As there are no large-scale modeling groups in Switzerland, my educated guess would be that their model would not be substantially different from those cited by the IPCC, and may likely include the Hadley Centre GCM. I provide an assessment of the climate models used in the U.S. National Assessment in my manuscript published by the George C. Marshall Institute and appended to my original Senate testimony. Many of the same criticisms of these two models hold for other models as well. As for the prediction of local impacts, this study appears to use nested modelingan approach where higher resolution models are used to look at local fluctuations. These models are driven by the coarser resolution GCMs and, as a consequence, inherit their biases and errors. Thus, the local assessments are only as good as the large-scale forcing which, for GCMs, is not very accurate. RESPONSES OF DR. DAVID R. LEGATES, TO ADDITIONAL QUESTIONS FROM SENATOR CAMPBELL Question 1. In your testimony, you expressed concern over what you termed "land bias". That nearly three fourths of the Earth's surface is covered by water and goes largely unobserved. Therefore, much of our available data on global warming may not in fact be wholly accurate. You also mention that some countries actually sell their data to interested parties, also potentially tainting that information. What efforts are being made to correct these situations? Response. Clearly, it is virtually impossible to instrument the oceans in the same way we have instrumented land areas. We do have ship reports; however, they tend to be biased in a number of ways. First, ships, for obvious reasons, tend to avoid storms if at all possible. This provides a "fair weather bias" that affects our estimates. Second, most ships are moving targets (there are some reports from fixedposition ships) and provide air temperature estimates that are integrated over large areas and do not represent a single point. Third, ships are large metal objects that generate their own heat and have different characteristics than the open ocean. This problem is akin to the urbanization effect we see with land-based thermometers. Thus, our only real source of obtaining a spatially representative sample of global air temperatures is through remote sensing. Much of the work by Roy Spencer and John Christy has been based on attempting to compile a long-term temperature record using satellite remote sensing. Using their analysis, we see that satellite-derived air temperature has not exhibited a marked increase as suggested by landbased thermometers. This lack of a trend has also been observed with radiosonde data (balloons); traditionally, weather balloons are used twice daily around the world to sample the vertical profile of the atmosphere, including air temperature. As for the fact that countries have been selling their data, Dr. Mike Hulme of the Climatic Research Unit at the University of East Anglia relayed this information to me. His unit has been the source of many of the air temperature and precipitation time-series that have been displayed. These countries are largely Third World, which see the data as a potential source of income. Efforts are ongoing to encourage these countries to participate in the global telecommunication of weather data, largely through the World Meteorological Organization. In some cases, financial support has been supplied. I participated in the first protocol that allowed the U.S. and USSR to exchange data for climate research (back in 1990); such efforts have now been extended to an international scope. However, I would conclude that global cooperation in this area is still lacking. Question 2. You mention in your testimony that perhaps 20 percent or less of the observed global increase in temperature may be due to the activities of mankind. What are other likely causes of global warming? Response. I believe my intent was to state that 20 percent or less of the observed global increase in temperature was due to anthropogenic increases in greenhouse gases. Variations in solar output are an obvious source of some of the changes in global temperatures we have seen. Dr. Sallie Ballunias probably can offer comment that is more up-to-date on this topic. However, I also would strongly argue that much of the observed global increase in air temperature is due to the effect of urbanization. Over time, weather stations that originally were sited in open, rural settings have become increasingly surrounded by sprawling urban areas. Several researchers have documented time-series of air temperature for rural versus urbanized stations and have found that air temperature increases with urbanization, while little change occurs with rural observations. This effect is well documented; the "urban heat island" occurs due to a decrease in evaporation and an increase in absorption of solar radiation that results when forests and grasslands are replaced by cities. While urbanization technically can be considered as a humaninduced ef fect, I strongly differentiate increased temperatures due to urbanization from a rise in air temperature resulting from increased greenhouse gases. Thus, urbanization, in my view, is largely responsible for most of the air temperature rise that we have seen in the observed, land-based air temperature record. I would further argue that land surface changes (such as urbanization, but also including deforestation and desertification) have probably a bigger effect on the Earth's climate than atmospheric constituents. Land surface interactions are a big component of the surface energy balance, although they are not well represented within climate models. Models are more tuned to study the radiative balance of the atmosphere, which is probably why the models are very sensitive to changes in greenhouse gases. Natural climatic variability is also another likely source of rising air temperatures. In the late 1800's, we emerged from a relatively cool period known as the "Little Ice Age”. It is therefore not unexpected that air temperatures would rise during the last century after the end of a period during which colder temperatures were experienced for 300 to 400 years. Before then, the Medieval Warm Period exhibited globally warmer air temperatures. I would note that many civilizations thrived during this period even though they were in a lesser position than we are to adapt to climate change. Fig. 1. Locations of the 17986 terestrial air temperature stations contained in the edited and merged data set. Twelve mean monthly surface wir lemperatures are available for each station From: Legates, D.R., and C.J. Willmott (1990). Mean Seasonal and Spatial Variability in Global Surface Air Temperature. Theoretical and Applied Climatology, 41(1):11-21. Theor. Appl. Climatol. 41. 