ASSESSING PREDICTABILITY OF HYDROLOGICAL PROCESSES (ON THE EXAMPLE OF FROZEN SOIL WATER CONTENT DYNAMICS)

Abstract. A method has been developed for assessing the limits of predictability of the frozen soil water content (according to observations at the Nizhnedevitskaya water balance station). The method is based on the analysis of the convergence of a given probabilistic measure (the variance of the calculated soil water content at a given date) to its stable value. The soil water content was simulated by the physically based model of heat and water transfer in a frozen soil column during a autumn-winter seasons. To assess variability of the modelled soil water content at a given date, the boundary meteorological conditions for the autumn-winter period were simulated by the Monte Carlo procedure using a stochastic  weather generator. The initial conditions were assigned as the constant soil temperature and soil moisture values over the
1-meter soil column. The predictability of the soil water content in the one-meter layer of the studied soils has occurred to be about 1.5 months; it means that for the forest-steppe conditions, the soil water content before the beginning of soil freezing cannot serve as an indicator of soil water content before spring. Numerical experiments have shown that the soil water content predictability: (1) grows with an increase in the thickness of the considered soil layer and its depth; (2) decreases for coarser soils as compared to finely dispersed soils; (3) is more sensitive to changes in the soil texture than to changes in the climatic norms of precipitation and air temperature