552 Doklady Earth Sciences, Vol. 366, No. 4, 1999, pp. 552–556. Translated from Doklady Akademii Nauk, Vol. 366, No. 2, 1999, pp. 248–252. Original Russian Text Copyright © 1999 by Arpe, Bengtsson, Golitsyn, Mokhov, Semenov, Sporyshev. English Translation Copyright © 1999 by åÄàä “ç‡Û͇ /Interperiodica” (Russia). The reasons for changes in the Caspian Sea level (CSL) have been intensely investigated during the last few decades [1]. Scientific investigations were spurred by a sharp drop of the sea level by 1.7 m between 1933 and 1940, with a further drop of 1.3 m by the end of 1979, which had substantial social and economic con- sequences. The slow lowering of the CSL was related to the growing irreversible water consumption in the Caspian Sea basin, whose influence on the river dis- charge was found to be predominant in comparison with that of climatic factors. Consequently, a further lowering of the sea level was regarded as inevitable. The CSL began to rise unexpectedly in 1978, which had still more catastrophic consequences in the coastal territories developed during the previous lowering. The CSL continued to rise until mid-1995, resulting in a total increase of 2.5 m. According to the satellite mea- surements of the TOPEX–POSEIDON program [2], the average annual CSL began to fall rapidly in July 1995 and, by the beginning of 1997, was 40 cm lower than its maximum height. From 1997 to August 1998 (the latest data), the level was virtually constant. The CSL increments are chiefly determined by the difference between two large values: the river discharge and the apparent evaporation (the precipitation–evapo- ration difference) from the sea surface. A high correla- tion coefficient between these values, equal to 0.82 dur- ing the period 1900–1992, points to the crucial role of river discharge in interannual variations of the CSL. The correlation between the apparent evaporation and the CSL during the same period is also statistically sig- nificant (–0.46). One should note the anthropogenic influence both on the average annual value and on the annual trend of the river discharge. In particular, from the end of the 1940s to the middle of the 1960s, water reservoirs in the Volga basin were filled with a total vol- ume of about 200 km 3 . This work is based on the data of many years of observation of the Volga River dis- charge [3] and precipitation over the water catchment area [4] with an average monthly resolution. The Volga discharge constitutes 82% of the total river discharge, and the correlation coefficient between the average annual series of these values is 0.96 for the period from 1900 to 1992. A promising method for investigating the reasons causing the variations in CSL is the analysis of numer- ical experiments with general circulation models for the atmosphere and the ocean. Such investigations were first carried out in 1994 in the framework of a special subproject called the “Caspian Climate” of the largest International Atmospheric Intercomparison Project (AMIP) [5]. In the AMIP, numerical experiments with the use of general circulation models (GCM), average monthly values of the sea surface temperature (SST), and the position of the sea ice boundary for the period 1979–1988 were specified as boundary conditions. The majority of GCM with a high spatial resolution have differently reproduced specific features in the annual trend and interannual variability of the observed precip- itation over the Volga and Ural water catchment areas, as well as the CSL rise during the analysis period, which coincided with the period of the observed CSL rise [5, 6]. This communication is based on the results of more prolonged numerical experiments with the GCM ECHAM4 [7] of the Max-Planck-Institute for Meteo- rology (Germany), one of the most successful models in reproducing the climatic characteristics of the Cas- pian region from the results of AMIP. Average monthly values of the SST and the sea ice boundaries [8], as well as changes in the content of greenhouse gases in the atmosphere, were specified as boundary conditions for the atmospheric model in all of the calculations per- formed. Two experiments were carried out for the 1903–1994 period and two for the 1951–1994 period. The spatial resolution of the model was approximately 2.8° in latitude and longitude. The Caspian Sea occu- pies 6 cells in the model grid, while 31 cells are assigned to the Volga catchment area. Analysis and Modeling of the Hydrological Regime Variations in the Caspian Sea Basin K. Arpe, L. Bengtsson, Academician of the RAS G. S. Golitsyn, Corresponding Member of the RAS I. I. Mokhov, V. A. Semenov, and P. V. Sporyshev Received January 11, 1999 Max-Planck-Institute for Meteorology, Hamburg, Germany Oboukhov Institute of Atmospheric Physics, Russian Academy of Sciences, Pyzhevskii per. 3, Moscow, 109017 Russia Voeikov Main Geophysical Observatory, ul. Karbysheva 7, St. Petersburg, 194018 Russia GEOPHYSICS