The aim of the paper is to develop a physically based dynamic model describing the hydrothermal processes occurring within a soil-vegetation-atmosphere system throughout the whole year on the scale of an agricultural field or natural ecosystem. The main emphasis of the paper is on an analytical mathematical formalism. Its application does not allow the development of a general, simple and convenient analytical model for a whole year, which is why, in the paper, the year is divided into four periods: the warm period; the period of soil freezing; the pre-spring period and the spring snow melt period. For each period, an analytical model was developed, then all four models were linked into one model package named ANDY-SWASH allowing the permanent calculations throughout a year and year by year. ANDY-SWASH includes the following models: VAST describing the processes of heat and water exchange within a soil-atmosphere-vegetation system during the warm period of a year; FRED which simulates the dynamics of soil freezing depth during the cold period; SOT for the calculation of the vertical profile of soil temperature before the spring snow melt and MIR simulating the spring snow melting with subsequent infiltration and surface run-off. Principal advantages of the models are that being; physically based and well provided with parameters they have relatively simple mathematical algorithms and require; standard meteorological data as forcing factors. The model package; was validated with respect to available soil water storage using standard records of the agrometeorological station at Petrinka from 1955 to 1988 (forest-steppe zone, the Kursk region, Russia) and observational data for Central-Chemozem State forest-steppe reserve from 1955 to 1973 situated not far from the station. Model calculations for different types of the land surface, including fallow, deciduous forest, grassland and agricultural crops, were performed. Simulated values of available water storage in the uppermost 1 m soil layer were verified against measured values. The root mean square errors of the calculations are equal to 25, 30, 33 and 51 kg m −2 for fallow, grassland, crops and forest, respectively. Analysis of the obtained results shows that the package ANDY-SWASH is suitable for calculating annual and interannual dynamics of soil water storage in different agro- and non-forested ecosystems of the forest-steppe zone with satisfactory accuracy.