launched weather balloorigi,-This is in stark contrast to the observed surface air temperature rise of 0.6° 0.2°C that has occurred over the entire twentieth century. A blue-ribbon panel convened to address this apparent discrepancy concluded that the temperature of the lower atmosphere might have remained relatively constant while an increase in near surface air temperature was observed. Some have argued that the surface warming is a delayed response to warming that had earlier occurred in the troposphere, although the abrupt warming of the troposphere is not consistent with expected scenarios of anthropogenic warming. The National Academy of Sciences (NAS) concluded that the difference between surface air temperatures and those of the troposphere was real but inconsistent with anthropogenic warming scenarios. In particular, the NAS only considered whether the satellite and surface records could both be correct and yet contradictory; they never addressed the issue of whether the surface records could, in fact, be biased. Another problem in tying the observed increases in air temperature to an anthropogenic cause is timing. Most of the warming in the observed record occurred during two periods: 1910 to 1945 and 1970 to present. Much of the warming actually predates the rise in anthropogenic trace gas emissions, which makes it difficult to ascribe anthropogenic causes to the entire record. Indeed, we know that our observed record began in the late 1800's when air temperature measurements were sparse and more prone to bias. This timing also coincides with the demise of the Little Ice Age-a period of cooler-than-normal conditions that lasted from the middle portion of the last millennium to about the mid-1800's. Thus, it is unclear how much of the observed warming should be attributed to anthropogenic increases in atmospheric trace gases and how much of it is simply natural variability or measurement bias. MODELING THE COMPLEX CLIMATIC SYSTEM IS AN EXTREMELY DIFFICULT TASK In theory, therefore, climate models should be our best ability to study climate change. With models, we are not constrained by biased and limited observing systems or by contamination by other signals; but rather, we can alter the simulated climate and see "what if' while holding everything else constant. Such models, however, are predicated on their ability to replicate the real climate-after all, if climate models cannot replicate what we observe today, how can their prognostications of climate change possibly be expected to be transferable to the real world? Although I am not a climate modeler, much of my research has focused on comparing observations with climate model simulations of present-day conditions. Thus, I am very familiar with what climate models can and cannot do. I am dismayed by the fact that much of the rather limited success in simulating average conditions by most climate models is achieved at the expense of changing some parameters to highly unrealistic values. For example, some models drastically change the energy coming from the sun to levels that are well beyond those that solar physicists have observed. Many models employ what are called "flux adjustments", which can only be described as finagling factors to make the average, present-day surface air temperatures look reasonable. One has to question why such overt deviations from reality are necessary if, in fact, the models are able to realistically represent our climate system. In defense of climate modelers, I will say that they have a very difficult and daunting task. The climate system is extremely complex. Clouds, land surface processes, the cryosphere (ice and snow), precipitation forming mechanisms, the biosphere, and atmospheric circulation, just to name a few, are complex components of the global climate system that are not well understood or modeled appropriately at the scale employed by general circulation models. In essence, the climate change response can be directly affected by our parameterizations of many of these components. For example, an important question that now is being asked is "Why is the warming exhibited by transient climate models not being seen in the observed record?" There has been much discussion on the impacts of aerosols, black soot, high altitude clouds, and other so-called "wild cards" in the climate system-are they masking the climate change signal or should they be adding to it? How climate modelers treat these unknown processes in their models can affect dramatically the model simulations. Indeed, there are likely additional issues that we have not yet encountered. CLIMATE MODELS CANNOT REPRODUCE A KEY CLIMATIC VARIABLE: PRECIPITATION Despite these issues, do climate models well represent the Earth's climate? On three separate occasions in 1990, 1996, and again in 2000-I have reviewed the ability of state-of-the-art climate models to simulate regional-scale precipitation. In general, the models poorly reproduce the observed precipitation and that characteristic of the models has not substantially changed over time. One area where the models have been in continued agreement has been in the Southern Great Plains of the United States. In all three studies, the varied models I have examined agree that northeastern Colorado receives substantially more precipitation