The Study On Estimation Technology Of Forest Volume Based On Rapid Eye Imag·

Forest volume as an important indicator of forest resources,the traditional forest resource investigation cycle is long,work intensity,operation is more difficult,urgent need for a quick and easy way to replace the traditional forest resource survey.How to make full use of 3S technology,establish the estimation model of forest volume and improve the prediction accuracy of forest volume is the hot issue of forestry remote sensing science and technology workers.In this study,Yanqing District of Beijing was used as the monitoring area,and the eighth national forest resource survey data and Rapid Eye image and DEM elevation map were used as original data.The measured forest volume of the forest land were selected as the dependent variable,and 15 independent variables were selected according to the Pearson correlation analysis.The 12 independent variables were obtained by Pearson correlation analysis.Index,to form a "optimal" subset of the forest volume estimation model,and to ensure the accuracy and reliability of the model fit.The Remote Sensing estimation model of forest volume based on the principle of mathematical statistics and remote sensing image.The details and results are as follows:(1)The forest volume estimation model was established by stepwise regression method,principal component regression analysis method and partial least squares regression method.Using the same sample to establish a nonparametric KNN(K-Nearest Neighbors)estimation model.(2)Set the comprehensive evaluation index,use the modeling data to compare the advantages and disadvantages of the four methods,and finally select the partial least squares regression method and KNN method to invert the forest volume in Yanqing area.Inversion of forest volume in Yanqing area was carried out by partial least squares regression and KNN method.Among them,the forest volume of Yanqing area was 310.332 million m3 with KNN method,and the estimation accuracy was 82.4%.The estimation of forest volume by the partial least squares regression method is 316.592 million m3,and the overall estimation accuracy is 84.1%.The overall results are still relatively impressive.This indicates that the forest volume estimation technology of Yanqing area based on Rapid Eye image has some reference value for forest resource monitoring and investigation.