WASHINGTON ROUNDTABLE ON SCIENCE & PUBLIC POLICY

Global Warming and the Hydrologic Cycle: How are the Occurrence of Floods, Droughts, and Storms Likely to Change?
By David Legates

Washington, D.C.

The George C. Marshall Institute

The George C. Marshall Institute, a nonprofit research group founded in 1984, is dedicated to fostering and preserving the integrity of science in the policy process. The Institute conducts technical assessments of scientific developments with a major impact on public policy and communicates the results of its analyses to the press, Congress and the public in clear, readily understandable language. The Institute differs from other think tanks in its exclusive focus on areas of scientific importance, as well as a Board whose composition reflects a high level of scientific credibility and technical expertise. Its emphasis is public policy and national security issues primarily involving the physical sciences, in particular the areas of missile defense and global climate change.

The Washington Roundtable on Science and Public Policy

The Washington Roundtable on Science and Public policy is a program of the George C. Marshall Institute. The Roundtable examines scientific questions that have a significant impact on public policy and seeks to enhance the quality of the debate on the growing number of policy decisions that look to science for their resolution. The opinions expressed during Roundtable discussions do not necessarily represent those of the Marshall Institute or its Board of Directors. Additional copies of this transcript may be ordered by sending $7.00 postage paid to: The George Marshall Institute 1625 K Street, NW Suite 1050 Washington, D.C. 20006 Phone: 202/296-9655 Fax: 202/296-9714 E-mail: info @marshall.org Website: www.marshall.org

Global Warming and the Hydrologic Cycle: How are the Occurrence of Floods, Droughts, and Storms Likely to Change? by David R. Legates

The George Marshall Institute Washington, D.C.

David R. Legates, Ph.D., is the Director of the Center for Climatic Research and an Associate Professor of Climatology at the University of Delaware. He participated in the joint USA/USSR Working Meeting on Development of Data Sets for Detecting Climatic Change that led to a protocol for the first climatic data exchange between the USA and the USSR. He also has twice been called to testify before US Senate subcommittees. Dr. Legates has published more than 100 journal articles, monographs, proceedings papers, and other publications on climate change and precipitation, surface water hydrology and hydroclimatology, and the use of statistical/numerical methods in climatology.

Global Warming and the Hydrologic Cycle: How are the Occurrence of Floods, Droughts, * and Storms Likely to Change?
David R. Legates April14, 20041
Jeff Kueter: I am pleased today to have with us Dr. David Legates from the University of Delaware as part of our continuing series, the Washington Roundtable on Science and Public Policy. The Washington Roundtable is designed to bring to scientists to Washington, D.C. to interact with congressional staff, members of the think-tank community, industry and whatnot on topics of controversy in the policy realm. Today Dr. Legates will talk about the enhanced hydrological cycle and its relationship to climate change. David Legates: Thank you for hosting me and thank all of you for coming. Today I want to talk about the impact of global warming on the hydrologic cycle. Two weeks ago, as you are probably aware, we witnessed a hurricane, or at least we think it might have been a hurricane, in the South Atlantic, a phenomenon that we never before have seen. The question is, Is this a harbinger of things to come, or is this maybe just one of those rare events that occurs once every fifty years or so, and this happens to be the year in which it occurs? Now I want you to consider the following: What are the weather events that result in the most deaths and incur the biggest economic impacts? Are they changes to the mean field the average conditions? No, they are changes to the extreme conditions. Thus, what we really want to look at is how these extreme conditions might change; that is, the occurrence of floods, droughts, mid-latitude storms (including noreasters), tornados, thunderstorms, tropical storms, tropical cyclone-hurricanes, heat waves and cold spells. Today, I am going to focus on the hydrological cycle and changes in storminess and changes in precipitation, but I wont talk about heat waves and cold spells. Since four of these major events are related to
The views expressed by the author are solely those of the author and may not represent those of any institution with which he is affiliated. 1 Dr. Legates presented this Roundtable on Capitol Hill on April 12 and April 14, 2004.
*

the hydrological cycle, we want to see how they might be changing and how they might change in the future. The question is: How are the occurrence of floods, droughts and storms likely to change in the future? On May 28, a movie called The Day After Tomorrow opens in theaters worldwide. The same producer who brought us Independence Day, where the earth was destroyed by aliens, brings us a movie in which the earth is nearly destroyed by natural forces such as tornadoes, massive snow storms, hail the size of grapefruit, and other extreme climatic events. I mention the aliens in particular because there is an unexpected connection here. But we want to address whether the movie represents, to some extent, what we can expect to see in the future, or whether this is just going to be a Hollywood representation of fantasy, The Day After Tomorrow has been turned into a novel written by Whitley Strieber, a name that may be familiar to some of you. In fact, the screenplay itself comes loosely from a book written by Art Bell and Whitley Strieber entitled The Coming Global Superstorm whose premise is that we can expect a superstorm nearly half the size of a continent. This superstorm will create all sorts of havoc, including fifty-foot tidal waves and fifty feet of snow in New York City. So who are these people? Art Bell is a top-rated talk show host, which hopefully is not his credential for writing about climatology. He used to have a show that ran from about 2 to 6 a.m. and discussed the supernatural, aliens, UFOs, and so forth. Whitley Strieber is a best-selling author, probably not noted so much for The Coming Global Superstorm as for his books Communion and Confirmation, which describe his contacts with aliens. So there is a strong alien connection to the movie here. But my point is that while the movie may be very entertaining, we cannot use it as an indication of future conditions. Instead, I want to go back to our original question: What really is the scientific evidence and what is the science behind what might be changing our future? The Intergovernmental Panel on Climate Change (IPCC) Report is an important element in this. Three assessments have been produced and I am going to talk primarily about the Third Assessment Report, which came out about three years or so ago. I will also allude to the earlier report of 1996 on occasion. Most people dont realize that each IPCC report is not a unitary document, but it is made up, in fact, of two reports the first being the scientific assessment and the second, the Summary for Policymak-

