DOWNSCALING OF A LARGE-SCALE SURFACE TEMPERATURE FIELD FOR THE MOSCOW REGION

E. V. Dmitriev, K. G. Rubinshtein, and A. I. Chavro

The reconstruction of a small-scale field of daily mean surface temperatures for Moscow from the hydrodynamic short-term forecast on a larger scale is considered. For this purpose, a statistical model is developed to solve the inverse problem with a minimum rms error and to estimate an a priori solution error and the reliability of the model. Reanalysis data are used as an approximation of the short-term forecast. The proposed statistical model made it possible to solve the inverse problem with the rms error of about 2.09ºC or to reconstruct 57% of daily surface temperatures at meteorological stations of Moscow. Numerical experiments show that this method is able to detect an extreme situation in the atmosphere and thus to predict situations when the probability of a bad solution is large. The method of the optimal planning of experiments allows the dimension of the input vector of the statistical model to be effectively reduced.