摘要
利用哨兵影像、数字地形数据及森林实地样方调查数据,分别构建K-近邻(KNN)模型、随机森林(RF)模型、极值梯度增强(XGBboost)模型、Stacking模型,实现对黄河三角洲人工刺槐(Robinia pseudoacacia)林生物量的估算。结果表明,相较于K-近邻模型、随机森林模型、极值梯度增强模型,集成学习Stacking模型明显提高了生物量估测的精度(R2=0.61、RMSE=13.42 t/hm2)。
Using sentinel images,digital terrain data and forest field quadrat survey data,K-nearest neighbor(KNN)model,random forest(RF)model,extreme gradient enhancement(XGBboost)model and Stacking model were constructed respectively to estimate the biomass of artificial Robbin pseudoacacia forest in Yellow River Delta.The results showed that the integrated learning Stacking model significantly improved the accuracy of biomass estimation compared with K-nearest neighbor model,random forest model,and extreme gradient enhancement model(R2=0.61,RMSE=13.42 t/hm2).
作者
汪逸聪
WANG Yi-cong(College of Hydrology and Water Resources,Hohai University,Nanjing 210098,China)
出处
《湖北农业科学》
2023年第7期143-148,176,共7页
Hubei Agricultural Sciences
基金
国家自然科学基金面上项目(41471419,31971579)。