ers. Most people assume the Summary for Policymakers is just that, a summary that takes the science, distills it, removes some of the scientific jargon, and essentially says, Here is the essence of what the scientific document says. That couldnt be farther from the truth. It was created by a completely different process in which each nation has one vote. So if enough nations decide that the earth is flat, for example, then the document shall read, The earth is flat, regardless of the science. So we have a situation where sometimes and I will show you a couple of these cases coming up the scientific document is completely at odds with what we see in the Summary for Policymakers. Warmer temperatures will lead to a more vigorous hydrological cycle; this translates into prospects for more severe droughts and/or floods in some places and less severe droughts and/or floods in other places. Several models indicate an increase in precipitation intensity, suggesting a possibility for more extreme rainfall events. IPCC politics document (1996) I would like to examine some of the statements in this 1996 IPCC Summary for Policymakers. Warmer temperatures will lead to a more vigorous hydrologic cycle. That concept of a vigorous hydrologic cycle is what we refer to as an enhanced hydrologic cycle. There will be prospects for more severe droughts and/or floods in some places and less severe droughts and/or floods in others. When you stop and think about that statement, it really doesnt say anything. There could be more floods, there could be less, more drought or less. It doesnt necessarily say much. Several models, though, indicate an increase in precipitation intensity, suggesting a possibility for more extreme rainfall events, which, if true, could have serious ramifications. The more recent politics document, the Summary for Policymakers (2001), indicates Global warming is likely to lead to greater extremes of drying and heavy rainfall and increase the risk of droughts and floods that occur in many different regions. Now there is no longer waffling; this document indicates that floods and droughts are likely to increase in frequency. Again, if that is true, it may have dire consequences for human life and future economic impacts.

Estimates of confidence in observed and projected changes in extreme weather and climate events
Confidence in observed changes th (latter half of the 20 century) Likely Changes in Phenomenon Higher maximum temperatures and more hot days over nearly all land areas Higher minimum temperatures, fewer cold days and frost over nearly all land areas Reduced diurnal temperature range over most land areas Increase of heat index over land areas More intense precipitation events Increased summer continental drying and associated risk of drought Confidence in projected changes st (during the 21 century) Very likely

Very likely

Very likely

Very likely

Very likely

Likely, over many areas Likely, over many Northern Hemisphere mid-to high latitude areas Likely, in a few areas

Very likely, over most areas Very likely, over many areas

Likely, over most mid-latitude continental interiors. (Lack of consistent projections in other areas)

Not observed in the few analyses available Insufficient data for assessment

Increase in tropical cyclone peak Likely, over some areas wind intensities Increase in tropical cyclone mean and peak precipitation intensities Likely, over some areas

Figure 1

Figure 1, a table from the Summary for Policymakers document, shows the current assumptions and confidence in projected changes over the next 100 years. Summer drying and drought are likely to be in evidence now over a few areas and likely in the future to be in evidence over many more regions. For the increase in cyclone peak wind intensities, that is, will hurricanes and tropical cyclones become more intense, we have not observed that to have happened yet, but the conclusion is that it is likely to happen in some areas. And finally, for the increase in cyclone mean and peak precipitation intensity heavy flooding events occurring with major rainfall events of a tropical nature there currently is insufficient data for an assessment, but it is likely to occur in the future.

Figure 2

Now I want to talk a little bit about theory. Based on the theory, what can we say is likely to happen? Figure 2 shows the amount of water vapor in the atmosphere when the atmosphere is at 100% relative humidity, i.e., when it is saturated. The curve has an exponential increase; that means that if temperatures increase, then the potential for water vapor in the atmosphere at saturation increases. Therefore, with a slightly warmer atmosphere, we can get more water vapor into the atmosphere and when we get precipitation, we can possibly get more precipitation out. So theoretically we could say that, in a global sense, we should expect to see an increase in precipitation. But I cant necessarily say that is true for any particular location because two things are required on a macro scale to get precipitation: moisture in the air and a mechanism to release it. As it has been said, You cant get blood from a turnip, if there isnt much moisture in the atmosphere, you arent going to get much water out. But secondly, regardless of the moisture content of the atmosphere, we still need a mechanism in place to release that moisture, and all of those mechanisms result in cooling the air down to saturation. Some of the most humid places, for example in North Africa, are associated with desert conditions. There is a lot of moisture in the air, but there is nothing to get that moisture to condense.

Now when we examine precipitation in hydrological studies, we encounter three big problems. The first is the bias in precipitation gage measurement, the second is our perception of what we mean by floods and droughts, and the third is climate-modeling problems or complexities associated with large-scale climate models. I will first start with gage measurement biases. I am partial to this issue because it was part of my Ph.D. dissertation and I have done a lot of work on it. One of the things I attempted to do with my dissertation was to put together the highest resolution climatology we had to date. We recognized that when we use a standard raingage, which is little more than a container that catches water, a bias in the measurement occurs, and that bias is largely a function of the wind speed and whether the precipitation falls in the form of rain or snow. In my dissertation, I found that the global bias would be about 11%, which is consistent with a study that was undertaken in the Soviet Union at about the same time. But the bias is not constant; it is different for different places and in areas where there are higher winds and more snowfall, the bias becomes greater. I later examined the effect of precipitation gage measurement biases on climate change detection. I took a hypothetical station, near Madison, Wisconsin, that had no change in its time series, by definition, over a 100year period, and I introduced into that record a slight change in wind speed and temperature. I decreased the wind speed by one knot (one nautical mile) an hour over one century and increased the temperature by one degree Celsius, simulating the potential urbanization that had occurred. Assume, for example that the station was located outside of a small town in 1900 A.D., which grew into a large urban metropolis area by 2000 A.D. A one degree Celsius rise and a decrease of wind speed due to urbanization are not uncommon. But with a simple one-knot decrease in the wind speed and a one degree Celsius increase, the change in the gage measurement bias caused an apparent 6.4% increase in the precipitation just from the change in the siting characteristics alone. So part of the problem is that when we look at time series of precipitation, we are seeing trends in precipitation, but we are also seeing the effect of changes in the siting characteristics around the gage. This makes it difficult to separate the local environment change from the climate change that we are trying to study. The second problem is our perception of floods and droughts. Generally we define floods as streamflow above a certain level. Increases in

