关键词: 沙尘暴, 过程预报, 周期分析, 相似系数

Abstract: With observation data on sandstorm time and stations, and reanalyzed data of 500 hPa NCEP/NCAR from 1954 to 2003 in Gansu province, we analyzed the regional sandstorm characters per five days in spring. Taking seven 500 hPa areas of spring in Gansu province as key studying regions we analyzed the periods of sandstorm weather. There were about 30 pentads (150 days) in these key regions having passed prominence test (α=0.01), which indicated that the 500 hPa geopotential height mean field in spring certainly existed obvious periods about 30 pentads in some extent. The typical fields where regional sandstorm frequently occurred in spring were decided to caculated the similarity coefficients with geopotential height anomaly fields of every pentads in October to December year after year. Result showed that the historical probability in sandstorm accurately forecasting with the \!nearly 30 pentads period method\" was within 71.5%~76.4%, the highest 87.5% and the lowest 60.0%, but the neglecting probability was still at 19.8%~26.0%. Nevertheless, the forecasting method has relatively high capability in forecasting regional sandstorm in Gansu province, especially the sandstorm events corresponding to typical geopotential height fields.

Key words: sandstorm, weather process forecasting, periods analysis, similarity coefficient