MONTHLY METEOROLOGICAL FORECASTING WITH THE MGO HYDRODYNAMIC-STATISTICAL METHOD. II. PROBABILISTIC FORECAST: ANALYSIS AND INTERPRETATION OF ENSEMBLE DISTRIBUTION, FORECAST CALCULATION AND VERIFICATION

V. M. Mirvis, V. P. Meleshko, V. M. Gavrilina, V. A. Matyugin, and T. Yu. L'vova

Methods of probabilistic forecast calculation and verification are considered by the example of ensemble hydrodynamic forecasts of surface air temperature for a period up to a month based on the MGO atmospheric general circulation model (T42L14). Prognostic ensemble distributions are calculated from 20-year (1979—1998) hindcasts and compared with the daily observational data of 70 stations over the former USSR. (The NCEP/NCAR reanalysis is used to calculate the hindcasts.) Nine-member ensembles are formed by the perturbation of initial fields. It is shown that parameters of the forecast ensemble distribution must be corrected. The scores of probabilistic surface air temperature forecasts are given for three equiprobable categories, including the independent tests in 1999—2005. Examples of potential economic value of forecasts are given.

Joomla templates by a4joomla