应用UAVSAR数据及改进极化水云模型对热带森林地上生物量反演  

Application UAVSAR Data and Improved Polarized Water Cloud Model for Aboveground Biomass Inversion in Tropical Forests

作　　者：段云芳 罗洪斌 岳彩荣[1] 罗广飞 王宁 余琼芬 郭喜龙 孙妙琦 Duan Yunfang;Luo Hongbin;Yue Cairong;Luo Guangfei;Wang Ning;Yu Qiongfen;Guo Xilong;Sun Miaoqi(Southwest Forestry University,Kunming 650224,P.R.China)

机构地区：[1]西南林业大学,昆明650224

出　　处：《东北林业大学学报》2024年第1期54-60,共7页Journal of Northeast Forestry University

基　　金：云南省教育厅项目(2018JS330);国家自然科学基金项目(42061072);云南省重大科技专项(202002AA100007-015)。

摘　　要：针对热带森林地上生物量遥感估测中容易饱和区域大尺度森林地上生物量估测精度低的问题,以非洲加蓬中部的洛佩达国家公园为研究区,以NASA提供的L波段全极化机载合成孔径雷达(SAR)数据和LiDAR网格化森林地上生物量产品为数据源开展森林生物量估测方法研究。采用极化分解方法提取森林的多种散射机制,从中选择反映森林结构差异的地面散射特征和森林体散射的特征构建体-地散射比,采用极化水云模型(PWCM)进行森林地上生物量反演和精度评价。为了提高PWCM模型的适应性,建模过程中根据体散射分量(V_(ol))分段进行模型的参数优化。结果表明:以Freeman三分量极化分解后得到的体散射(V_(ol))、表面散射(O_(dd))、地-干散射(D_(bl))为基础构建的6个体-地散射比在极化水云模型估算森林地上生物量中,以μ_(VG2)作为体-地散射比时估测效果最好,模型决定系数(R^(2))为0.60,均方根误差(R_(MSE))为127.78 Mg/hm^(2);在此基础上,进一步根据体散射分量分段优化极化水云模型,模型决定系数(R^(2))增加到0.74,均方根误差(R_(MSE))降低了约20%,预测精度从50.76%提升至60.28%,并改善了低值高估、高值低估问题,在地上生物量高达450 Mg/hm^(2)时未出现饱和现象。The problem of low accuracy of forest aboveground biomass estimation at the regional scale,which is easy to be saturated in the remote sensing estimation of tropical forest aboveground biomass,taking the Lopeda National Park in central Gabon,Africa as the study area,using NASA’s L-band polarimetric airborne SAR data and LiDAR gridded forest aboveground biomass products as data sources to carry out a research of forest biomass estimation method.Firstly,the polarimetric decomposition method was used to extract various scattering mechanisms of the forest,and the ground scattering characteristics and volume scattering characteristics of the forest that best reflect the differences in forest structure were selected to construct the volume-ground scattering ratio.Based on this,the polarimetric water cloud model(PWCM)was used for forest aboveground biomass inversion and accuracy evaluation.In order to improve the adaptability of the PWCM model,the model parameters were optimized according to the segmentation of the volume scattering component(V_(ol))during the modeling process.The results showed that among the six volume-ground scattering ratios constructed based on the volume scattering(V_(ol)),surface scattering(O_(dd)),and ground-dry scattering(D_(bl))obtained after Freeman’s three-component polarimetric decomposition,the best estimation effect was achieved when usingμVG2 as the volume-ground scattering ratio.The model coefficient of determination(R^(2))was 0.60,and the root mean square error(R_(MSE))was 127.78 Mg/hm^(2).Furthermore,based on this,the polarimetric water cloud model was further optimized according to the segmentation of the volume scattering component,and the coefficient of determination(R^(2))increased to 0.74,the root mean square error(R_(MSE))decreased by about 20%,and the prediction accuracy increased from 50.76% to 60.28%.The problem of overestimation at low values and under estimation at high values was improved,and there was no saturation phenomenon even when the aboveground biomass reached

关 键 词：机载SAR 森林地上生物量 极化水云模型 体地散射比 体散射分量分段 

分 类 号：S771.8[农业科学—森林工程]