基于CUACE模式产品的订正方法比较研究

摘要

分别利用一元线性回归、分季节消除偏差订正和滚动偏差订正的方法,结合实况观测资料和2014年3月-2015年2月CUACE模式产品的评估结果,对2015年3月-2016年2月中国气象局下发的CUACE模式产品进行了订正.结果表明,利用一元线性回归方程订正的效果最差,利用分季节消除偏差订正的方法能降低冬季PM10、SO2和NO2的RMSE、MB和MNB值,对于预报效果有提升作用,但该方法对其他季节的订正效果较差.而利用滚动偏差订正的方法,能够降低SO2和NO2预报值的ERMS、MB和MNB,而且对于除O3外的其他污染物相关系数具有明显的提升作用.%Combining with the observed data and the assessment results of CUACE model products from March 2014 to February 2015,the results of China Meteorological Administration CUACE model products from March 2015 to February 2016 were revised by using simple linear regression,eliminating bias correction and rolling bias correction.Results show that the correction effect of the simple linear regression is the worst,and the method of eliminating bias correction can reduce the RMSE,MB and MNB values of PM10,SO2 and NO2 in winter,which can improve the prediction effect,but not suitable in other seasons.The RMSE,MB and MNB of SO2 and NO2 can be reduced by using the rolling bias correction method,and the correlation coefficient can be obviously improved except O3.