摘要
文章基于2023年兴安盟地区国家级气象台站的逐日气象观测资料,与中央气象台下发的国家级智能网格预报产品进行对比检验,并利用机器学习方法探索归纳订正方法,得出结论。结合CMA-GFS数值预报模式结果以及各类地面观测实况,通过集成学习方法建立了温度产品订正模型。该模型在最高气温和最低气温的订正上均表现出显著效果,订正后准确率显著提高,误差明显降低。该订正方法具有较高的研究价值和实际应用意义。
Based on the daily meteorological observation data of the national-level meteorological stations in Xing'an League area in 2023,this paper conducts a comparative verification with the national-level intelligent grid forecast products issued by the Central Meteorological Observatory,and explores and summarizes the correction method by using Machine Learning method,then comes to conclusions.Combining the results of the CMA-GFS numerical forecast model and various ground observation facts,a temperature product correction model is established by an integrated learning method.The model can achieve excellent correction effects on both the highest and lowest temperatures,and the accuracy rate increases and the error is reduced significantly after correction.This correction method has good research value and practical application significance.
作者
李雪雪
LI Xuexue(Meteorological Bureau of Keyouqian Banner,Xing'an League 137400,China)
出处
《现代信息科技》
2025年第4期80-86,共7页
Modern Information Technology
基金
内蒙古自治区气象局科技创新项目(nmqxkjcx202412)。
关键词
智能网格
预报检验
数值预报修订
机器学习
intelligent grid
forecast verification
numerical forecast correction
Machine Learning
