Remote Sensing Estimation And Spatial And Temporal Distribution Characteristics Of Forest Biomass In The Jianghuai Watershed Area

Forest biomass is an important indicator that directly reflects the productivity and carbon sequestration capacity of forest ecosystems,and an important parameter for calculating the carbon sink and carbon cycle of terrestrial ecosystems.The Jianghuai Watershed area is an important forestry production area in the Middle and Lower Reaches of the Yangtze River.The study of forest biomass in this region is an important basis for monitoring carbon dynamics in this region.In this paper,the Huangfu Mountain Forest Farm,the most representative primary secondary forest in the Jianghuai Watershed area,is used as the research area.Aimed at the problems of poor estimation accuracy of biomass in complex canopy structure and high canopy closure forest areas,and the insufficient applicability of forest biomass estimation models in complex forest areas.Combined with the satellite data of Gaofen-1（GF-1）and Gaofen-6（GF-6）,combined with auxiliary data such as Digital Terrain Elevation Model（DEM）and forest resource data,the spectral information,texture features,and main features of high-resolution images are extracted.Component analysis and other characteristics.Established based on different tree species（masson pine,slash pine,sawtooth oak and multiple tree species）,different data（GF-1,GF-6,GF-1 and GF-6）and different estimation methods（multiple stepwise regression,BP neural network,Random Forest）biomass estimation model,and carried out inversion mapping of biomass and analyzed its temporal and spatial characteristics.The research results are as follows,（1）Better forest biomass estimation results can be obtained based on the high score data.The R² range of the GF-1 satellite data estimation result is 0.346～0.541,RMSE range is 12.53～42.17t/hm²,and the R² range of the GF-6 satellite data estimation results is 0.277～0.601 and RMSE range is 13.23～33.34 t/hm².The combination of the two data significantly improves the accuracy of biomass estimation,with R² ranging from 0.384 to 0.651 and RMSE ranging from 13.10 to29.81 t/hm².（2）Based on the GF-1 satellite date and GF-6 joint data to the random forest model was the most effective,with R² of 0.904 and RMSE=10.521 t·hm^-2;followed by BP neural network,with R² of 0.825 and RMSE=18.641 t·hm^-2;the effect of multiple stepwise regression is the worst,R²of 0.528 and RSME=28.011 t·hm^-2.（3）The characteristics of the biomass distribution showed that the central hilly areas of the forest areas is higher than the peripheral plain areas,and the biomass growth rate in 2018-2020 is higher than that in 2012-2018,with the maximum biomass in 2018 and 2020 are 246.741 t/hm²,265.614 t/hm²,and the biomass per unit area is 22.494 t/hm² and 24.611 t/hm²,the biomass is the most widely distributed in the interval of 50-75t/hm²,accounting for 34.88%of the entire forest area.Among them,the forest biomass of plain terrain,flat slope and no-slope woodland had the highest proportion,reaching 79.50%,86.41%and 30%,respectively.Forest biomass and meteorological elements in Huangfu Mountain all had strong correlations,with the highest correlation of 0.214（p<0.01）with sunshine duration,followed by 0.210 for maximum temperature and 0.203（p<0.01）for average temperature,and a negative correlation of-0.141（p<0.01）with average minimum temperature.