Font Size: a A A

Remote Sensing Monitoring Of Dynamic Change Of Vegetation Coverage In Reservoir Basin Based On Mixed Pixel Decomposition

Posted on:2022-06-09Degree:MasterType:Thesis
Country:ChinaCandidate:D Y GaoFull Text:PDF
GTID:2480306605955879Subject:Hydraulic engineering
Abstract/Summary:PDF Full Text Request
Vegetation is an important part of the reservoir basin,which has the functions of water and soil conservation,soil consolidation and sand prevention,and runoff reduction.Percentage Vegetation Cover(PVC)is an important indicator reflecting the growth status of vegetation.Therefore,remote sensing monitoring of vegetation coverage in a reservoir basin is of great value for the ecological environment evaluation and soil and water conservation of the basin.This paper uses the Chushandian Reservoir basin in Henan Province as the research area,and the five phases(2006,2011,2013,2017,2020)Landsat TM/OLI remote sensing images from 2006 to 2020 are used as the data source.Purify the pure pixels of Yuan Yuan,and then use the fully constrained least squares linear mixed model and the bilinear mixed model(FM)to quantitatively invert the vegetation coverage in the Chushandian Reservoir watershed of Henan Province,and compare the unmixed models After accuracy,a higher-precision model(bilinear model after purification of pure pixels)was used to analyze the dynamic changes of vegetation coverage in the watershed,and the results are as follows:(1)Extraction of pure pixels based on pure pixel index and Euclidean Square Distance(ESD)According to the distribution characteristics of the ground features in the basin,the type and number of end members are determined.The remote sensing image is subjected to minimum noise separation transformation and pure pixel index calculation,and four pure end members of non-arable land vegetation,cultivated land,water body and bare soil are extracted,and the distance from each pure pixel to the end member spectrum center is calculated by the European square distance,And count the average distance and standard deviation within the end members,and use ?±3? as the confidence interval to achieve the effect of further purifying the pure pixels of each end member.(2)Inversion of vegetation coverage based on bilinear mixed pixel decomposition modelThe pure pixels extracted by the pure pixel index and the purified pixels are respectively applied to the linear mixed model and the bilinear mixed model to compare the unmixing accuracy before and after the purification of the pixels.Among them,the RMSE of the FCLS model was reduced from 0.030 to 0.026,and the RMSE of the FM model was reduced from 0.021 to 0.014.The FM model after end-member purification has the highest inversion accuracy of vegetation coverage,with an overall accuracy of 95.8%.Then,the model was used for unmixing to realize the inversion of the long-term series of vegetation coverage in the Chushandian Reservoir watershed in Henan Province.(3)Analysis on the dynamic change of vegetation coverage in the Chushandian Reservoir watershed of Henan ProvinceThe FM-ESD model was used to invert the vegetation coverage in the five periods in the study area,and the dynamic change analysis of the long-term series of vegetation coverage changes in the Chushandian Reservoir watershed was carried out.The results showed that between 2006 and 20 20,the vegetation coverage in the basin increased from 63.4%to 67.4%.In addition,qualitative and quantitative analysis of vegetation coverage in the Chushandian Reservoir basin was carried out by means of difference analysis,transfer matrix,and overall trend analysis.At the same time,combined with the topography(elevation,slope,aspect),climate(temperature,rainfall)and human activities,the changes in vegetation coverage in the basin are analyzed.The results show that the vegetation coverage is higher in the area with an elevation range of 250350 meters,while the vegetation coverage is the lowest in the range of 69-150 meters than the overall level.The vegetation coverage in the area with a slope of 0-5°is at a relatively low level as a whole,while the overall level of vegetation coverage in the area with a slope of 10-30.80 is relatively high.At the same time,the vegetation coverage in the basin also fluctuates with the change of slope.The vegetation coverage is highest in the area with the south slope,fluctuating between 0.71 and 0.78,followed by the southeast and east slopes,while the west and northwest slopes have lower vegetation coverage.The area change range of vegetation coverage of different grades also decreases with the increase of elevation and slope.In addition,temperature,rainfall and human factors also have a certain impact on the change of vegetation coverage.
Keywords/Search Tags:Vegetation Coverage, Mixed Pixel Decomposition, Euclidean Square Distance, Dynamic Change
PDF Full Text Request
Related items