February 7, 2005

No proof is sufficient

Filed under: Climate Models, Droughts

In the face of evidence resulting from their own research, some scientists refuse to abandon their preconceived idea: that computer-based climate models reliably forecast future climate.

We’ve come across research published in the peer-reviewed journal Geophysical Research Letters, the title of which pushes the limits of scientific acceptability: “Forty-five years of observed moisture in the Ukraine: No summer desiccation (yet).” It appears we have now entered a phase of the global climate change debate wherein scientists begin to trumpet their personal bias even if it runs contrary to evidence compiled by the scientific entity they represent or, even more astounding, if it runs counter to research results they themselves produce!

An example of the first type is illustrated by NOAA hurricane researcher Christopher Landsea’s recent letter resigning from participation in the UN Intergovernmental Panel on Climate Change. Landsea described how Kevin Trenberth participated in a press conference last fall in which he expressed a personal view that global warming was likely responsible for the high level of hurricane activity along the U.S. coastline last year.

Landsea was dismayed that Trenberth — who serves as a the convening author of the chapter in the upcoming IPCC Fourth Assessment Report dealing with observed climate changes (including hurricanes) — expressed an opinion in marked contrast with the IPCC’s findings in its Third Assessment Report. In 2001, the IPCC concluded there is no indication global warming is affecting hurricane strength or frequency. Landsea concludes (correctly, we think) that if the lead author’s expressed opinion runs counter to observed data, then there is little chance the observations will receive fair treatment in the upcoming 2007 Fourth Assessment Report.

An example of the second type of bias also can be found in Geophysical Research Letters. It appears under that headline “Forty-five years of observed moisture in the Ukraine: No summer desiccation (yet)” atop research published by Rutgers University Environmental Sciences professor Alan Robock and colleagues.

Appending a parenthetical “yet” to the end of the paper’s title is wholly unjustified by its content and represents nothing more than an editorial effort to prejudice the reader. Someone casually browsing article titles in the table of contents can be expected to conclude that even though observations don’t show anything to be amiss, it is only a matter of time before they do.

While this may reflect the authors’ personal opinion, there is absolutely nothing in the paper that lends credible support to that opinion. As such, the peer-reviewers of this paper failed in their role and in their responsibility to the general scientific community by allowing a paper’s title to bear little relevance to its content.

The paper analyzes what is described as the “longest data set of observed soil moisture available in the world.” Summer soil moisture data from the topmost meter of soil (that which is available for use by plant life) were collected from a network of 141 soil moisture observing stations in the agricultural region of the Ukraine beginning in 1958. The data were aggregated to provide a history of observed soil moisture for the region from 1958 to 2002. With that history in hand, the authors set out to compare it with climate model simulations of the same thing. The authors hypothesize:

Most global climate model simulations of the future, when forced with increasing greenhouse gases and anthropogenic aerosols, predict summer desiccation in the midlatitudes of the Northern Hemisphere. This predicted soil moisture reduction, the product of increased evaporative demand with higher temperatures overwhelming any increased precipitation, is one of the gravest threats of global warming, potentially having large impacts on our food supply. We use our extended data set here to evaluate these model simulations.

We provide the observed soil moisture history from the world’s longest data set of observed soil moisture as Figure 1. We note an upward trend (signifying more moisture) over the course of the 45-year period of record. Most of the rise occurs in the first half of the record. This trend of increased soil moisture is present despite a small upward trend in temperature and a downward trend in summer precipitation. The authors believe this may be driven by a downward trend in evaporation cause by increased tropospheric aerosols, but this is merely their theoretical explanation of what appears to be going on.

Ukraine Soil Moisture

Figure 1. Time history of the summer soil moisture content averaged across soil moisture observing stations throughout the Ukraine, 1958-2002. Two independent determinations were made, one using data collected from fields growing winter cereals (red line), the other from fields growing spring cereals (blue line). (Source: Robock et al., 2005)

These research results don’t appear to square with climate models, so Robock and colleagues collected soil moisture simulations for the Ukraine region using several climate models. The models were run with observed changes in atmospheric composition for the past century or so and then coupled with one of the most extreme future “forcing” scenarios (SRES A2) set forth by the IPCC for the next 100 years.

In yet another disservice to readers, the authors describe the A2 scenario as “essentially business as usual, with increasing greenhouse gases and sulfate aerosols.” The term “business as usual” apparently means something different to the authors than it does to the rest of us. Total climate forcing prescribed by SRES scenario A2 during the 21st century is 6.74 W/m2. This contrasts with NASA climatologist Jim Hansen’s recent report that “the growth rate of climate forcing by measured greenhouse gases peaked near 1980 at almost 5 W/m2 per century” and “this growth rate has since declined to ~3 W/m2 per century.”

In other words Robock’s “business-as-usual” scenario represents a growth rate that is more than twice the observed value. In fact, it is a rate of growth that has never occurred before! How can this be business as usual?

Figure 2 shows the climate model simulations of soil moisture with the observations overlain atop them. None of the models bears any resemblance to the observations. According to the authors, “All of the runs clearly show a weak upward soil moisture trend in the Ukraine before 1960 and a strong downward trend after 1960. The GFDL-R30 and the CCCma model runs simulate less summer desiccation than the HADCM3 and CCSR/NIES models simulate more. However, none of the simulations is well matched with the observations.”

Ukraine soil moisture modeled

Figure 2. Comparison between climate model simulations (colored lines) and actual observations (black line) of Ukrainian summer soil moisture. The models do a poor job at capturing the observed behavior. (Source: Robock et al., 2005)

The authors go on to add, “This is not unexpected, as there should be a random component to interannual precipitation variations, especially for a region as small as the one studied here.” Might we ask, then, if the authors believed the models couldn’t be expected to simulate observations why was comparison of models and observations one of the stated purposes of their paper? How can they conclude, “The new 45-year data set of Ukrainian soil moisture has proven useful for evaluation of climate model simulations, as illustrated here”? Instead, this sounds like an a posteriori cover-up of the fact that the models performed poorly when compared with actual observations, apparently to the chagrin of the authors.

Nowhere in their research is there an indication that summer desiccation is imminent unless one carelessly includes what can be expected from climate models run under an extreme forcing scenario, results that illustrate how climate simulations bear little resemblance to actual climate observations. Clearly, from the results of the analyses presented in their own paper, there is no scientific justification for the word “yet” to be appended to the title of their paper. Its presence exposes the authors’ disappointment at their failure to generate support for model-based predictions of climate doom and gloom.

Instead of bemoaning what observed reality indicates, the authors would better serve their peers were they to stress the inaccuracies of climate modeling, especially related to prediction of regional-scale impacts. We have little hope their peers will suggest that they remedy their error (yet).


Hansen, J.E., and Sato, M., 2001. Trends of measured climate forcing agents. Proceedings of the National Academy of Sciences, 98, 14778-14783.

Robock, A., Mu, M., Vinnikov, K., Trofimova, I. V., Adamenko, T. I., 2004. Forty-five years of observed soil moisture in the Ukraine: No summer desiccation (yet). Geophysical Research Letters, 32, L03401, doi:10.1029/2004GL021914.

No Comments

No comments yet.

RSS feed for comments on this post. TrackBack URI

Sorry, the comment form is closed at this time.

Powered by WordPress