Estimation Of Aboveground Biomass Of Forestland In Daning River Basin

The aboveground biomass of forest land is an important index for the study of forest land resources.It is an important parameter to evaluate the productivity and health status of forest ecosystem,which affects the benefits of forest ecological services.In order to estimate the aboveground biomass of large area forest with high accuracy,it is necessary to make full use of the advantages of multi-source remote sensing data,select the appropriate model,and realize the estimation range extrapolation.The Daning River basin is an important tributary of the Yangtze River.The terrain is mainly high mountains and middle high mountains,with high altitude.Affected by the impoundment of the Three Gorges Dam,Daning River Basin has a certain impact on the ecology and environment of the region,especially vegetation.Therefore,it is necessary to carry out quantitative research on aboveground biomass of woodland in this area,and provide basis for the ecosystem protection of Daning River Basin.Taking the forest land in Daning River Basin as the research object,the sentinel-1a radar data,sentinel-2 multispectral data and DEM data in October 2018were selected as important data sources.Through data preprocessing,the principle of forest height inversion by sentinel-1a radar data was analyzed,and the average tree height of forest land in Daning River Basin was estimated,and the random forest regression model was used to retrieve 2018 Aboveground biomass of woodland.The main conclusions are as follows?1?The average tree height of forest land was retrieved by sentinel-1a radar data.This paper mainly uses backscatter intensity method and dual polarization phase difference method to retrieve the average tree height in Daning River Basin.Compared with the measured tree height,it is found that the average tree height accuracy?R²=0.716,RMSE=7.36 m,MAE=5.554 m?obtained by backscatter intensity inversion is greater than that by dual polarization phase difference inversion?R2=0.642,RMSE=10.105 m,MAE=7.171 m?.?2?Random forest regression model was used to estimate the aboveground biomass of forest land in Daning River Basin.The tree height,sentinel-2 multispectral data and DEM data retrieved from sentinel-1a radar data were used to extract the factors affecting the forest biomass.The correlation between the above ground biomass and the measured forest land biomass was calculated.The boruta feature selection method was used to select the feature set involved in the calculation.The random forest regression model was used to estimate the aboveground biomass of the Daning River Basin Compared with the multivariate regression model,the retrieval accuracy of random forest regression model is higher than that of multiple regression model.The biomass results before and after the average tree height factor was added Through comparison,it is concluded that the inversion accuracy of the model after adding the average tree height factor of forest land is higher;the retrieval ability of random forest regression model is evaluated for different coniferous and broad-leaved stands,and the accuracy of coniferous forest is the highest(R²=0.896,RMSE=6.218 t.hm^-2,MAE=12.740 t.hm^-2),which can be used for the inversion of aboveground biomass in Daning River Basin.?3?The results show that the biomass of woodland in Daning River Basin is high,ranging from 11.49 t.hm^-2 to 246.97 t.hm^-2,with an average value of 193.62 t.hm^-2,showing that the biomass on both sides of the river is low,and the biomass is higher in the mountainous area which is a distance from the river.?4?The results showed that the aboveground biomass of forest land increased with the increase of altitude and decreased with the increase of slope.The farther away from the river,the higher the aboveground biomass of forest land.