干旱区资源与环境2026，Vol.40Issue(1)：99-111,13.DOI:10.13448/j.cnki.jalre.2026.009
基于随机森林模型的金昌市土壤有机质空间预测方法研究
An approach to the spatial prediction of soil organic matter based on random forest model:A case of Jinchang City
摘要
Abstract
This research mainly focuses on the performance differences between the Random Forest(RF)model and the Ordinary Kriging(OK)model in predicting the spatial distribution of soil organic matter,and analyzes whether adding soil attributes as environmental covarietes can improve the prediction accuracy of the random forest model.Taking Jinchang City,Gansu Province as the research area,vegetation,terrain,and climate indicators as environmental covariates,and combined with soil properties,the RF model is used to spatially predict the soil organic matter content in Jinchang City,and the calculation accuracy is compared with the OK model.The results show that the RF-A model with increased soil attributes(R2=0.58,RMSE=2.947)exhibits better prediction accuracy than the RF-B model with terrain,climate,and vegetation attributes as environmental variables(R2=0.51,RMSE=4.731)and the OK model(R2=0.45,RMSE=4.01).This indicates that adding soil attributes as training variables during model training can improve the accuracy in predicting the spatial distribution of soil organic matter.关键词
随机森林/土壤有机质/金昌市/土壤属性Key words
random forest/soil organic matter/Jinchang City/soil properties分类
农业科技引用本文复制引用
GUO Ruipeng,GAI Aihong,GAO Yanhong,LI Miao,LI Yingying,ZHAO Pengwei,WANG Zishen..基于随机森林模型的金昌市土壤有机质空间预测方法研究[J].干旱区资源与环境,2026,40(1):99-111,13.基金项目
国家自然科学基金项目(42075120) （42075120）
甘肃农业大学科研基金(GSAU-JSFW-2024-187)资助. （GSAU-JSFW-2024-187）
