凌汛期地面温度精细化预报方法研究

摘要

Refined temperature forecast plays an important role in ice flood prevention. This paper designed 6 schemes for refined forecast of 3?hourly surface temperature.The modeling data were based on the products of AREM model and observed temperature during ice flood period (November to next March)in 2006—2010. The six schemes were used to forecast the surface temperature of 4 stations surrounding the Yellow River in Inner Mongolia during 2009—2010. With comparison of 6 schemes, the results show that all schemes improve the forecast skill and the best interpretation method is step wise regression. Using the best interpretation scheme,the mean absolute error of 3-72 h fore?cast temperature at the four stations reduces by 1.47℃.The averaged prediction accuracy ratio with error less than 1℃ increases 14.6% and the accuracy with error less than 2℃ increases 26.1%. Zhungeer Station has the best forecast effect,with 1.25℃ mean absolute error,and the accuracy ratios with error less than 1 ℃ and 2 ℃ are 46.5% and 80.2%,respectively.%气温的精细化预报对防凌减灾具有重要意义.基于AREM数值模式预报产品,利用2006—2010年黄河凌汛期(11月至翌年3月)实测温度进行建模,对2009—2010年度凌汛期黄河内蒙古河段周边4个站点的每3 h地面温度进行精细化预报研究.6个精细化预报方案效果比较表明:各预报方案都能改进模式预报效果,效果最好的是逐步回归方法.逐步回归方法对各站3~72 h平均预报的平均绝对误差降低1.47 ℃,1 ℃准确率提高14.6%,2 ℃准确率提高26.1%;改进效果最明显的站点是准格尔站,改进后的72 h预报的平均绝对误差为1.25 ℃,1 ℃准确率为46.5%,2 ℃准确率为80.2%.