利用2014年9月-2015年8月环境保护部对外发布的兰州市6种污染物实况监测数据,对同时期CUACE模式的24 h预报结果进行误差分析,并通过误差滚动线性回归订正方法进行检验订正.结果表明:(1)兰州市的首要污染物以PM10为主,其次是PM2.5;(2) CUACE模式对SO2的预报及对O3、NO2、PM2.5和PM10预报为2级时,等级预报准确率较高,预报结果可直接使用;(3)模式对O3和CO预报1级时,采用10d误差滚动订正后等级预报准确率可提高1.1％ ～5.5％;(4)模式预报其它要素的其它级别时采用5d或10 d误差滚动订正后再加上或减去一定值后,等级预报准确率可提高8.7％ ～75％.%Based on the observation data of six kinds of pollutants issued by the Ministry of Environmental Protection from September 2014 to August 2015 in Lanzhou of Gansu Province,the deviation of 24 h forecast results of the CUACE model was analyzed firstly,then the test and correction were carried out by using the error rolling linear regression correction method.The results are as follows:(1) The primary pollutant in Lanzhou was PM10 and followed by PM2.5.(2) For SO2,the forecast results of the CUACE model could be used directly.For 03 、NO2 、PM2.5 and PM10,when the forecast values were level 2,the accuracy of forecast results was higher,which could be used directly also.(3) When the forecast values of the CUACE model were level 1 for O3 and CO,the accuracy of level forecast could be increased by 1.1％-5.5 ％ after the deviation correction based on ten-day error rolling linear regression correction method.(4) For other elements,when the forecast values of the CUACE model were other levels,the five-day or ten-day error rolling linear regression correction method was adopted to correct the deviation,then add or subtract a certain value,the accuracy of level forecast would be increased by 8.7％-75％.
展开▼
