Font Size: a A A

Study On Remote Sensing Monitoring Technology Of Multi-source And Multi-scale Shrub Encroachment Into Grassland

Posted on:2024-06-25Degree:MasterType:Thesis
Country:ChinaCandidate:C Q ZhangFull Text:PDF
GTID:2531307121482874Subject:Cartography and Geographic Information System
Abstract/Summary:PDF Full Text Request
The Zheng Xiang Bai Banner in Inner Mongolia is located on the Hunshandake Sandland’s southern edge.It is characterized by a typical arid and semi-arid climate,making it the essential ecological barrier in the Beijing-Tianjin-Hebei region.In recent years,climate change,overgrazing,and other phenomena have accelerated grassland stratification,dramatically impacting the region’s ecological environment and grassland protection.As a significant landscape symbol of grassland degradation,shrub-encroached grassland has significantly pressured the protection of arid ecosystems.Current research on shrub-encroached grassland stratification mainly focuses on its physiological and ecological characteristics.However,using remote sensing technology to identify and monitor crucified grassland is still in the exploration stage.The increasingly abundant global remote sensing data products and domestically produced high-resolution remote sensing data provide strong data support for identifying and monitoring crucified grasslands.Therefore,this study selected the Zheng Xiang Bai Banner,Inner Mongolia Autonomous Region,as the study area.And used high-resolution data from 2020 as the remote sensing data source.Based on the quantitative extraction perspective,the random forest feature selection algorithm was used to evaluate the importance of feature variables.And then,a multi-scale shrub vegetation coverage estimation model was constructed as well.The study then researched remote sensing identification technology for crucified grasslands in 2020 at different scales.Using Landsat data from2000 and 2020 as the remote sensing data source,a simple linear iterative clustering segmentation algorithm was used.Based on the idea of direct object change detection,combined with the random forest classification algorithm to extract the spatial distribution of crucified grasslands in Zheng Xiang Bai Banner in 2000.The study then conducted remote sensing monitoring of grassland stratification in Zheng Xiang Bai Banner over the past two decades to provide technical support for supervising,managing,and rationalizing arid and semi-arid grasslands in northern China.The main conclusions of the study are the following:(1)An efficient and convenient model for estimating shrub vegetation cover was proposed in this study.By utilizing the advantages of multiple data sources and the assistance of the random forest feature selection algorithm,different contributions of multiple data sources at multiple scales were discovered to explain changes in shrub cover.Based on this,four models were used to estimate the multi-scale shrub vegetation cover,including the linear,binary quadratic,Regress,and Random Forest regression models.And the quantitative extraction results were evaluated.It was found that the binary quadratic model had the best performance with R~2 precision evaluation indices of 0.757,0.818,and 0.894 for 3×3,5×5,and 7×7 windows,respectively.The accuracy of the estimation models improved by 5.95%,23.84%,and 20.74%at the three scales compared to the next best model.(2)The spatial distribution of shrub-encroached grassland in Zheng Xiang Bai Banner in 2000 was obtained based on the change detection methods.The superpixel algorithm of simple linear iterative clustering was used for object-oriented segmentation of images in 2000 and 2020,and the"band difference method"was used for change detection.The overall accuracy of change detection was 85.8%,and the distribution of shrub-encroached grassland change areas in Zheng Xiang Bai Banner in2000 was obtained,mainly concentrated in the north and south directions.Random Forest was used to classify the change areas,and the Kappa coefficient,mapping accuracy,user accuracy,and overall accuracy were all above 0.8,which indicated good performance.Combining the unchanged areas,the spatial distribution of shrub-encroached grassland in Zheng Xiang Bai Banner in 2000 was obtained later.(3)Monitoring the grassland shrub encroachment in Zheng Xiang Bai Banner from 2000 to 2020 was achieved.By comparing the spatial distribution of shrub grassland in 2020 and 2000,one of the data collected from the multi-scale is based on the multi-scale shrub vegetation coverage estimation model,and the other one is based on change detection methods.Combining the data obtained from statistical analysis of image change areas and land use transition matrix,the overall degree of shrub encroachment in Zheng Xiang Bai Banner in the past twenty years was determined.The degree of shrub encroachment of the whole Zheng Xiang Bai Banner was lightly scrubbed grassland.However,in the northern sand dunes,the degree of shrub encroachment in the central region was relatively light,while that in the southeast and southwest was heavier.The distribution of shrub-encroached grassland in the hilly areas of southern Zheng Xiang Bai Banner in 2020 expanded slightly compared to that in 2000.From 2000 to 2020,the southern part of Zheng Xiang Bai Banner degraded from non-shrub-encroached grassland to moderate-to-heavy shrub-encroached grassland,indicating a clear phenomenon of grassland shrub encroachment.It was also found that it is possible to reverse from lightly shrub-encroached grassland to non-shrub-encroached grassland.
Keywords/Search Tags:Shrubland encroachment, Multi-source data, Multi-scale, Random Forest, Image segment, Change detection
PDF Full Text Request
Related items