以机载数据替代卫星影像的方式提高森林生物量估测精度,是目前林业遥感研究的重点领域.以高山松(Pinus densata)林为研究对象,进行无人机(Unmanned Aerial Vehicle,UAV)多光谱影像数据采集;结合36块样地实测数据,采用变异函数确定高山松地上生物量最佳观测窗口;提取并筛选出相关性较强的5个因子,分别建立PLS和RF模型,对飞行区高山松地上生物量进行估测.结果表明,高山松地上生物量最佳观测窗口为球状模型的变程α值5.2 m;RF模型(R2=0.90、RMSE=17.96 t/hm2、P=84.98%)优于PLS模型(R2=0.55、RMSE=38.94 t/hm2、P=71.13%);基于RF模型,飞行区高山松地上生物量均值为130.48 t/hm2,总生物量为8 343.53 t.
Estimation of Pinus densata Above-ground Biomass Based on UAV Multi-spectrum
Using airborne data instead of satellite images to improve accuracy of forest biomass estimation is key field of forestry remote sensing research.UAV multi-spectral image data acquisition was carried out in Pinus densata forests.Based on measured data of 36 sample sites,the best observation window for P.densata above-ground biomass was determined by variance function.Five factors with strong correlations were extract-ed and screened,and PLS and RF models were established respectively to estimate P.densata above-ground biomass in flight area.Results showed that the best observation window for P.densata above-ground biomass was variable range α value of 5.2 m in spherical model.RF model(R2=0.90,RMSE=17.96 t/hm2,P=84.98%)was superior to PLS model(R2=0.55,RMSE=38.94 t/hm2,P=71.13%).Based on RF model,average P.densata above-ground biomass in flight area was 130.48 t/hm2,and total biomass was 8 343.53 t.
国家科技期刊平台
NSTL
万方数据
