The article presents a set of hierarchical models that allow simulating, extending and restoring series of average daily and average monthly water levels of small and medium-sized lakes in the North-West region of Russia. The objects under study are lakes of the Baltic, White and Barents Sea basins. One of the key problems is the lack of initial hydrometeorological information in the region under consideration. This issue is resolved using the author's program and expert analysis, which allow selecting representative meteorological stations. The article searches for alternative predictors of potential prognostic regression models. This search is based on multivariate statistics methods and the water balance approach. As a result of combining both approaches, predictors were identified that allow replacing insufficiently accurate data on atmospheric precipitation with such characteristics as: relative air humidity, water vapor saturation deficit. The models obtained in the article allow simulating the level regime of lakes with an optimal lead time of 3 days for models constructed using daily discrete data, and 1 month for models constructed using monthly data. In order to adapt the obtained models to changing hydrometeorological conditions, the existence of a relationship between the parameters of cyclonic activity and the water levels of the studied lakes was investigated. The establishment of such a relationship made it possible to proceed to the determination of numerical values of meteorological characteristics that are an indicator of the passage of a cyclone. Atmospheric pressure and relative humidity became such indicative and accessible characteristics.
Anna Batmazova (Wed,) studied this question.