摘要
根据2004年中尺度数值预报模式MM5输出产品和临沂市环境监测中心逐日监测资料建立了各污染物浓度预报方程,在2005年夏季的试报中,该方程的试报效果较差,其预报准确率明显低于其历史拟合率。为了提高预报准确率,利用逐步回归筛选的因子及统计模型研究中的有关数据,探讨了使用卡尔曼滤波方法制作空气污染物浓度预报的问题。分析发现,利用卡尔曼滤波方法制作空气质量预报可以取得比较满意的效果。
Based on the output data of meso-scale model MM5 and the monitoring data of air during the summer of 2004 in Linyi, a statistical equation was constructed by stepwise-regression to forecast the concentration of air pollutants. Trying the equation during summer of 2005 ,The forecast accuracy of the equation was obviously lower than its fitting rate in history,so the Kalman filtering method was introduced to improve the forecast accuracy with the initial condition given by stepwise-regress. It had been found that the Kalman filtering method could give better forecast results than statistical equation.
出处
《气象科学》
CSCD
北大核心
2007年第1期57-62,共6页
Journal of the Meteorological Sciences
基金
山东省科技发展计划项目(编号:2004GG2208142)
山东省气象局科研课题(编号:2004sdqxz02)共同资助
关键词
MM5模式产品
空气质量预报
逐步回归
卡尔曼滤波
Output data of MM5 Air quality forecast Stepwise-regression Kalman filtering method