Study Of Land Cover Classification And Spatio-temporal Evolution In Horqin Sandy Land

Under the influence of natural environment change and human activities,the desertification of Horqin sandy land is aggravated,which seriously restricts the sustainable development of local ecological environment.Therefore,the study of sand-fixing vegetation based on remote sensing technology is helpful to curb the desertification process and improve the sustainable utilization of vegetation resources in arid and semi-arid areas as well as the monitoring efficiency of spatial and temporal dynamic changes.The study area is located in the southern margin of Horqin Sandy Land,which has the characteristics of the zone specificity and serious mixed growth of vegetation.In order to explore the availability of hyperspectral remote sensing data in vegetation identification,this paper started from the characteristics of "same thing and different spectrum",and enhanced the separability of the original spectrum of different communities through continue removal and first derivation method,and realized the extraction of identification parameters conductive to classification.Meanwhile,by comparing the classification accuracy of different classification models in multi-spectral images,based on the classification map of high-quality Landsat TM/OLI remote sensing images from 1987 to 2017 and based on the ecological vegetation succession research method,the dynamic evolution rules of vegetation communities in the study area in the past 30 years were systematically analyzed.The main conclusions are summarized as follows:1.According to the phenological characteristics and species composition of the dominant species of Artemisia halodeudrou,four succession stages(i.e.,four communities)were studied by using space instead of time.By analyzing the characteristics of canopy reflectivity of four different Artemisia halodeudrou communities,it is found that with the stability of the communities,the reflectivity of the"Red edge" continued to rise,and the overall curve in September was lower than that in May.The first derivative and continuum removal transformation effectively enhanced the absorption characteristics of the original spectrum and the difference of the red edges between the communities.The optimal identification parameters were the red edge characteristics and the absorption depth,and the recognition effect in September was better than that in May.2.By analyzing the spectral characteristics of different communities,it can be seen that the community will enter the wilting period in September,and the poplar has a long growth cycle,which makes it easy to distinguish from other communities at the red edge.The canopy reflectance of the community of shami-tshapaga is higher at the "Green peak" and "Red valley",but the phenomenon of "Double peak" is not obvious.As the indicator group cluster of the flowing semi-flowing dune,the vegetation composition of the community of Agriophyllum squarrosum-Artemisia halodeudrou community is simple,the leaves are small,and it is easy to distinguish.The width-to-depth ratio can be used to distinguish the communities of Artemisia halodeudrou and Caragana microphylla Lam and Leymus chinensis Tzvel-Phragmites australis community.By comparing the vegetation index characteristics of the five communities,it can be seen that the differences presented by RVI follow the general law of community growth.3.It can be seen from the classification accuracy that the classification accuracy of each model is related to the Decision Tree?SVM?KNN?K-means,in which the overall accuracy of decision tree classification reaches up to 95.24%,kappa coefficient is 0.946,and the high classification accuracy highlights the adaptability of decision tree classification in the study area.4.There is a vegetation structure characteristic of "Semi-shrub-Meadow-Shrub".Combined with the results of dynamic attitude analysis of land use,and Semi-shrub and Meadow are the main vegetation types in the study area.During the period from 1987 to 2017,the major land cover types were still evolving in two directions:forward and backward,and reverse succession accounted for 28.68%of the total evolution.From the perspective of the change sequence,the land cover change in the study area is mainly characterized by the continuous decrease of Semi-shrub and the continuous expansion of Alpine meadow,which reflects the continuous instability and vulnerability of the ecological environment in the study area.The SDE method reflects the accumulation and migration of the center of gravity of different vegetation communities during the process of change.From 1987 to 2017,except for the northward migration of Forest,Meadow and Crop land with great human influence,the center of gravity of other vegetation communities has obvious southward migration.Multi-spectral data can acquire desertification vegetation information timely,accurately and nondestructively.Contributes to the development and management of vegetation resources at regional scale.It is of great significance to the allocation of vegetation resources such as ecological environment.However,hyperspectral data has a great advantage in extracting the index information such as leaf structure,physiological parameters and biomass of near-surface vegetation.It can provide basic data for the analysis and research of multi-spectral data.Meanwhile,in the study of land cover characteristics and space-time pattern of Horqin sandy land based on spectral information and GIS spatial analysis,multi-spectral data realized dynamic monitoring and analysis of vegetation resources.This study can provide an effective reference for establishing the connection between point scale and surface scale remote sensing data and desert vegetation identification.