million tons of milled grain, 1 million more than the surplus expected by the U.S. Department of Agriculture.

  6. Nat says:

    Robert Hirsch is rightly concerned by the issue of Peak Oil, so much so in fact that he thinks we shouldn’t let issues like Climate Change get in the way of the exploitation of unconventional fuels such as tar sands… As serious as Peak Oil is, Climate Change is even more fundamentally serious in my opinion.

  7. Lou Grinzo says:

    Besides being accurate easily understood, Dust-Bowlification also has the benefit of being more socially acceptable than FUBAR-ification.

  8. Nat says:

    oops.. wrong Robert Hirsch… was thinking of this one:

  9. Mossy says:

    Thanks for the information on one of the authors, Nat.

  10. Lou Grinzo says:

    Nat mentions Hirsch’s peak oil background, which is indeed very important. In fact, in peak oil circles one of the more important contributions is the 2005 report “Peaking of World Oil Production: Impacts, Mitigation, and Risk Management”, a.k.a. “The Hirsch Report” [].

    A key point to keep in mind is that the peak oil conversations ON AVERAGE are dominated by geologists, both real and of the armchair variety, which means they have a deep aversion to anything that comes out of an economist’s mouth. This is particularly true when economists (including me) start talking about how we can and will (attempt to) adjust to higher oil prices. Even if you’re not a cornucopian (and I’m definitely not), any suggestion you make that a worldwide peak and decline in oil production can be handled even somewhat gracefully is rejected by the wanna be geologists and doomers. (I’m not saying that Hirsch or anyone contributing to this USGS report are in that category, just to be clear.)

    I think it’s fair to say that many, if not most, of the PO crowd think PO is worse than CC because of its immediacy and its potential to devastate economies like the US’ that are so highly dependent on oil. As the PO camp will point out, we’ve been on more or less the infamous “undulating plateau” of oil production for several years, so we might be at peak oil right now, with the only mysteries being how soon and how quickly the production decline will happen.

    The PO vs. CC debate is like arguing which is worse, getting very sick right away with an ailment that will last for a few years to a decade (PO) or continuing to poison ourselves with a 100% certainty of having some symptoms now that will worsen until we’re much sicker than we would be in the first scenario in a few decades. Depending on your assumptions, you can make a reasonable argument for either case; I came into this sustainability discussion via the PO door, but I’ve been convinced for some time that CC is the much greater threat.

    Of course, it’s not an either-or situation. We’ll have to deal with both illnesses at once, since there’s no evidence we’ll get serious about reducing our oil dependency or our CO2 emissions by enough to avoid either problem.

  11. Mike says:

    The Moscow Warming Hole
    Filed under:

    * Climate Science

    — stefan @ 26 October 2011

    This week, PNAS published our paper Increase of Extreme Events in a Warming World, which analyses how many new record events you expect to see in a time series with a trend. It does that with analytical solutions for linear trends and Monte Carlo simulations for nonlinear trends.

  12. Joe Romm says:

    Yes, a different guy.

  13. mark t says:

    Hirsch is a statistician, not a geologist (full disclosure, I am both a hydrologist and geologist). He has done some very good hydrologic statistical work in the past, and is co-author of an outstanding statistical text published by the USGS. However, good statistics do not produce good answers if you test the wrong hypothesis. Clearly, the hydrologic cycle is affected by temperature, and temperature is a regional variable. CO2 is a global variable, not a regional one. The effects of changes in annual and seasonal temperatures on the hydrologic cycle must be examined regionally, but the regions must make hydrologic sense. A contour map of annual rainfall in the US shows a dense clot of contours running from Minnesota to Texas. Areas to the west are much drier, areas to the east are much wetter. With an increase in temperature there is an increase in water vapor content, but relative humidity often stays about the same. This means dry regions tend to get drier as evaporation is greater, and wet areas tend to get wetter, as there is more water vapor to condense. Including areas west of the N-S rainfall boundary with areas east of the boundary includes negatively and positively correlated regions in one region, invalidating the test of statistical significance. This paper probably did not go through the usual rigorous USGS review process. The statistical methods are fine, but I agree with Trenberth, the questions tested are ‘absurd’. The paper is an embarrassment to the Survey.

  14. Joan Savage says:

    I question their method of selection of data. They left out data from unregulated streams (e.g. that leaves out data from much of the Mississippi Valley) and from urban areas defined as more than 150 persons per square kilometer (e.g. that leaves out many small watersheds, such as along the Gulf coast and Atlantic coast). They therefore did not integrate total volume of precipitation over time or area. They excluded data from streams that with 85 years of data had largely gone dry in the last ten years of the study (“must have had five annual peak discharges between 1999 and 2008” (Appendix, Site and peak selection).

    With the huge data set available to the USGS, more meaningful observations might be made, such as integrating total stream flow from watersheds, and with that, developing ‘flood’ equivalencies for regulated and urban areas.

    That level of hydrological analysis takes the kind of statistical skills that are used interpolate surface temperature.

    Joe Romm already drew attention to the peculiar approach of dividing the 48 states into quadrants. Why not watersheds? Or historic climatological regions?

    This is not a complete analysis of the methodology or of data selection in particular. Just some observations.

  15. You’re right – It is a “vast generality” and you need to know what group of geologists are talking. Although most in the extractive industries are deniers, the majority of us in academia, federal (like the USGS), and state agencies are more clear headed about this.