than northwestern Louisiana! That is in marked contrast with reality where Louisiana is obviously wetter than Colorado. But the important ramification of this is that if precipitation is badly simulated in a climate model, then that will adversely affect virtually every other aspect of the model simulation. Precipitation affects the energy, moisture, and momentum balances of the atmosphere and directly affects the modeling of the, atmosphere, the hydrosphere, the biosphere, and the cryosphere. In turn, a bad representation of these components will again adversely impact the precipitation simulation. In short, anything done wrong in a climate model is likely to be exhibited in the model simulation of precipitation and, in turn, errors in simulating precipitation are likely to adversely affect the simulation of other components of the climate system. Given its integrative characteristic, therefore, precipitation is a good diagnostic for determining how well the model actually simulates reality, especially since simple "tuning" adjustments cannot mask limitations in the simulation, as is the case with air temperature. If we examine climate model output a bit further, we uncover another disturbing fact-climate models simply do not exhibit the same year-to-year or even within-season variability that we observe. Precipitation in a climate model does not arise from organized systems that develop, move across the Earth's surface, and dissipate. Instead, modeled precipitation can best be described as "popcorn-like", with little if any spatial coherency. On a year-to-year basis, both air temperatures and precipitation exhibit little fluctuation, quite unlike what we experience. This is particularly important because it is the climatic extremes and not their means that have the biggest adverse impacts. Simply put, climate models cannot begin to address issues associated with changes in the frequency of extreme events because they fail to exhibit the observed variability in the climate system. I attach a piece I wrote regarding the climate models used in the National Assessment and their evaluation with my climatology, which further highlights our uncertainties in climate models. In fact, the National Assessment itself recognized that both the Canadian Global Coupled Model and the Hadley Climate Model from Great Britain used by the, Assessment provide more extreme climate change scenarios than other models that were available and that had been developed in the United States. Neither model is reasonably able to simulate the presentday climate condi tions. OUR OBSERVATIONAL CAPABILITIES ARE IN JEOPARDY Given that our observational record is inconclusive and that model simulations are fraught with problems, on what can we agree? In my view, there are two main courses of action that we should undertake. First, we need to continue to develop and preserve efforts at climate monitoring and climate change detection. Efforts to establish new global climate observing systems are useful, but we need to preserve the stations that we presently have. There is no surrogate for a long-term climate record taken with the same instrumentation and located in essentially the same environmental conditions. Modernization efforts of the National Weather Service to some extent are undermining our monitoring of climatic conditions by moving and replacing observing sites, thereby further introducing inhomogeneities into these climate records. Some nations of the world have resorted to selling their data, which has adversely impacted our assessments of climate change. However, given that oceans cover nearly three-quarters of the Earth's surface, we need to exploit and further develop satellite-derived methods for monitoring the Earth's climate. We also need to better utilize the national network of WSR88D weather radars to monitor precipitation. But foremost, we need to focus on developing methods and policy that can directly save lives and mitigates the economic devastation that often is associated with specific weather-related events. Climate change discussions tend to focus on increases in mean air temperatures or percentage changes in mean precipitation. But it is not changes in the mean fields on which we need to place our efforts. It would be rather easy to accommodate even moderately large changes in mean air temperature, for example, if there were no year-to-year variability. Loss of life and adverse economic impact resulting from the weather occurs not when conditions are "normal"; but rather, as a result of extreme climatic events: heat waves, cold outbreaks, floods, droughts, and storms both at small (tornado, thunderstorm, high winds, hail, lightning) and large scales (hurricanes, tropical storms, nor'easters). The one thing that I can guarantee is that regardless of what impact anthropogenic increases in atmospheric trace gases will have, extreme weather events will continue to be a part of our life and they will continue to be associated with the most weather-related deaths and the largest economic impact resulting from the weather. Ascertaining anthropogenic changes to these extreme weather events is nearly impossible. Climate models cannot even begin to simulate storm-scale systems, let alone model the full range of year-to-year variability. Many of these events