Abstract:

Using strong cooling data [the lowest temperature decline rate of 24 (48) h was ≥8 (10)℃ and the lowest temperature was ≤4℃] of five meteorological stations in eastern Hexi corridor during 1961 to 2010, climatic characteristics of temporal and spatial distribution and intensity of strong cooling were analyzed systematically by statistical methods. The results showed that strong cooling times were more obvious in high elevation mountainous and northern desert edge than that in oasis plain area. Strong cooling times of age and year had an overall reducing tendency. Strong cooling weather occurred mainly from January to May and from September to December, mostly in April. The variable rate of strong cooling times of each strength was relatively large, and strong cooling times reduced rapidly with the cooling intensity increased. The strong cooling intensity of 24 h showed an overall weak reducing tendency, and that of 48 h showed a weak increasing tendency. Selecting ECMWF numerical forecast grid field data at 20:00 day by day from January to May and from September to December during 2004 to 2013, forecast factors were selected initially and precisely by Press criterion and stepwise regression method. The monthly lowest temperature forecast equations of each region were built with optimal subset regression, and the overall situation and optimal forecast equations were determined finally with the CSC double grading criterion. The forecast equations passed significance level examination of α=0.01. Forecast marginal value and forecast rank of strong cooling were determined with the biggest approaches principle. Forecast fitting rate of 24 and 48 h was 80.0%-83.3% and 81.3%-86.2%, respectively, and the forecast accuracy rate of 24 and 48 h was 71.4%-75.0% and 73.3%-77.1%, respectively, reaching high fitting and forecast level, and could provide objective and effective guiding products for strong cooling forecast and early warning.

Key words: strong cooling, climate characteristics, ECMWF, numerical forecast