基于Landsat 8影像的汝城县森林地上生物量遥感估算研究

Research on remote sensing estimation of aboveground biomass based on Landsat 8 Image in Rucheng County

精确估测森林生物量是分析森林碳动态和碳循环的基础。本研究采用汝城县森林资源连续清查数据,结合Landsat 8遥感影像,分析了森林地上生物量的空间自相关和空间异质性,并选取显著相关的植被指数因子,分别构建普通最小二乘模型、空间滞后模型以及地理加权回归模型,并绘制汝城县森林地上生物量的空间分布图。结果表明:通过对森林地上生物量的空间效应分析,发现样地生物量的空间自相关和空间异质性不容忽视。与普通最小二乘回归相比,空间滞后模型和地理加权回归模型可以减少空间效应对森林地上生物量估测的影响。地理加权回归模型可以最大程度地减少过高或过低估计,估测森林地上生物量的精度最高,决定系数达到0.756,均方根误差和平均相对误差最小,分别为17.288 t·hm-2和-8.542%。因此使用Landsat 8遥感影像结合地理加权回归方法在改善森林地上生物量的估测中具有巨大潜力。

Accurate estimation of forest biomass is the basis for analyzing forest carbon dynamics and the carbon cycle.In this study,the continuous inventory data of forest resources in Rucheng County were combined with Landsat 8 OLI remote sensing images to analyze the spatial autocorrelation and spatial heterogeneity of above ground biomass,and significantly related vegetation index factors were selected to construct ordinary least squares model,spatial lag model and geographic weighted regression model,and then the spatial distribution map of above ground biomass in Rucheng County was drawn.The results showed that the spatial autocorrelation and spatial heterogeneity of the biomass in the plot couldn′t be ignored through the analysis of the spatial effect of the aboveground biomass in the forest.Compared with ordinary least squares regression,the spatial lag model and geographically weighted regression model could reduce the impact of spatial effects on forest aboveground biomass estimation.Geographically weighted regression models can minimize over-or underestimation to the greatest extent.The accuracy of estimating aboveground forest biomass was the highest,the coefficient of determination was 0.756,and the root mean square error and average relative error were the smallest,with 17.288 t·hm-2 and-8.542%respectively,which had great potential for improving the estimation of forest aboveground biomass.