flood frequencies can result from increasing rainfall events, but it can also occur due to changes in timing or the form of precipitation. If the rain falls when the soil is saturated or very dry, on some types of soil, there can be more runoff than there would be if the soil were in a different moisture condition. Temperature also is important; if precipitation falls in the form of snow, it will lay there until it melts. If it falls in the form of rain, it will be dealt with immediately, either as runoff or infiltration. Due to urbanization, we replace grasses and trees with asphalt, macadam and other hard surfaces. These surfaces do not allow infiltration, so virtually all of the rainwater becomes runoff, which changes the flooding characteristics. Channelization of streams and rivers, the building of levies, dredging activities, reinforcement of banks, for example, also change the streamflow. The state climatologist of Louisiana, for example, characterizes the Mississippi in southern Louisiana as a freeway with no on-ramp because the US Army Corps of Engineers have built levees and channelized the river. In fact, water that is supposed to flow into the Mississippi from the Comite River basin actually flows in the opposite direction; it is no longer allowed to break the levee. We have changed the hydrologic patterns of the region and, as a result, we have changed in the flooding conditions. So changes in flooding conditions can result from climate changes in precipitation and also from changes in the local characteristics. The same thing happens with drought. Drought is a streamflow below a certain level and increases in drought frequencies can be induced by decreases in rainfall events, but also by urbanization, increased demand for water, and, in particular, an increase in water-intensive activities, such as irrigation, residential or industrial activities. Such problems with observations led Tom Wigley to make the famous statement, But the data are dirty. When you start looking at climate change trends as seen in the observed record, it is full of all sorts of measurement problems and biases. What we want to do is clean that up and get a record that does not have these biases. But where can we get that? From general circulation models (GCMs), of course mathematical representations of the climate system. Using GCMs, we can play what-if scenarios, we can change certain things, see what happens, but more importantly, we can hold things constant, so that measurement biases and changes in the landscape are not an issue. A wide variety of problems are associated with trying to measure precipitation with a GCM. I often say, if you really want to look at the

efficacy of a climate model, dont look at air temperature, look at precipitation. Anything you do wrong in a climate model will show up in precipitation. If you get the mountains in the wrong place, the atmospheric circulation wrong, the amount of moisture going into the atmosphere wrong, the condensation rate wrong in summary, anything you do wrong shows up in precipitation. And now the kicker: anything wrong in the precipitation affects everything else. If I have too much precipitation in a region, I release too much energy from condensing water, which in turn affects the energy balance. Water is part of the mass balance of the hydrological cycle and part of the energy balance, and as a result, getting precipitation wrong will necessarily affect everything else you do.
Zonally-Averaged Global Precipitation by 31 GCMs

Figure 3

Figure 3 is from the Atmospheric Model Intercomparison Project by Larry Gates done a couple years ago, which looks at thirty-one GCMs. I have seen modelers look at this and say, By golly, thats pretty good. We are able to see the low latitude tropics, the dry subtropics, increases in precipitation in mid-latitude, and the dry deserts of the poles. My response is that if a climate model couldnt get the gross characterization of the globe like that, it wouldnt be worth anything at all, so I am not surprised that it

can do that. My concern is that from model to model, there is considerable uncertainty in the level of precipitation. There is uncertainty in our estimation of precipitation, but the differences in Figure 3, I contend, are not trivial. Global Precipitation Estimates Model GFDL GISS NCAR UKMO January 3.13 in. 3.89 in. 3.76 in. 3.54 in. July . 3.35 in. 4.06 in. 4.24 in. 3.93 in.

Global Precipitation Estimates Model GFDL GISS NCAR UKMO Observed January 3.13 in. 3.89 in. 3.76 in. 3.54 in. 4.37 in.
Figure 4

July . 3.35 in. 4.06 in. 4.24 in. 3.93 in. 3.44 in.

Figure 4 is from a study I did several years ago. I examined four different models to compare summertime versus wintertime precipitation. You can see they give us different numbers for globally averaged monthly precipitation; I took data for the given month from the entire globe and averaged it to a single average rainfall depth. In this case, all four models agree, not on the numbers, but they do agree that July is wetter than January. There is one hitch. When you compare that with the observations, exactly the opposite pattern occurs. We can argue from theory that the wetter month should be January, since its summer in the southern hemisphere. Thats where you have more water, thats where you have warmer temperatures, thats where we should see a more vigorous hydrologic cycle. And we do, in the observations. But we dont see that at all in the models and that is a major concern. Now when we start to look spatially at model simulations of precipitation, things get even more complex. Figure 5 shows precipitation over

summer. This is by Doherty and Mearns; using my global precipitation climatology, they compared the two models used in the National Assessment. Model Minus Observed: Precipitation

Figure 5

The Hadley Centre climate model is at the top and at the bottom is the model from the Canadian Climate Centre. What we can see here, if we focus on just the extreme blues and extreme reds and yellows, is that these are areas where we are off by three, six, nine, fifteen millimeters a day. Now if you are off in a model by one millimeter of precipitation a day, that equates to thirty millimeters a month, which is more than an inch of rainfall. In Oklahoma in the central United States, we get about thirty-six inches of rain a year. Divided by twelve, that is about three inches a month. So an error of just one millimeter is an error in precipitation estimation by one-third of the monthly total. That is significant. And in fact some of these numbers are physically impossible because it just doesnt rain that much. But I want to compare this to our model estimates of precipitation and the assumed changes under global warming.

10

Prognostications for 2030: Precipitation

Figure 6

Figure 6 is a scenario for 2030, twenty-five years in the future and thirty years from when the models were run. The colors represent differences from the control run (2000) and green is in the middle: little to no change. Figure 6 is predominantly green, showing relatively small changes. But Figure 5 has lots of blues, reds, yellows, which represents lots of uncertainty. Thus, the prognostications from Figure 6 represent a small signal in a large sea of uncertainty. But look at the scale we are using. In figure 6, the maximum was ten, the minimum was minus five. But in figure 5, we have extremes of plus or minus fifteen, so the scale was stretched on figure 6. So, what we see in figure 5 is the noise, the uncertainty in the model, and there is lots of it. The signal that we get out of the model is figure 6 and it is much less than the uncertainty in the model. So if the model has more uncertainty than the signal it is simulating, it is very difficult to argue that the signal is significant.

