Abstract
The study of the Aral Sea water level and volume dynamics is an urgent scientific task due to the need to understand the mechanisms of natural, anthropogenic processes. In particular, the study of water balance component dynamics of Small Aral basin is the most important task in planning scenarios of water use, water protection in the region. In the proposed work, on the basis of machine learning methods, two statistical models were developed: a model that takes into account the variability of the monthly values of Syrdaria runoff and corresponding change in Small Aral water volume. In the low availability conditions of data from field observations, obtained operational estimates, which compose water balance component are the most important source of information about ongoing changes in studied basin. The proposed technique can used to obtain initial conditions in hydrodynamic modeling experiments, as well as to calculate climatic scenarios for development of the Aral hydrological system.