最优子集回归在夏季高温极值预报中的应用

Application of optimal subset regression in summer high temperature extremes prediction

利用2003-2007年6-9月ECMWF格点场资料,使用差分法、天气诊断、因子组合等方法构造出能反映本地天气动力学特征的预报因子库.考虑气温变化的连续性,采用Press准则初选因子,用最优子集回归建立6-9月0-72 h≥35℃高温预报方程.2008年预报系统投入业务应用,检验证明预报效果较好,提供了有效的地市级客观预报指导产品.业务系统与MICAPS对接,实现全自动化,输出的预报产品客观、定量.

Based on the June-September date of ECMWF grid field from 2003-2007 and using the difference method,weather diagnosis and the combination of methods that can reflect the structure of the local characteristics of the dynamics of the weather factor,the continuity of the temperature changes was considered and factors were initially elected with the Press criterion.Using the June-September forecast equation of 0-72 h day-to-day extreme maximum temperature and the equation to forecast temperature≥35℃in summer,the forecast system was investigated to understand the service application in 2008 and the examination proved that the forecast effect was better.The effective instruction of objective forecasting of prefecture level was provided.The forecast system was docked with MICAPS and complete automation was realized.The output forecast product was objective and quantitative.