Study On Dynamic Monitoring Of Forest Carbon Storage In County Scale

Forest carbon storage is almost estimated through ground survey data,there is a difficulty of large statistic work and modeling complexity.How to calculate forest carbon storage in a way of both quickly and accurately,and finally achieve dynamic monitoring has always been a research focus in the current forest field both in China and abroad.Qingyuan county of Zhejiang province which has a title of the first ecological environment county in China was taken as research object in this thesis,after fusing medium and high resolution remote sensing image,layer classification was adopted to classify the fuse image.Firstly,forest land and other type land were classified from remote sensing data in layer classification;Secondly,tree species land type were divided from forest land;Finally,land type classification result and forest tree species classification result were combed as regional type image.Classification based on support vector machines and fuzzy k-means clustering algorithms was adopted in land use/cover classification,and maximum likihood was used in tree species classification.To dynamic monitoring of forest carbon stock,according to average carbon density of various tree species to map distribution of forest carbon storage in the first step;The carbon storages in 2013 and 2014 were respectively estimated by In VEST(Integrated Valuation of Ecosystem Services and Tradeoffs,InVEST)combining with the average carbon density of forest tree species and the annual growth of carbon storage;Finally,the annual dynamic monitoring of forest carbon storage was realized by using the method based on the carbon storage replacement.In addition,overlaying forest right data,the annual dynamic change of forest carbon storage and carbon storage can be achieved even to every forest land owner.The study results are shown as follows:(1)Layer classification in fusing multi datasource.Compared to single datasource,combining multi datasource has higher classification accuracy.Layer classification architecture can be used in different regions of the image using different algorithms,by improving the local accuracy,improved the overall classification accuracy.Land use/cover classification based on remote support vector machine and fuzzy k-means clustering is an effective computer automatic classification algorithm,which can quickly and accurately classify the remote sensing image into forest land and other land types.The classification method has the characteristics of small sample size data and sample selection.(2)Estimation of above-ground carbon storage based on forst survey data which is mainly based on the dominant tree species in the forst survey data.The average carbon density of the dominant tree species are calculated by selecting appropriate classes data,then building carbon model from average single tree,and the total aboveground carbon storage in the region can be estimated by the average carbon density multiplied by the total area of the dominant tree species in the region.(3)Dynamic monitoring of forest carbon storage based on InVEST mode,the carbon storage module of the InVEST model has the advantages of simple operation and less input variables.The changes of carbon stocks can be obtained by using the method of carbon stock replacement when two periods of forest carbon storage are estimated by the model,then the dynamic monitoring of forest carbon stocks can be realized.The dynamic monitoring of carbon storage at county,township(town)and village level can be done with this method.The result shows that the forest carbon storage of Qingyuan(a county),Songyuan(a town),Kengxi(a villiage)in 2013 was respectively 2.8063 ×107t,2.2279 ×106t,1.8963×105t,and the carbon sink of Qingyuan,Songyuan,Kengxi was increased 5.0314×105t,-3.2036×104t,6.1547×103t from 2013 to 2014.(4)Adding forest right data to the remote sensing image can not only correct the forest right data and coupled the forest right data and remote sensing data,but also can map spatial distribution of forest carbon storage when combining with species average carbon density data.According to forest land ownership relationship,forest carbon storage value of forest right certificate can be estimated,and the personal scale of forest carbon storage monitoring will be achieved.In this thesis,a dynamics monitoring of forest carbon storage in county was proposed,layer classification in fusing multi datasource was adopted.This method combined simple model in InVEST model and low cost,large scale of remote sensing image.The method reflected the changes of forest carbon storage in the spatial distribution amd date series.In the processing of remote sensing image,medium and high resolution remote sensing image,and forest right survey data were combined not only to ensure the accuracy of forest resources spatial distribution,but also reduce the workload of outside survey and the damage to forest ecological resources.