统计降尺度方法在北京月尺度预测中的应用

Application of Statistical Downscaling Method to Forecasts in Monthly Scale

利用SDSM(statistical downscaling method)方法对北京47年(1961—2007年)的最低、最高气温和降水变化情况进行模拟评估,在此基础上对2008年北京奥运期间和2009年国庆期间天气变化进行实际预测应用。结果表明,SDSM方法具备模拟气温和降水等要素的能力。从年际变化模拟的情况上看,SDSM模型对气温模拟的效果好于降水,其中对于月平均最低(最高)气温模拟的效果好于最低(最高)气温极值的模拟。模型模拟的逐年极端最高(最低)气温结果在整体上偏低于实况气温,体现出气温极值模拟能力的不足。SDSM模型模拟的降水量整体上小于实测值,对降水极大值模拟能力更弱。对奥运会和国庆期间北京天气预测结果表明,模型对日最高、最低气温和降水的数值预测能力较差,预测值偏低于实际值,但升温和降温过程发生的时段能够准确的预测。

Using a statistical downscaling method（SDSM） we simulated and evaluated the changes of minimum and maximum temperatures,as well as precipitation during 47 years（1961-2007） in Beijing,and on this basis,we predicted the weather during the 2008 Beijing Olympic Games and during the 2009 National Day.The results show that,the SDSM has the ability of simulating temperature and precipitation.From the point of inter-annual simulation variability,the SDSM simulated temperature is better than precipitation, in which the simulated mean minimum（maximum） temperature is better than the simulated lowest （highest） temperature extremes.Meanwhile,the simulated annual maximum（minimum） temperatures are overall lower than the observed indicating there are systematic errors which show less ability to simulate extreme temperatures.The SDSM-simulated precipitation on the whole is less than the measured values, and the large values in the simulation of precipitation are even more serious.Weather forecasts during the Beijing Olympic Games and the National Day indicated that the predicted values for maximum and minimum temperatures,and precipitation are lower than the actual values,but the heating and cooling processes can be accurately predicted by the SDSM.