Practical predictability of the standardized precipitation index on monthly and seasonal timescales

Abstract

The SL-AV global semi-Lagrangian model of Hydrometcenter of Russia is used for considering the issues of practical predictability of the Standardized Precipitation Index (SPI) recommended by WMO. The index values are computed using the actual data (observations and reanalysis) taken as a reference and the forecast (hindcast) data interpolated to the stations (236 stations on the CIS territory). The analysis of practical predictability is based on diagnostic verification as well as on the model verification measures recommended by WMO. The statistically significant useful signal was detected on monthly and seasonal integration intervals. No useful information is found for the second- and third-month forecasts. A case study for the Republic of Kazakhstan (July 1989) demonstrates the dependence of forecast skill on the atmospheric circulation patterns. It is revealed that in the case of meridional atmospheric circulation forms the model resolution increases and, in some cases, not only moderate but also severe drought can be predicted.