11

Figure 7

But there is an even bigger problem. As I said earlier, we are not really interested in changes in the mean per se, we are interested in changes in the extreme. Figure 7 is from an article by Soden in the 2000 issue of the Journal of Climate. The heavy dark line is observed precipitation. The thin line is model ensemble average in year and one model standard deviation is shown by these crossed lines. Soden concludes that GCMs differ with respect to the observations and that they also lack coherence among themselves. We saw that in the spatial distributions between the Hadley Centre Model and the Canadian Climate Centre Model. The Climate Centre Model had lots of drying in the central United States, the Hadley Centre Model doesnt. But Soden says, and this is important, Even the extreme models exhibit markedly less precipitation variability than observed If we want to determine how the extremes might change and our climate model doesnt model the extremes, it is impossible to determine from that model how the extreme conditions are likely to change. Soden goes on, if the GCMs are in error, this deficiency would presumably reflect a more fundamental flaw common to all models. Now, I further argue that those errors that we have seen in the models are not trivial. For example, if I take one millimeter of rainfall, condense it out of the troposphere, compute how much energy is given off by that millimeter of rainfall, and turn that into a temperature change, we find that one millimeter of rainfall is almost 0.4 of a degree temperature change in the troposphere (Figure 8). If you prefer English units, a tenth of an inch is 1 degrees Fahrenheit.

12

P ( w L g ) / ( Cp p ) = T 1 mm of rainfall 0.39C in air temperature or 0.1 inch of rainfall 1.77F in air temperature for the troposphere
Figure 8

Now when we think about the climate change scenarios and the climate change prognostications for air temperature, we can see that being off by just a little bit in precipitation yield errors that are larger than our climate change signal! This is my fundamental argument: if you do things wrong with precipitation, it shows up everywhere else; in this case, it can quite dramatically induce major problems with temperature. So when climate models are tuned to try to get the air temperature right, we may have a fundamental problem because if the models are not adequately simulating the precipitation, the mere process of model calibration and tuning may introduce considerable biases into the model simulation. But enough of the theory what is the observational evidence? In particular, what do we see and what do we expect to see with regard to changes in total precipitation, changes in precipitation frequency and intensity, changes in flood and drought frequency, changes in tropical storm frequencies and intensities, and changes in extra-tropical or mid-latitude storm frequencies and intensities? The first thing we are going to look at is changes in precipitation totals. Precipitation increased by 0.5 to 1% per decade over mid- and high latitudes of the Northern Hemisphere continents increased by 0.2 to 0.3% per decade over the tropical land areas, but not over the past few decades decreased by about 0.3% per decade over Northern Hemisphere subtropical land areas no systematic changes over the Southern Hemisphere insufficient data over the worlds oceans. IPCC politics 2001

13

Trends in terrestrial precipitation estimates over the past century. Since 1980, when the most spatially complete assessments of global precipitation became available, few regions show marked trends in mean precipitation. (from New et al., 2001)

Figure 9

Figure 9 shows more recent results by Michael New and colleagues from the University of East Anglia in Norwich, England. At the top is a long-term time series of the last century for global precipitation over land; the middle is Northern Hemisphere precipitation over land, and the bottom is Southern Hemisphere precipitation over land. None of these signals has a statistically significant slope, so none of them shows significant increases or changes. The conclusion is that since 1980, when the most complete assessment of global precipitation is available, few regions show marked trends in mean precipitation. We do not see large-scale changes in precipitation, except only in selected areas, and that goes back to my argument: it is not just the moisture in the atmosphere but the concomitant mechanism to release that moisture. Concerning changes in precipitation frequency and intensity, the IPCC Summary for Policymakers (2001) states In the mid-and high latitudes of the Northern Hemisphere over the latter half of the 20th century, it is likely that there has been a 2 to 4% increase in the frequency of heavy precipitation events.

14

That is significant and important, if true. Much of that statement comes from Karl and Knight (Bulletin of the American Meteorological Society, 1998), where they examined precipitation regimes in the United States. They concluded that The precipitation regimes in the United States are changing disproportionately across the precipitation distribution. The proportion of total precipitation derived from extreme and heavy events is increasing relative to more moderate events. This analysis presents a problem. Their procedure is to create bins, each representing sequentially heavier rainfall events, and compute the frequency of precipitation events for each bin. Now assume that precipitation is increasing by the same proportion for each category. Some of the events in each category would be moved to the next higher category. In that case, each bin would lose events (moving to the next higher category) but also would gain events (moving up from the next lower category). Thus, only the bin at the end may exhibit a trend, since there is no bin higher than it. We may erroneously conclude that only the precipitation in the highest category is increasing, which clearly is not the case. The other important point to note is that they focus on the proportion of total precipitation. We want to know particularly if it is the frequency of the precipitation events that is changing, not the proportion of the total. But Kunkel et al., (Journal of Climate, 1999) actually looked at that question. For the Midwest, they concluded that heavy event frequencies from 1896-1906 were higher for all ten-year periods except 1986-96. Their conclusion was Interpretation of the recent trends must account for the possibility of significant natural forcing There is no implication in these results that the upward trends will necessarily continue.

15

Figure 10

Last year, they produced another study (Geophysical Research Letter, 2003). Figure 10 shows the changes from 1895-2000 in one-day storm events and five-day storm events. The return period simply means the event is likely to occur once every year on average, once every five years on average, or once every twenty years on average. We can see almost a U shape in both of these graphs. The values for one-day events in the late 1800s and early 1900s are as high as they are now. This is an important issue to note because the values for the late 1800s and the early 1900s must be natural variability; it cannot be anthropogenic change. They concluded Frequencies at the beginning of the 20th Century were nearly as high as during the late 20th Century suggesting that natural variability cannot be discounted as an important contributor to the recent high values.

16

So we are not sure that precipitation frequency intensity is changing. What about changes in floods, droughts, tropical and extra-tropical storms? It is the last three that fall into our category of the major events that lead to the most deaths and the largest economic impact. The IPCC Summary for Policymakers (2001) says Global warming is likely to lead to greater extremes of drying and heavy rainfall and increase the risk of droughts and floods that occur in many different regions. This suggests that we can expect more floods and drought events. But if we look at the same year, the IPCC science document, the Technical Assessment by scientists, indicates Over the 20th century there were relatively small increases in global land areas experiencing severe drought or severe wetness. Thus, we havent seen these things happen. To continue Changes are dominated by inter-decadal and multidecadal climate variability There are years when it went wet and years when it went dry, years when we have had dust bowls, years when we have had floods. There is considerable variability and most of the changes are dominated by that variability. They go on In some regions the frequency and intensity of droughts have been observed to increase in recent decades. In some cases, for example, the Sahel region in Africa, we have seen the effects of desertification. That is likely a feedback from the surface increasing drought event, not a change in the climate. So compare the politics document with the science document and you can see the two dont match. The Summary for Policymakers, therefore, is not a good summary of the scientific assessment.

