seasonal NDVI data in northern latitudes, from 1982 to the end of 1990, are mapped in Fig. 3a. Data were averaged from May to the end of September, to approximate the main active growing season of land vegetation in the Northern Hemisphere. In Eurasia, a band of increasing NDVI extends from Spain in a northeasterly direction across Asia to the western Pacific Ocean. In this band, central Europe, southerr. Russia, and a broad region near Lake Baikal in Siberia are most affected. Outside this band, northern Scandinavia, northern China, and northeastern Siberia are also strongly affected. In North America, a band of increasing NDVI extends from Alaska in a southeasterly direction to the Great Lakes, thence northeasterly to Labrador. In this band, northwestern Canada is most strongly affected. Outside this band, the continental United States (excluding Alaska) and the area around the Hudson Bay show little change in NDVI In general, the regions of greatest increase in NDVI are inland from die oceans, except in the Arctic, and are north of 50° N. The prominent bands of increased NDVI referred to above in both Eurasia and North America, correspond generally to areas of high NDVI (Fig. 3b). Thus most of the areas where changes in NDVI amplitude and seasonality were observed are also regions of significant vegetation density. Notable exceptions are several Arctic regions in Eurasia where NDVI rose sharply from low initial values. We believe the increasing trend in photosynthetic activity of the northern high latitudes, inferred from satellite observations of NDVI amplitude and phase, to be robust despite varying satellite overpass times and the lack of an explicit atmospheric correction. These effects, however, could modify the magnitudes of NDVI amplitude and estimates of the active growing season duration. Analyses of station temperature trends during 1961-90 indicate pronounced warming over substantial areas in Alaska, northwestern Canada and northern Eurasia'. The greatest warming, up to 4°C, has occurred in winter. Only slightly less warming has occurred in the same regions in spring, but considerably less warming in summer and even less in autumn'. Associated with warming at high latitudes is an approximate 10% reduction in annual snow cover from 1973 to the end of 1992, especially an earlier disappearance of snow in spring (Table 1 of ref. 4). Where snow-lines have retreated earlier due to enhanced warming, we expect an early start of the active growing season. The winter and spring warming in the interior of the continents of Asia and North America in the 1980s may be a result of natural causes not yet explained, but its timing is consistent with an enhanced greenhouse effect caused by build-up of infrared-absorbing gases in the atmosphere. The unusual warming which peaked near 1990 was of global extent. Although it amounted to a departure of only a few tenths of a degree from previous record temperatures". it was associated with far greater warming in the spring months at high northern latitudes. Biospheric activity there, based on our analysis, increased remarkably as a result of this warming, suggesting that small changes in global temperature may reflect disproportionate responses at the regional level, and may be accompanied by positive feedbacks which can markedly influence processes such as photosynthesis and litter decomposition. Received 29 August 1996; accepted 3 March 1997. 1. Keeling, C. D. Chin, J. F. S. & Whorf, T. P. Increased activity of northern vegetation inferred from atmospheric CO, measurements. Nature 382, 146-149 (1996). 2. Chapin, F. S., Zimov, S. A., Shaver, G. R. & Hothic, S. E. CO, fluctuation at high latitudes. Nature 363, 585-586 (1996). 3. Chapman, W. L. & Walsh, J. E. Recent variations of sea ice and air temperatures in high latitudes. Bull Am. Meteorol. Soc. 74, 33-47 (1993). 4. Groisman, P. Ya, Karl, T. R. & Knight, T. W. Observed impact of snow cover on the heat balance and the rise of continental spring temperatures. Science 263, 198-200 (1994). 5. Tucker, C. L in Advances in the Use of NOAA AVHRR Data for Land Applications (ed. D'Souza, D.) I19 (European Economic Union Press, Brussels, 1995) 6. James M. E. & Kalluri, S. N. V. The Pathfinder AVHRR land data set an improved coarse-resolution data set for terrestrial monitoring, Int. J. Remote Sens. 15, 3347–3364 (1994). 7. Tucker, C. I. Fung, 1. Y. Keeling, C. D. & Gammon, R. H. Relationship between atmospheric CO, variations and a satellite-derived vegetation index. Nature 319, 195–199 (1986). & Asras, G., Fuchs, M., Kanemas, E. T. & Hatfield, J. L. Estimating absorbed photosynthetic radiation and leaf area index from spectral reflectance is wheat Agron J. 76, 300-306 (1984). 9. Myocai, R. B., Ha, F. G., Sellers, P. J. & Marshak, A. L. the interpretation of spectral vegetation indenes LEEE Trans. Geosci. Remote Sens 33, 481-486 (1995). 10. Tucker, C. J. Newest W. W. St Degne, A E AVHRR data sets for determination of desert spatial exter. Int. J. Remote Sens. 15, 3547-3566 (1994). 11. Rea, C.R. N. & Chen, J. Inter-satellite calibration linkages for the visible and near-infrared channels of the advanced Very High Resolution Radiometer on the NOAA-7, -9, and -11 spacecraft. Int. J. Remote Sen 16, 1931-1942 (1995). 12. Los, S. a. Calibration adjustment of the NOAA AVHR Normalized Difference Vegetation Index without recourse to consponent channel 1 and 2 data. Int. J. Remote Semm 14, 1907-1917 (1993). 13. Holben, B. N. Characteristics of maximum value composite images for temporal AVHRR data. Int. J Remote Sens 7, 1417-1437 (1986). 