Statistical Modeling of Extreme Precipitation in Summer in the Baikal Region with the Use of the Correlation Theory of Random Fields

Abstract

A probabilistic model of the sums of extreme precipitation was developed for heavy showers covering large areas in the Baikal region. The spatial correlation functions of precipitation fields over one day and the entire rain period were studied. Data on the Iya River basin were used to evaluate the errors in estimating the mean values over a specified contour of extreme precipitation. Errors in the interpolation of precipitation estimates for the Baikal region were evaluated in the absence of observation data with estimation of the errors of the obtained values by Drozdov–Shepelevskii formulas. These errors amounted to about 10–15% and more. The specific features of precipitation field structures were studied with the use of their expansion in series in natural orthogonal functions for one day and over a rain period for different samples: for the entire combination of cases and for samples containing 10 and 30 maximal precipitation totals for each weather station, i.e., for extreme events. It was found that, if the observation data are limited to a range of maximal values, the structure of precipitation fields is simplified and the first 4–5 decomposition components are enough for its description. The obtained results are of importance for forecast problems and for the construction of simulation models of precipitation fields for further use in deterministic runoff simulation.