17

Now lets look at flooding events. Lins and Slack (1999) examined percentiles in streamflow and they found Trends are most prevalent in the annual minimum to median flow categories and least prevalent in the annual maximum category. They examined changes in streamflow by locating streams that are devoid of anthropogenic influences (channelization, urbanization and so forth). They concluded that the low flows are getting higher and the median flows are getting higher which isnt a bad thing but the high flows, the flood events, are in fact, least prevalent and show the least increase. Their conclusion is Hydrologically, these results indicate that the conterminous US is getting wetter, but less extreme. On the other hand, Groisman, Knight, and Karl did a study in 2001 that was very similar to their precipitation analysis. Their conclusion about changes in stream flow for the continental United States was that they saw significant increases in stream flow, particularly for the highest flow events. So on the one hand we have Lins and Slack saying high flow events arent changing and on the other hand, Groisman, Knight and Karl saying they are. How do we reconcile the apparently contradictory statements? It turns out they are not contradictory at all; the two research groups are just answering different questions. Lins and Slack asked the th question: Are trends occurring in stream flow percentiles? Is the 95 percentile of stream flow getting much larger, that is, is the distribution in an event that occurs once in twenty years, for example, changing relative to one that occurs once in five years? Groisman, Knight and Karl ask: Of the total volume of water that changed, how much of that water came from a specific percentile? But when we have high flow conditions, we expect to have lots of water; small percent change to a large number gives us lots of water. So it is true that more of the water is coming in high flow conditions, but that is not really the question we want to ask. The question is: How are the frequencies of these events changing? Thus, Lins and Slacks question is really the one that we want to answer and, therefore, the high flow events really arent increasing as much as the median and lower flows.

18

That is consistent with an overall increase in precipitation, but not an increase in storminess or the extreme events. Finally, lets look at changes in storm frequencies, tropical, extratropical. The 1996 IPCC scientific document indicates In the few analyses available, there is little agreement between models on the changes in storminess that might occur in a warmed world. Conclusions regarding extreme storm events are obviously even more uncertain. You cant say anything; it was just not possible at that time. Did things change, though, over the next five years of research? The IPCC scientific technical document (2001) now indicates There is no compelling evidence to indicate that the characteristics of tropical and extratropical storms have changed. Changes globally in tropical and extra-tropical storm intensity and frequency [have] no significant trends evident over the 20th century. Conflicting analyses make it difficult to draw definitive conclusions about changes in storm activity, especially in the extra-tropics. So apparently not much has changed. Two recent researchers, Henderson -Sellers et al. and Goldenberg et al. conclude In the two regions where reasonably reliable records exist (the North Atlantic and the western North Pacific) there is no clear evidence of long-term trends. Henderson-Sellers et al. (1999) Bulletin of the American Meteorological Society and There have been various studies investigating the potential effect of long-term global warming on the number and strength of Atlantic-basin hurricanes. The results are inconclusive. Goldenberg et al. (2001) Science

19

Thus, we cant conclude that any of these changes are occurring. Henderson-Sellers also concludes It is emphasized that the popular belief that the region of cyclogenesis will expand with the 26C [sea surface temperature] isotherm is a fallacy. Henderson-Sellers et al. (1999) Bulletin of the American Meteorological Society

But

A hurricane or tropical storm is what we call a latent heat engine; that is, evaporating water condenses and gives off energy that churns up the winds, the circulation, and gives us a cyclonic organization that characterizes hurricanes. For this to occur you need warm water, and if the warm water area gets bigger, you may have more reasons for cyclogenesis and deeper storms, which should yield an intensification of storm events. Their conclusion: Its a fallacy. That will not occur. Furthermore The very modest available evidence points to an expectation of little or no change in global frequency. Henderson-Sellers et al. (1999) Bulletin of the American Meteorological Society There have been a number of other studies, which I wont go into detail, but three suggest an increase of tropical storm activity in a warmer world, while two different studies argue for a decrease in tropical storm ac2 tivity in a warmer world. The conclusion is that we dont yet know what might happen and we havent seen any changes taking place. In the Southern Hemisphere, fewer analyses have been completed, but they suggest a decrease in extra-tropical cyclone activity since the 1970s. IPCC science document (2001) From the studies that have been completed, extra-tropical cyclone activity has become less extreme.

Krishnamurti et al. (1998), Walsh and Ryan (1999), and Meehl et al. (2000) suggest an increase in tropical storm activity in a warmer world while Bengtsson et al. (1996) and Yoshimura et al. (1999) argue for a decrease in tropical storm activity.

20

Recent analyses of changes in severe local weather (e.g., tornadoes, thunderstorm days, and hail) do not provide compelling evidence to suggest long-term changes. One thing we find as we look at tornado records is that over the last thirty or forty years, there has been a dramatic increase in the weaker tornadoes, the F0s, F1s and F2s. Has that been a climate change signal? It turns out the answer is no, for two reasons. One is that there is much more interest in chasing storms, particularly by meteorology students at various universities, including the University of Oklahoma, where I worked for 9 years. On the first big chase day in spring, nobody came to class; they were all out trying to get video for CNN. There is also much more interest in chasing storms and many more spotters helping out the National Weather Service by being their eyes in the field. Probably the biggest impact, however, has been the development of Doppler weather radar. With Doppler radar, we can see the circulation in the middle of a wheat field in Kansas when formerly it may have produced a small tornado that disappeared five minutes later without anybody seeing or recording it. Now we may record it on Doppler and so it becomes part of the tornado statistics. So, in general, trends in severe weather events are notoriously difficult to detect because they are rare events and it is difficult to detect changes in the rare event portion of the spectrum. I am now going to read a quote by Sinclair and Waterson (Journal of Climate) in two parts. Doubled CO2 leads to a marked decrease in the occurrence of intense storms [in the extratropics]. Thats the second time weve heard that stated. But they give us an exception: One exception is in the South Pacific, where there is a suggestion of an increased incidence of cyclones at the intense end of the spectrum. Lets investigate the South Pacific.