11-21 (1990) Theoretical and Applied Climatology by Springer-Verlag 1990 1 Department of Geography, College of Geosciences, University of Oklahoma, Norman, Oklahoma, U.S.A., and 2 Center for Climatic Research, Department of Geography, University of Delaware, Newark, Delaware, U.S.A. Mean Seasonal and Spatial Variability in Global Surface D. R. Legates' and C. J. Willmott2 With 7 Figures Received November 23, 1988 Revised February 3, 1989 551.524.2 Summary Using terrestrial observations of shelter-height air temperature and shipboard measurements, a global climatology of mean monthly surface air temperature has been compiled. Data were obtained from ten sources. screened for coding errors, and redundant station records were removed. The combined data base consists of 17986 independent terrestrial station records and 6955 oceanic grid-point records. These data were then interpolated to a 0.5° of latitude by 0.5° of longitude lattice using a spherically-based interpolation algorithm. Spatial distributions of the annual mean and intraannual variance are presented along with a harmonic decomposition of the intra-annual variance. 1. Introduction Virtually every component of the earth-atmosphere system influences and is influenced by surface air temperature (temperature of the air at a standard height above the ground). Radiative properties of the atmosphere, the availability and state of water, wind currents, surface albedo, solar angle, and clouds, for example, all are directly related to the temperature of the air within the planetary boundary layer (Willmott. 1987). Surface air temperature, in other words, is a "state" variable that expresses a current or integrated condition of the atmosphere within the boundary layer. Due to its integrated nature, surface air temperature is used in a wide variety of climatological applications. Climate models, for instance, use surface air temperature in the estimation of ground, sensible, and latent heat fluxes as well as for computing atmospheric counterradiation (Washington and Parkinson, 1986). In turn, the model-simulated air temperature field often is used to evaluate the performance of these models. Climatic change too, allegedly induced by greenhouse gases, urbanization, or other environmental factors is manifested largely in the surface air temperature field (cf., Jones and Kelly, 1983; Jones et al., 1986; Hansen and Lebedeff, 1987). Other investigations employ surface air temperature to delineate weather types or climatic regions, estimate evapotranspiration, or evaluate human comfort (Oliver and Fairbridge, 1987). Owing to the importance of surface air temperature and our incomplete knowledge of it, improved representations continue to be needed, especially at the large scale. Large-scale, surface air temperature climatologies have been deficient because of spatially uneven terrestrial station distributions, near complete absence of reliable oceanic measurements, and coarse grid resolutions. The air temperature climatology presented here represents an improvement inasmuch as it consists of a dense network of terrestrial stations and includes shipboard measurements. It may, in fact, be the highest resolution, global air temperature climatology available. Remotely-sensed estimates of surface air temperature are not considered since the technology is not yet mature and long-term means are not available. Their exclusion also allows this climatology to serve as an independent ground truth against which remotely-sensed data may be compared. 2. Station and Shipboard Observations 2.1 Terrestrial Measurements Global archives of shelter-height air temperature have been compiled by Wernstedt (1972), Willmott et al. (1981), and the National Center for Atmospheric Research (Spangler and Jenne, 1984) and they are used in this study. Wernstedt (1972) and Willmott et al. (1981) encoded and published monthly climate averages for 10687 and 13461 stations, respectively. Monthly averages, however, had to be computed from 2721 monthly timeseries contained in the National Center for Atmospheric Research (NCAR) archive. Using only these three archives, adequate spatial coverage can be achieved for most of the terrestrial surface with the exceptions of Antarctica, Australia, New Guinea, China, and other parts of the Far East. To improve the spatial resolution in these regions, additional monthly averages were obtained from eighty-one stations in Antarctica (van Rooy, 1957; Schwerdtfeger, 1984), forty-eight stations in Australia and New Guinea (CSIRO, 1962-71; ADND, 1965), and 417 stations in China and the Far East (Nuttonson, 1947; Terjung et al., 1985). Virtually all the data were used in order to achieve a dense spatial resolution. These data then do not represent climatic normals but, rather, they are based on time-periods of differing lengths. Most of the data were compiled between 1920 and 1980 and so this climatology is generally representative of that sixty-year period with a bias toward the data-rich latter years. Potential coding errors were identified by interpolating monthly averages (see next section) for each station location using only the surrounding stations. When the absolute difference between a recorded and interpolated monthly value was greater than 5°C, the recorded observation was checked for accuracy. Station location (i.e., the encoded latitude and longitude) also was evaluated to ensure that it was located within the recorded political division (country, state, or province). For the NCAR stations (which were recorded only to the nearest tenth of a degree), an atlas was consulted (Rand McNally & Co., 1980). Monthly climatic averages for 27415 stations then were merged into a single database. Many of the stations, however, were represented within more than one archive; therefore, it was necessary to combine or delete “redundant" records. Redundant records either 1) had the same latitude and longitude or 2) were located within 0.05° of latitude and longitude from one another and had virtually identical station names. Airport and downtown stations, however, were not considered redundant. It also was assumed that no two records within a single source were redundant. Redundant records were merged into a single record on the basis of record length. Each redundant record first was classified into one of three categories: 1) records for which the time period (dates) was known, 2) records for which only the total number of years was known, or 3) records of unknown duration. Redundant records of unknown duration were deleted in favor of dated records or those of known duration. When redundant records all were of unknown duration, arithmetic averages were taken to obtain the merged record. Duration only records were discarded in favor of dated records unless the dated record was of climatically short duration (i.e., less than ten years). Redundant duration-only records were merged by taking a weighted average where the number of years of record served as the weights. Dated records similarly were merged, that is, using the length of the non-overlapping portion of the records as the weights (Legates, 1987). After editing and merging, 17986 independent station records were obtained. Station locations are mapped on a cylindrical equal-area projection to facilitate the comparison and interpretation of regional station densities (Fig. 1). This simple projection translates latitude (8,) only (not longitude) according to |