Study On Forest Extraction Based On Multi-source Data Collaboration

In this paper,we study the forest land extraction by multi-source data to achieve the purpose of monitoring the dynamic change of forest.The study is significant to maintain and enhance the regional ecosystem service function.It is an important foundation and scientific support for improving the woodland supervision ability,carrying out ecological space planning,fulfilling the ecological benefit compensation,strengthening the woodland protection and utilization management,scientifically predicting vegetation growth.This study laid the foundation for regional economic and social development,urban and rural construction,land use and ecological environment protection.In this paper,the mainly contents areas follows:1?We use GF1,GF2,ZY3 three kinds of high resolution remote sensing data sources to carry out the woodland extraction research of 2015 which based on the knowledge discovery method and visual interpretation correction.First of all,it explores the spectral,texture,shape and remote sensing derived features of high resolution remote sensing data.Secondly,according to the characteristics of knowledge to establish a knowledge database for forest resources extraction.Finally,the study use google earth images and field survey data for accuracy verification.The results show that the accuracy of GF1,GF2,ZY3 and total extraction are 95.1%,94.5%,91.4% and 92% respectively.Kappa coefficients are 0.9,0.89,0.83,0.84 respectively.Indicating that the extraction results are good,and can be used for the next study.2?We carry out the study of biomass estimation.First of all,based on the results of the extraction,the linear regression model between biomass and stock volume is established by combining the hectares of reserves and the mainly tree species information in the forest land change data of 2015 to carry out the estimation of biomass.Secondly,spatial analysis of biomass is carried out by DEM,slope and aspect data.The results show that the distribution area of biomass of woodland is in the range of 500-1000 m above sea level,15-25o slope and 112.5-157.5o aspect.3?We carry out the dynamic analysis of multi-source data from 2011 to 2015.First of all,the forest security data of 2011,the land data of 2013 and the results of 2015 are divided into three periods(2011-2013,2013-2015,2011-2015)for the convergence of three data sources analysis.It is found that the area of woodland decreased by 115.53 hm2 in 2011-2013,increased by 8701.17 hm2 in 2013-2015 and 7546.02 hm2 in 2011-2015.The reasons for the change are mainly divided into human and natural factors.This change in the area is basically realistic.Secondly,the response of vegetation to terrain factors is studied by DEM,slope and aspect data.The results show that the distribution area of the forest land is in the range of 200-1000 m above sea level,8-25o slope and 112.5-202.5o aspect.