西北工程技术学报2025，Vol.24Issue(2)：130-136,7.
基于机器学习的宁夏地区地表温度反演
Research on Land Surface Temperature Retrieval in Ningxia Region Based on Machine Learning
摘要
Abstract
Aiming at the demand for high-precision retrieval of land surface temperature in the Ningxia Region,this study utilizes brightness temperature data derived from the thermal infrared bands of the MODIS satellite,combined with auxiliary data,to carry out surface temperature retrieval in the Ningxia region.Five machine learning models suited for regression tasks—Gradient Boosting Decision Trees(GBDT),Extreme Gradient Boosting(XGBoost),Light Gradient Boosting Machine(Light GBM),Support Vector Regression(SVR),and Random Forest(RF)—are employed for the retrieval study.Model performance is evaluated using multiple metrics,and the optimal model is selected to analyze the retrieval results for land surface temperature in Ningxia.The results show that the XGBoost model exhibits higher accuracy and robustness,with a root mean square error of 1.22 K and a mean absolute error of 0.91 K in cross-validation with MODIS land surface temperature products;ground-based validation yielded a root mean square error of 2.07 K and a mean absolute error of 1.47 K.Additionally,seasonal analysis indicates that the XGBoost model performs better in autumn and spring compared to summer and winter.This research not only confirms the superiority of XGBoost in land surface temperature retrieval but also provides a reliable method for regional-scale temperature estimation,offering significant reference value for related research.关键词
地表温度反演/机器学习/热红外遥感/MODIS/宁夏Key words
land surface temperature retrieval/machine learning/thermal infrared remote sensing/MODIS/Ningxia分类
信息技术与安全科学引用本文复制引用
肖柳瑞,毛克彪,郭中华,代旺..基于机器学习的宁夏地区地表温度反演[J].西北工程技术学报,2025,24(2):130-136,7.基金项目
中央级公益性科研院所基本科研业务费专项(Y2025YC86) （Y2025YC86）
宁夏科技厅自然科学基金重点项目(2024AC02032) （2024AC02032）
