Operational long-range weather prediction in the Soviet Union is reviewed. Methods for producing forecasts at the 5- and 10-day, monthly and seasonal range are described in terms of the synoptic, statistical and hydrodynamic tools available to Soviet forecasters. Skill scores for these forecasts published by the Soviets are summarized and examined.
Skill scores for Soviet operational forecasts of mean seasonal (about two months) temperature anomaly and precipitation category are computed separately for regions, seasons and years and compared to persistence skill scores. In addition, forecast-observation sets for the sign of the mean temperature anomaly are tested for “no skill.” The forecasts for the sign of the mean temperature anomaly are found to be best by region for the Arctic and by season for March through April, but generally do not outperform persistence, exhibit demonstrable skill, or show an improvement trend over the verification period. Forecasts of the mean precipitation category are shown to be consistently better than persistence, but to have quite modest skill scores.