Temporal error estimate for statistical downscaling regional meteorological models

Resumen

Statistical downscaling models, which are applied for retrieval of small-scale geophysical fields from largescale fields, allow obtaining a priori estimate of variance of the solution error and some other statistical characteristics of error. However, at given instants or even time periods the solution error of the considered problem can be much higher than its estimate that may be very important when using the results of downscaling. This paper is dedicated to testing the stochastic parameter known as "model reliability" as an indicator of temporal changes of the solution accuracy. For this purpose we considered several basic geophysical applications of downscaling. We show below that the probabilistic parameter "reliability of model" can be used for forecasting time points when the small-scale field is retrieved with high errors.