基于随机森林算法的短期降水预测及对农业生产的影响——以庆阳市为例  

Prediction of Short-term Precipitation Based on Random Forest Algorithm and Its Impact on Agricultural Production--With Qingyang City as an Exemplar

作　　者：张思远 王才士[1] 范楠 ZHANG Siyuan;WANG Caishi;FAN Nan

机构地区：[1]西北师范大学,兰州730070

出　　处：《智慧农业导刊》2024年第19期10-13,共4页JOURNAL OF SMART AGRICULTURE

基　　金：国家自然科学基金项目(12261080)。

摘　　要：准确有效地预测降水量有利于农业生产发展的规划、水资源管理以及自然灾害的预防等方面,对于干旱半干旱地区作用更为显著。该文利用庆阳市2023年1月至2024年1月的降水数据,基于包装法中的递归特征消除,迭代移除不重要的特征,后使用随机森林模型对该数据进行分析和预测。结果表明,通过对2种方法的整合使用,能够使模型具有良好的预测性能,且对庆阳市降水时刻与降水量作出较好的预测。该文研究内容对其他地市的降水量预测具有参考价值,也对当地的水资源合理利用以及促进当地社会经济可持续发展具有十分重要的意义。Accurately and effectively predicting precipitation is conducive to the planning of agricultural production development,water resources management and prevention of natural disasters,and is more significant in arid and semi-arid areas.This paper uses the precipitation data of Qingyang City from January 2023 to January 2024,iteratively removes unimportant features based on recursive feature elimination in the packaging method,and then uses the random forest model to analyze and predict the data.The results show that by integrating the two methods,the model can have good prediction performance,and can make a good prediction for precipitation time and precipitation amount in Qingyang City.The research content in this paper also has reference value for precipitation prediction in other cities,and is also of great significance to the rational use of local water resources and the promotion of sustainable local social and economic development.

关 键 词：包装法 递归特征消除 特征选择 随机森林 降水预测 

分 类 号：P457.6[天文地球—大气科学及气象学]