21

Figure 11

Figure 11 shows a map of topography, what you and I recognize as the earths surface. The mountains are in red, the lowlands in green, the water in white.
A standard T42 truncation scheme (Biasutti et al., 2003). The contour interval is 400m. Negative altitudes of 200m occur west of the Andes while the highest peak in the Andes is only 3000m high.

Figure 12

22

But what a climate model recognizes as the earths topography is quite a bit different. Figure 12 is a spectral model using what is called T42 truncation scheme. Here the Andes mountains are just sort of a general bump; we dont see the individual mountains, and in fact the Andes only reach an altitude of 3000 meters. The mountains of East Africa are represented by a large plateau and the Himalayas appear as simply a rise over a large area. So we really cannot simulate mountains and irregular topography well. But worse yet, what does the model see over the ocean? Several years ago, I talked with Tony Broccoli from the Geophysical Fluid Dynamics Lab in Princeton and he said, We were looking for precipitation patterns over the oceans, and we saw some weird things happen. We could not figure out what those weird things were. Since everyone blocks out the topography over the oceans, I decided to plot it and see what it looks like. And thats what we see in figure 13.
Surface Elevation (m) Represented in an R30 Climate Model (2.25 of Latitude by 3.75 of Longitude)

Figure 13

In these models, topography must be smooth, so the model produces waveforms out over the Pacific and the Atlantic. Note that all the oceans have topographical variations on the order of several hundreds of meters. The Andes Mountains dont stop at the edge of the Pacific Ocean;

23

they extend all the way across and you can still see the remnants of them as we approach Australia. So a wave pattern occurs in topography where in the real world you have a vast expanse of thousands and thousands of miles of water. No topographical variations. But in the model, the air is forced up and down, with the rising motions enhancing the condensation process. So increased storminess in this area could simply be a result of topographical forcing and this is incorrectly represented in climate models over the South Pacific. Now we come back to the second part of Sinclair and Watersons comment: Reductions in average cyclone central pressure that have been used in other studies to promote the possibility of enhanced storminess under greenhouse warming, are more likely the result of global-scale sea level pressure falls rather than any real increase in cyclone circulation strength. Sinclair and Watterson (1999) Journal of Climate If we are looking for a cyclone or for the development of storms, we need to look for an area of low pressure. But is the pressure in certain regions, in the average, decreasing over time? Lower pressures imply more storminess and higher pressures imply less storminess, because high pressure produces clear skies. Their conclusion is that increased storminess is more likely the result of falling sea-level pressure on a global scale. There are two physical ways that could happen. One is that we could be losing substantial amounts of mass of the atmosphere. We are losing hydrogen to space, but thats insignificant. The atmosphere still has essentially the same mass as it did hundreds of years ago. So we are not seeing that as a change. The other possibility is that the atmosphere is expanding, changing the potential energy of the molecules, and hence changing their gravitational force on the earths surface. We do not see that happening significantly either. How else could that happen? Well, in the climate models, it happens because the models do not conserve mass! Some models in fact do not conform to one of the basic premises of the atmosphere, the law of mass conservation. What the modelers do is run the model for a bit, stop and re-inflate the atmosphere, sort of like running on tires with a slow leak. The real world doesnt work that way, but some climate models do and

24

Sinclair and Wattersons argument is that the drops in general pressure that modelers have seen over time are actually the tires deflating. They are not seeing a real process in the atmosphere or a simulated real process in a model; what they are seeing is a model fault, in that the model simply cannot reproduce the law of mass conservation. There is no firm evidence that climate has become more variable over the last few decades. IPCC Politics (1990) There is no evidence that extreme weather events, or climate variability, has increased in a global sense, through the 20th century, ... data and analyses are poor and not comprehensive. IPCC Science (1996) Variability in much of the Northern Hemisphere's midlatitudes has decreased as the climate has become warmer. Some computer models also project decreases in variability. Karl et al. (1997) In fact, we have seen less variability, so in many cases, the argument might be that in a warmer world, we will see fewer, not more, extreme events. This is actually good news, because it is extremes that cause the most economic damage and cause the most deaths. Now for the report of the water sector for the United States National Assessment. Remember that the National Assessment was put together by Mike McCracken and Tom Karl and a few others through the National Assessment scientific team and it was commissioned by a group including Al Gore. Bruce Hayden gave a presentation for the water sector in Atlanta and I was fortunate to be there when he made his presentation. Bruce said, There has been no trend in North America-wide storminess or in storm frequency variability for the period 18851996 It is not possible to attribute regional changes in storm climate to elevated atmospheric carbon dioxide.

25

[Model] projections of North American storminess shows no sensitivity to elevated carbon dioxide. Statements about storminess based on [model] output statistics are unwarranted. Little can or should be said about change in variability of storminess in future, carbon dioxide enriched years. Hayden (1999) Journal of the American Water Resources Association Report of the Water Sector for the US National Assessment One would assume that would make it into the National Assessment. But the National Assessment proclaimed, despite Haydens assertions, It is likely that the observed trends toward an intensification of precipitation events will continue. Thunderstorm and other intensive rain events are likely to produce larger rainfall totals. Projections [for hurricanes] are that peak wind speed and rainfall intensity are likely to rise significantly. US National Assessment (2001) The statements of U.S. National Assessment run directly counter to every argument we have seen earlier. The National Assessment went on to say, While it is not clear how the numbers and tracks of hurricanes will change, projections are that peak wind speed and rainfall intensity are likely to rise significantly. US National Assessment (2001) Here again is the argument by Henderson-Sellers et al., which was available to the National Assessment: It is emphasized that the popular belief that the region of cyclogenesis will expand with the 26C [sea surface temperature] isotherm is a fallacy. The very modest available evidence points to an expectation of little or no change in global frequency. Henderson-Sellers et al. (1999) Bulletin of the American Meteorological Society

