基于实况资料的Stacking回归模型下游气温预报方法  

Downstream Temperature Forecasting Method of Stacking Regression Model Based on Observation Data

作　　者：邓世有 潘影 DENG Shiyou;PAN Ying(Kaiyang Meteorological Station of Guiyang City,Guizhou Province,Kaiyang 550300,China;Baiyun District Meteorological Station of Guiyang City,Guizhou Province,Baiyun 550014,China)

机构地区：[1]贵州省贵阳市开阳县气象局,贵州开阳550300 [2]贵州省贵阳市白云区气象局,贵州白云550014

出　　处：《山地气象学报》2024年第5期34-40,共7页Journal of Mountain Meteorology

摘　　要：【目的】目前大多数气温预报模型是基于数值预报建立的。这种模型存在一个主要问题,即预测精度完全受数值预报精度的影响,导致预报员过度依赖该模型,缺乏对天气实况资料的认知。【方法】该文利用2013-2022年的贵州省自动气象站资料,在考虑气温上下游的相关性的基础上,使用夏季气温实况资料得到了安顺市西秀区日最高和最低气温与省内其他台站之间相隔24 h的皮尔逊相关系数。然后,利用机器学习块选择了Stacking回归模型,建立本地未来24 h的气温预报方法。【结果】(1)上下游最高和最低气温相关性均通过了0.005的显著性检验,表明西秀区24 h气温变化主要受到上游毕节、大方、播州、开阳和贵阳等地的影响;(2)该文所建立的Stacking回归模型能够很好地预测24 h最高和最低气温的变化趋势,在使用±2℃的温度检验方法下,准确率分别达到了83.7%和93.47%;(3)最高气温的预测准确率低于最低气温,反映出西秀区最高气温预报的难度较高。【结论】该方法能够有效降低对数值模式的过度依赖,同时在预测本地24 h气温时具有较高的准确率、稳定性和普适性。Nowadays,most temperature prediction models are based on numerical prediction.However,there is a major problem with this model,that is,the prediction accuracy is completely affected by the accuracy of numerical prediction,resulting in excessive reliance on the model and lack of awareness of the actual weather data.In order to solve this problem,using the summer temperature data of automatic weather stations in Guizhou Province from 2013 to 2022,this paper obtains the Pearson correlation coefficient with 24 h interval between the daily maximum and minimum temperatures in Xixiu District of Anshun City and other stations on the basis of considering the correlation between the upstream and downstream temperatures.Then,machine learning blocks are used to select the Stacking regression model and a local future 24 h temperature forecast method is established.The results show that:(1)The correlations between maximum and minimum temperatures in the upstream and downstream of Xixiu District all pass the significance test at 0.005 level,indicating that the 24 h temperature change in Xixiu District is mainly affected by the upstream Bijie,Dafang,Bozhou,Kaiyang,Guiyang and other places.(2)The regression model established in this paper can well predict the trends of maximum and minimum air temperatures in 24 h.Under the±2℃temperature test method,the accuracy rates reach 83.7%and 93.47%,respectively.(3)The prediction accuracy of maximum temperature is lower than that of minimum temperature,which also reflects the difficulty of maximum temperature forecast in Xixiu District.Overall,this method can effectively reduce the over-reliance on numerical models,and it has high accuracy,stability and applicability in predicting local 24 h temperature.

关 键 词：相关系数 Stacking回归模型 气温 预报方法 

分 类 号：P457.3[天文地球—大气科学及气象学]