14. Myneni, R. B., Tucker, C. J. Asrar, G., Keeling, C. D. & Nemani, R. R. Increased vegetation greenness amplitude and growing season duration in northern high latituds inferred from satellite-sensed vegetation index data from 1981-91. NASA Tech Meme. 104638 (NASA Goddard Space Flight Center, Grecabek, MD, 1996). 15. Myneni, R. B. Los, S. & Tucker, C. J. Satellite-based identification of linked vegetation index and sea surface temperature anomaly areas from 1982-1990 for Africa, Australia and South America Geophys. Res. Lett. 23, 729–732 (1996). 16. Kerling, C. D., Whorl. T. P., Wahlen, M. & van der Plicht, J. Interannual extremes in the rate of rise of atmospheric carbon dioxide since 1980. Nature 375, 666-670 (1995). 17. Heimann, M, Kerling, C. D. & Tucker, C. 1. in Aspects of Chimase Variability in the Pacific and Western Americas (ed. Peterson, D. H.) 277-303 (Geophys. Monog Ser, Am Geophys. Union, Washington DC. 1989). 18. Kauppa, P. E., Martikainen, K. & Kuusela, K. Biomass and carbon budget of European forests from 1971-1990 Science 256, 70-74 (1992). 19. Jacoby, G. C. D'Arrigo, R. D. & Davagamis, T Mangoban tree rings and 20th-century warming So: 273, 771-773 (1996). 20. Houghton, J. T. et al (eds) Climate Change 1995 1-365 (Cambridge Univ. Press, 1995). 21. Jones, P D., Wigley, TM. L. & Hiffa, K. R. in Trends 93 A Compendium of Dasa on Global Change (eds Boden. T. A., Kaiser, D. P., Sepanski, R. 1. & Stos, FW) (ORNLCDIAC 65. Oak Ridge, TN, 1994). 22 Piper, SC & Stewart, E. F. A gridded global data set of daily temperature and precipitation for terrestrial biosphere modelling Glob Biogeochem Cycles 10, 757-782 (1996) Acknowledgements. We thank S. C. Piper and E. F. Stewart for analysis of the station temperature data, SC Piper and T. P Whorf for discussions, and S. Los for help in the calibration of GIMMS NDVI data This work was supported by the Office of Mission of Planet Earth of NASA E F. Stewart's collaboration was made possible by funds from the Electric Power Research Institute and the USNSF Correspondence should be addressed to R.B.M. (e-mail: rmyneni@fcrua.bu.edu) Records of Total Rainfall and Extreme Rainfall for the United States Over the Last Century Q8. A8. On pages 4 and 5 of your written testimony, you discuss NOAA's Tom Karl review of the records of total rainfall and extreme rainfall for the United States over the last century and conclusions he reached based upon that review. Please provide a copy Mr. Karl's review and the underlying data used in that review. Please see the attached article by Tom Karl and Richard Knight, “Secular Trends of Precipitation Amount, Frequency, and Intensity in the United States," which was recently published in the Bulletin of the American Meteorological Society. Secular Trends of Precipitation in the United States Thomas R. Karl and Richard W. Knight NOAA/NESDIS/National Climatic Data Center, Asheville, North Carolina ABSTRACT Twentieth century trends of precipitation are examined by a variety of methods to more fully describe how precipitation has changed or varied. Since 1910, precipitation has increased by about 10% across the contiguous United States. The increase in precipitation is reflected primarily in the heavy and extreme daily precipitation events. For example, over half (53%) of the total increase of precipitation is due to positive trends in the upper 10 percentiles of the precipitation distribution. These trends are highly significant, both practically and statistically. The increase has arisen for two reasons. First, an increase in the frequency of days with precipitation [6 days (100 yr)'] has occurred for all categories of precipitation amount. Second, for the extremely heavy precipitation events, an increase in the intensity of the events is also significantly contributing (about half) to the precipitation increase. As a result, there is a significant trend in much of the United States of the highest daily year-month precipitation amount, but with no systematic national trend of the median precipitation amount. These data suggest 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. These changes have an impact on the area of the United States affected by a much abovenormal (upper 10 percentile) proportion of precipitation derived from very heavy precipitation events, for example, daily precipitation events exceeding 50.8 mm (2 in.). 1. Introduction In many areas of the United States during recent years, there has been a notable number of catastrophic flooding episodes. A few examples include the 1993 flooding event along the Mississippi, the New England floods during the autumn of 1996, the winter floods of 1997 in the Pacific Northwest and California, and the 1997 spring floods along the Ohio River and the Red River Valley. Previous work (Karl et al. 1996) has documented an increase in the proportion of the area of the United States affected by a much above-normal frequency of extreme precipitation events, for example, > 50.4 mm day1 (or 2 in.). A Corresponding author address: Thomas Karl, NOAA/NESDIS/ National Climatic Data Center, 151 Patton Ave., Asheville, NC 28801-5001. E-mail: tkarl@ncdc.noaa.gov ©1998 American Meteorological Society thorough analysis of how precipitation is changing in the United States, however, has not been provided. Changes in precipitation have most often been quantified in terms of changes in the total precipitation over long averaging periods, for example, annually, seasonally, and occasionally monthly. Such statistics (Karl et al. 1993; Groisman and Easterling 1994; IPCC 