26

So lets summarize: What can we say and what cant we say? Changes in Total Precipitation? We saw from theory that it may increase and that in certain cases it had increased. The answer I will give is that it is possible we have seen increases in precipitation and we might see increase in precipitation in the future, but it is certainly below measurement bias and natural variability. Precipitation is a highly noisy field; year to year there is quite a bit of variation, as we saw, and models just dont simulate that at all. Changes in Frequency and Intensity? No, we have not seen changes in precipitation frequency or intensity and this is in contradiction to the IPCC Science document (2001). Now in fairness, Kunkel et al.s research, some of the arguments made regarding Lins and Slacks findings, and the context of the questions that we really should be addressing was not available to the IPCC science assessment in 2001. Changes in Flood and Drought Frequencies? No, we have not seen changes in flood and drought frequencies and that is in complete agreement with the IPCC Science document (2001). Changes in Tropical Storm Frequencies? No, we have not seen changes in tropical storm frequencies and that is in agreement with the IPCC Science document (2001). Changes in Extratropical Storm Frequencies? No, we have not seen changes in extratropical storm frequencies and that is in agreement with the IPCC Science document (2001).

So I will leave you with two pieces of information: Global warming is likely to lead to greater extremes of drying and heavy rainfall and increase the risk of droughts and floods that occur in many different regions. IPCC Politics (2001)

27

The observed trends toward an intensification of precipitation events will continue. Thunderstorm and other intensive rain events are likely to produce larger rainfall totals. Projections [for hurricanes] are that peak wind speed and rainfall intensity are likely to rise significantly. US National Assessment (2001) I ask you the question: Where is the proof for any of these statements? I thank you very much for inviting me here today. Questions and answers
3

Question: What is the problem with the National Assessment estimate? Was it under political pressure? Legates: Well, I do not know enough about how exactly how it was put together, but I have some suspicions. But not being there, I cannot really comment directly that this is what people did or not. All I can say is it just seems to be a bit lacking in scientific balance and that is disconcerting because the IPCC and the National Assessment are erroneously held up as a scientific consensus but they dont often reflect the true state of the science and they generally give a very biased view the National Assessment in particular. I cant say why it happened, but it has happened. Question: You discussed some extreme weather events in the past, for instance, the dust bowl. How does that compare to what the models predict? Do the models predict the dust bowl? Is this a matter of natural variability or is there some other cause for it? Legates: The dust bowl was two different things. One centered on natural variability that caused an extended dry period. But at the same time there was ignorance of the ways in which Midwest land should be managed. They treated it like the humid east, they plowed the land and turning over the topsoil allows it to blow away under very dry conditions. So it was a combination of drier than normal conditions, we saw that in Kunkels graph where the 1930s showed a decrease in precipitation, and natural variability. But it was also accentuated by human activities. So I am not sure, in fair3

These questions are from Dr. Legates presentations on April 12 and 14, 2004.

28

ness to the models, whether they should be expected to simulate the dust bowl. But remember the models dont get year-to-year variability well. They have years that all look very similar to one another. The problem is that models tend to simulate precipitation much differently than the way precipitation occurs. Most models simulate what is called popcorn precipitation; it is largely convective and occurs with very little organized system structure. Our weather map is characterized by the passage of fronts and organized systems, but climate models do not have the spatial resolution to simulate fronts. Yet fronts are important weather-making phenomena. So if a model only simulates convective precipitation events, it is missing, by the nature of its coarse resolution, the ability to simulate other precipitation-forming mechanisms that are equally important. We could get changes, lets say, in precipitation equally if we simply change the largescale circulation. But models arent able to simulate that; its all popcorn convective rainfall. Thus, there is a tendency for every year to look like every other year, so long as the radiation input remains the same, as it generally does. But we are not going to see all of the myriad things that can happen in the real world. So to come back to your question, are they able to simulate the dust bowl? No, but thats probably too because they arent able to simulate the processes that existed either on the anthropogenic side or on the climate side. They are limited in both respects. Question: Could you go over this effect, the oceans being modeled with different sea levels? Legates: The sea-level of the oceans or do you mean the topography as simulated by climate models? Question: The topography of the oceans. Where do those various bumps that we see there come from? Legates: This is from a spectrally based model. That means we are trying to represent all fields by a series of sines and cosines, spherical harmonics, if you will. In this case, we are trying to describe a very rough field by a series of smooth curves. If you have ever seen the modeling of a hat using smooth curves, we can only simulate the hat up to the point at which the vertical gradient occurs. The more terms you use, the more you narrow that area but the more undulation you get in the vicinity of that step change. Thats what is known as the Gibbs phenomenon. So anywhere

29

we have very sharp gradients, it is going to be very difficult for a climate modeler or for spherical harmonics to resolve that dramatic change. But we have that in topography, the Andes in particular; the mountains rise considerably and then come back down to sea level and then theres a flat ocean for thousands and thousands of miles. So what happens is, therefore, that wave field extends across the ocean. You can see it dampening out as it goes. But nevertheless within the modeled oceans, we still have significant elevations to contend with. Question: So you are indicating that those bumps that we see every few hundred miles are artifacts of the process? Legates: Yes, these are the Andes, as they appear as they the harmonics dampen out. Here is essentially the Antarctic Peninsula, as the harmonics dampen out to the north. With distance, things decay. But the sharper the gradient, the longer that field is going to manifest itself into the spherical harmonics. Question: In general, the way people try to cure harmonics like that in the model, is to take more and more samples. Are you at that point? Legates: Figure 12 shows R30, which simply means for rhomboidal truncation at 30 wave numbers. Figure 11 is a little newer analysis with T42 truncation but we can still see the problem here: even with higher resolution, we cant resolve the Andes. There is actually a negative altitude of 200 meters, which will also propagate out all through the ocean as well. If we took more wave numbers, the oceans would dampen out, but there would be considerable problems along the steep gradients more so than we see here. That is the Gibbs effect. But to increase wave numbers significantly would require a lot of computational power, to be able to resolve all those wave numbers, and we just dont have the ability to do that now. Question: So all models have this problem? Legates: All spectrally based models do. There are two classes of models. Models that are spectrally based represent everything as waveforms horizontally while what are called grid-point models represent everything as little boxes. Horizontally you look at things going into and out of boxes. Even in both models there are boxes stacked up on top of one another. But in terms of the grid-point model again, you have fluxes into and out of the box, and you resolve the fluxes crossing the six sides of the box. Spec-

