County-level Forest Resource Information Extraction And Construction Of Management System

Forest resources are the basis of forestry management.With the rapid development of computer technology,Internet,and remote sensing technology,the combination of forestry and 3S technology is more and more close.The demand of forestry department for extraction and scientific management of forest resource information is also increasing,In order to achieve "one map","one set","one system" monitoring and "one platform" management of national,provincial and county forest resources,mastering the status and changes of forest resources,the effective extraction of forest resource information and the construction of forest resource management system at county level have become very important.This study takes Dongtai city,Jiangsu province as the research area,start from the needs of county-level forest resource information extraction and forest resource management information construction,using the theories and technologies such as remote sensing,image processing,deep learning,geographic information systems,and databases,extracting forest resources information from sub-compartment,forest farms and county-level regions through UAV and satellite remote sensing image,integrating data in multiple formats and build a multi-platform county-level forest resource management system to achieve scientific and effective management forest resources information.The main research results are as follows:(1)The number of trees extraction based on UAV image.By using the differential detector,using the Laplacian of Gaussian(LoG)and the Difference of Gaussian(DoG)perform spot detection on the sub-compartment's processed images of metasequoia and poplar to obtain the number of trees in sub-compartment.The results show that the error rate of the number of trees obtained by the LoG detection in metasequoia sub-compartment is 0.33%,and the DoG detection is 0.49%;the error rate of the number of trees obtained by the LoG detection in poplar sub-compartment is 0.24%,and the DoG detection is 0.48%.It is feasible to use the spot detection algorithm to extract the number of trees based on UAV images,and the accuracy of the LoG detection is higher than the DoG detection.(2)The vegetation information extraction based on UAV image.Using the Visible-band Difference Vegetation Index(VDVI)to extract the vegetation information of Dongtai Forest Farm in Dongtai City,Jiangsu Province,the threshold value is 0.0314,the vegetation extraction accuracy rate is 93.89%,the Kappa coefficient is 0.8522.The vegetation coverage of Dongtai forest farm in August 2016 is 79.65% by pixel ratio calculation(3)Tree species classification based on deep learning.Selecting poplar,metasequoia,and bamboo to classify tree species in the forest farm area,using the U-Net fully convolutional neural network for model training,the overall accuracy of the obtained pixel-based image segmentation model is 93.18%.The model performs inference through the ArcGIS deep learning module to obtain the tree species classification map of the forest farm.(4)Forest change monitoring based on satellite remote sensing image.By analyzing the Normalized Difference Vegetation Index(NDVI)time curve of four land types: forest,farmland,grassland,and wasteland,0.8 is selected as the threshold to distinguish whether the pixel is a forest.Acquiring the NDVI maximum synthetic images of 2015 to 2019 through the Google Earth Engine and perform forest extraction and coverage calculation.The results show that the forest coverage rates of 2015 to 2019 are 27.79%,31.26%,27.90%,29.77%,and 33.93%,that is,the forest resource coverage rate of Dongtai has shown a rising trend in the past five years.(5)Construction and implementation of county-level forest resource management system.Analyzing the user needs and functional needs of county-level forestry authorities,forest management agencies,and field investigators,and carry out feasibility analysis.According to the "Digital Forestry Standards and Norms" and the requirements of county-level forest resource update management,designing the basic geographic database,raster database,forest resource database,and survey update database,using ArcGIS Enterprise platform to build Dongtai county-level forest resource management system,mainly including desktop,web and mobile applications.Realizing user and data management,county-level forest resource data display,query,edit,update and statistical analysis,and other functions.The research shows that the application of remote sensing technology can greatly improve the efficiency and accuracy of forest resource information extraction,UAV image can effectively extract the number of trees and vegetation information and classify tree species;satellite remote sensing image can monitor the change of forest coverage on a large scale;The county-level forest resource management system in Dongtai basically realizes the visual expression,digital management and network sharing of forest resources by integrating the extracted forest parameters and basic forestry information.Providing the full-featured,interconnected and shared,efficient and convenient,stable and safe solutions for county forestry departments to investigate,update,and manage forest resources.