are extremely uncommon so that we cannot determine their statistical frequency of occurrence from the observed record, let alone determine how that frequency may have been changing over time. While we need to continue to examine existing climate records for insights and to develop reliable theory to explain plausible scenarios of change, the concern is whether we can enact policy now that will make a difference in the future. However, is there cause for concern that anthropogenic warming will lead to an enhanced hydrologic cycle; that is, will there be more variability in precipitation resulting in more occurrences of floods and droughts? The IPCC Summary for Policy Makers states: Global warming is likely to lead to greater extremes of drying and heavy rainfall and increase the risk of droughts and floods that occur with El Niño events in many different regions. However, if one reads the technical summary of Working Group I, we find that: There is no compelling evidence to indicate that the characteristics of tropical and extratropical storms have changed. Owing to incomplete data and limited and conflicting analyses, it is uncertain as to whether there have been any longterm and large-scale increases in the intensity and frequency of extra-tropical cyclones in the Northern Hemisphere. Recent analyses of changes in severe local weather (e.g., tornadoes, thunderstorm days, and hail) in a few selected regions do not provide compelling evidence to suggest long-term changes. In general, trends in severe weather events are notoriously difficult to detect because of their relatively rare occurrence and large spatial variability. The IPCC goes on to further state "there were relatively small increases in global land areas experiencing severe droughts or severe wetness over the 20th century". Karl and Knight, who conducted a detailed study on precipitation variability across the United States, concluded that as the climate has warmed, variability actually has decreased across much of the Northern Hemisphere's midlatitudes, a finding they agree is corroborated by some computer models. Hayden, writing for the Water Sector of the U.S. National Assessment, agrees that no trend in storminess or storm frequency variability has been observed over the last century and that "little can or should be said about change in variability of storminess in future, carbon dioxide enriched years." Soden concluded, "even the extreme models exhibit markedly less precipitation variability than observed." In addition, Sinclair and Watterson have noted that, in fact, climate models tend to indicate that increased levels of atmospheric trace gases leads to a "marked decrease in the occurrence of intense storms" outside the tropics and they argue that claims of enhanced storminess from model simulations are more the result of models that fail to conserve mass. Clearly, claims that anthropogenic global warming will lead to more occurrences of droughts, floods, and storms are wildly exaggerated. Thus, I believe it stands to reason that we need to focus on providing real-time monitoring of environmental conditions. This will have two benefits: it will provide immediate data to allow decisionmakers to make informed choices to protect citizens faced with these extreme weather events and, if installed and maintained properly, it will assist with our long-term climate monitoring goals. Such efforts are presently being developed by forward-looking states. For example, I am involved with a project, initiated by the State of Delaware in cooperation with FEMA, the National Weather Service, and Computational Geosciences Inc. of Norman, Oklahoma, to develop the most comprehensive, highest resolution, statewide weather monitoring system available anywhere. Louisiana and Texas also have expressed interests in using our High-Resolution Weather Data System technology for real-time statewide weather monitoring. Regardless then of what the future holds, employing real-time monitoring systems, with a firm commitment to supporting and maintaining longterm climate monitoring goals, proves to be our best opportunity to minimize the impact of weather on human activities. FINAL THOUGHTS: THE SCIENCE IS NOT YET IN In 1997, I had the pleasure to chair a panel session at the Houston Forum that included seven of the most prominent climate change scientists in the country. At the close of that session, I asked each panelist the question, "In 2002, given 5 more years of observations, 5 more years of model development, and 5 more years of tech nological advances and knowledge about the climate system, will we have an answer to the question of whether our climate is changing as a result of anthropogenic increases in trace gas emissions?" The panel, which consisted of both advocates and skeptics, agreed that we would have a definitive answer probably not by 2002, but certainly by 2007. I disagreed then and I continue to disagree today. I fear that the issue has become so politically charged that the political process will always cloud the true search for scientific truth. But more than that, I feel the climate system is far more complex than we ever imagined—so much so that we still will not have a definitive answer by 2007. I again thank the committee for inviting my commentary on this important topic. |