30

tral models have the problem I mentioned; but box models have another problem in that they also have problems with abrupt changes from one cell to another. Question: The word drought is being used and bandied about. What is the technical definition of drought? Is that what is going on with the Ogalla aquifer? Legates: Well, there are all sorts of issues. As you are probably aware, with the Ogalla aquifer, over-pumping leads to lowering in the well water levels. The water gets into the aquifer through percolation; excess precipitation moves down through the soil and winds up in the water table. So normally the excess water would replenish the aquifer and everything would be okay. If we start pumping the aquifer at a rate greater than its recharge, we change the moisture available in the aquifer, which lowers its level. We put much more demand on the water, which changes the amount of water available in the streams. So artificially we are creating a drought condition because of increased usage. Drought can be brought about by lack of precipitation, but it can also be brought about by increased use of the limited water resources you have available to you. Just because we wind up in drought conditions doesnt necessarily mean that precipitation is dropping; it can also mean that the demand for water has increased. Question: Another western issue with regional impact. Based on arguments by the US National Assessment, most of the water storage in the western states that comes from snowfall, like in the Sierra Nevada, is decreasing. So people are saying that in those states you had better get used to reduced snowfall, because it will never go up again. Legates: It is true that warmer temperatures obviously in marginal areas would decrease the amount of snow. At the mountaintops, the temperature is generally cold enough that any precipitation you get is going to come down in the form of snow. I have heard, for example, people talk about the melting of the entire Antarctic icecap. I dont know of any model, for example, that is predicting above-freezing temperatures at the South Pole. So clearly there is still going be an accumulation of ice down there. In fact, with rising temperatures and with more water available, you can get increased precipitation, hence increased storage. Thus, its likely that the Antarctic ice sheets would grow with increased air temperatures.

31

But that brings us back to the Sierra Nevada. The idea is, with slightly warmer temperatures in the midwinter, it is possible to get more snowfall. The issue, though, is most of their water in the year comes from the snowfall. Again if there is an increased demand with more population, that can lead to a lack of water, which isnt necessarily climatologically driven. So there is some truth in the argument that you have to be aware in the future that there may not be water reserves. But it may not be global-warming-induced; it may simply be increased demand. Again, there is that two-level issue associated with all water resources. There are changes in supply and changes in demand. So even though the supply may remain the same, if demand goes up, your reserves go down, and that is not necessarily an indicator that the supply has changed. Question: You talked about how the results from different researchers achieve different conclusions. Are they working on different assumptions of CO2 levels warming and that leads to different results? Legates: Not necessarily that they are working on different assumptions of CO2 warming, but a different understanding of the climate system. The climate system is inherently complex. If it were simply a radiation budget issue, they can say increased carbon dioxide will necessarily increase temperatures and that will necessarily lead to this, that and the other. But the climate is a series of feedback mechanisms, some positive, some negative. For example, if atmospheric circulation were to change, it may decrease the precipitation, which in turn, may decrease the evaporation, which may make the temperature rise even greater. By contrast, if it changes so you get more precipitation, it may decrease the temperature because more energy can go into evaporating moisture. So it is a really complicated series of what-if scenarios. And the various researchers are each determining what they think this is more important, but others are saying, no I think that is more important than this. So it gets down into an area of uncertainty. Of course, any model can predict what you what it to, if you make this model be more attuned to this and less attuned to that, you might be able to make it say what you want. But it doesnt necessarily mean any of it is right. It is just a very complicated issue and the different researchers use different assumptions on how the climate system may respond. Question: You were talking about the National Academy of Sciences report; it was basically put together in a matter of two weeks and is basically a regurgitation of the IPCC report. It should not be considered an independent report of any sort.

32

Legates: I would argue it is a regurgitation of the IPCC Summary for Policymakers rather than a scientific document. In fairness, thats what it was intended to be, a political document, not a scientific document. I think Tom Karl made that statement in his congressional testimony several years ago. Kueter: Thank you very much for coming.

33

RECENT WASHINGTON ROUNDTABLES ON SCIENCE AND PUBLIC POLICY

Lori Garver, Stewart Nozette, Richard Buenneke, & Robert Butterworth Evaluating the New Space Policy: A Panel Discussion (February, 2004) Prasanna Srinivasan, Roger Bate, & Henry Miller - Use and Misuse of Science in Regulating Chemicals: Unintended Consequences for Developing Countries (December, 2003) Ross McKitrick & Steve McIntyre: The IPCC, the "Hockey Stick" Curve, and the Illusion of Experience: Reevaluation of Data Raises Significant Questions (November, 2003) with CEI Hans Mark Sea-based Defense against Ballistic Missiles (October, 2003) Michael Canes Unraveling the Puzzle: Differing Economic Estimates of Climate Policy (September, 2003) Robert Gallo The AIDS Global Health Crisis: Strategies for Policy and Science for Its Resolution (September, 2003) Gregory Canavan Missile Defense for the 21 Century (September, 2003) Michael Gough, Roger Bate & Henry Miller A Discussion of Politicizing Science (July, 2003) Marc Landy & Charles Rubin Civic Environmentalism: Developing a Research and Action Agenda (June, 2003) James Oberg Toward a Theory of Space Power: Defining Principles for U.S. Space Policy (May, 2003) Willie Soon Was the 20 Century Climate Unusual? Exploring the Lessons and Limits of Climate History (May, 2003)
th st

The Marshall Institute Science for Better Public Policy

Board of Directors Robert Jastrow, Chairman Mount Wilson Institute (ret.) Frederick Seitz, Chairman Emeritus Rockefeller University William OKeefe, President Solutions Consulting Bruce N. Ames University of California, Berkeley Sallie Baliunas Marshall Institute Senior Scientist Gregory Canavan Los Alamos National Laboratories Thomas L. Clancy, Jr. Author Will Happer Princeton University Willis Hawkins Lockheed Martin (ret.) Bernadine Healy U.S. News & World Report John H. Moore President Emeritus Grove City College Robert L. Sproull University of Rochester (ret.) Chauncey Starr Electric Power Research Institute