1990, 1996), although quite useful for many applications, do not reveal important aspects of how precipitation changes within such a long averaging period. After all, most precipitation events in the midlatitudes last a few days at most. It would be remiss not to mention some notable work that has emphasized changes in precipitation intensity (Englehart and Douglas 1985; Diaz 1991; Yu and Neil 1991; Nicholls and Kariko 1992; Karl et al. 1995; R. Suppiah and K. Hennessy 1998, manuscript submitted to Int. J. Climatol.; Mearns et al. 1995). In these analyses, however, there has been no standard technique of investigating precipitation intensity. For example, R. Suppiah and K. Hennessy (1998, manuscript submitted to Int. J. Climatol.) calculate trends equivalent to the number of days with precipitation to understand how the frequency of precipitation contributes to changes of precipitation, but only the trends of the 90th and 95th percentiles of daily precipitation amount are used to calculate how the intensity of precipitation may be affecting the trend. Nicholls and Kariko (1992) define precipitation intensity as the mean rainfall per day, but define a rainfall event as any period of days with consecutive rainfall. Mearns and Giorgi (1995) analyze on a monthly basis the number of precipitation days, the average rainfall per rain day (what they define as intensity), and the average rainfall per day. Although there is no single method of analysis that can comprehensively cover all the important aspects of how precipitation changes over the course of time, it is fairly apparent that more consideration needs to be given to the type of questions various analyses can address. For example, a rather fundamental question might be related to how much of any precipitation increase or decrease is attributable to changes in the frequency of precipitation versus intensity of precipitation. For example, increased precipitation could be derived from simply more days during the year with precipitation, and they may be equally distributed for all quantiles' of daily precipitation amount. Alternatively, one could also envision a situation where the number of days with precipitation does not change, but the amount of precipitation changes for all, or a limited number of quantiles. 2. Data (HCNs). The data from these stations span the period 1910-96, but there is some missing data, and some stations do not have data back through 1910. To prevent missing data from introducing any bias, Karl et al. (1995) describe a procedure that was used to estimate missing data. Basically, a gamma function is fit to each station's daily data for each month of the year. To determine if precipitation occurs on any missing day, a random number generator is used such that the probability of precipitation is set equal to the empirical probability of precipitation during that month. If precipitation occurs, then the gamma distribution is used to determine the amount that falls for that day, again using a random number generator. The other two datasets that are included in this study are used primarily to serve as a cross check against the 182 daily dataset. This includes the climatological state divisional precipitation data (Guttman and Quayle 1996), which are monthly averages based on all reporting stations in the United States. In some years and months, this network reaches over 7500 stations. Most of these stations are cooperative weather stations that have not changed in instrumentation during the twentieth century, unlike the firstorder stations, which have been affected by new automated instruments and the introduction of wind shields (Karl et al. 1993). These data span the period of the HCNs data, but there is an uneven number of stations that enter and leave the network during the course of the twentieth century, possibly contributing to some bias in trends. The other dataset (TD3200) consists of 3091 stations in the United States that reported daily precipitation and passed our completeness criterion. The period of record is shorter for these data, spanning the years 1948-95, and each station had to have at least 80% of all data present. The TD3200 data were subjected to the standard National Climatic Data Center (NCDC) data checks as given in TD3200 documentation. There are several datasets that are used in this analysis. The primary dataset is the daily precipitation dataset used by Karl et al. (1996). This dataset consists of 182 stations across the contiguous United States. Of these 182 stations, 134 are part of the U.S. Historical Climate Network (HCN, Hughes et al. 1993). An additional 48 stations were added to improve data coverage in the western United States. 3. Methods Detailed station histories for all of these stations indicate that standard 8 in. precipitation gauges have been used throughout the twentieth century at all locations. This dataset is referred to as the HCN special network The value of any quantile (Q) in a sample is given by the ordered data values themselves. The order of the quantile is given by P = (i −0.5) n', where i = 1 ton, and n is the sample size. So Q(0.5) is the median, Q(0.25) is the first quartile, etc. a. Spatial averages The HCNs daily precipitation data as well as the TD3200 data were arithmetically averaged into 1° x 1° grid cells. These grid cells were then area weighted to calculate changes of precipitation for nine regions across the United States. A national average was derived from these nine regions by area weighting the values within each region on a monthly basis. All sea |