| The Engebei Ecological Demonstration Area is a typical ecologically fragile area,with serious soil erosion and great human disturbance.Research on this area can provide a reference for the scientific management of the area’s landscape and the improvement of ecological security,and at the same time provide a technical basis for evaluation of desert governance and ecosystem restoration along the Yellow River Basin.Based on this,this thesis takes the Engebei Ecological Demonstration Area as the research area,interprets the four-phase remote sensing data of the area in 2005,2010,2016 and 2021,obtains the landscape classification map,and analyzes the landscape pattern changes and ecological risks in the area.research.First,the dynamic degree and transition matrix are used to quantitatively describe the change characteristics of the landscape pattern in the demonstration area,and at the same time,the optimal analysis granularity of the area is studied.Conduct in-depth analysis,and use the moving window method to explore landscape heterogeneity,analyze the temporal and spatial variation characteristics of landscape indices,and finally build an ecological risk assessment model to explore the temporal and spatial distribution characteristics of landscape ecological risks in the demonstration area,and to analyze the spatial and temporal characteristics of ecological risks.correlation analysis.The main findings are as follows:(1)During 2005-2021,the area of grassland,other land and water area decreased,while the area of cultivated land,construction land,transportation land and forest land increased.Grassland and other land were always the dominant landscape in the study area;The change rate of transportation land is the fastest,reaching 28.54%;the dynamic degree of comprehensive land use first decreased and then increased,reaching the highest value of 2.16% in 2016-2021;the transfer of other land area and the transfer of cultivated land area are the land use in the study area.The main form of change;in the analysis of spatial heterogeneity of landscape pattern,each index showed obvious change characteristics.(2)From the landscape scale effect analysis,it is concluded that 50-70 m is the suitable grain size domain for the study area.Based on the suitable grain size domain,combined with the landscape area loss evaluation method,the optimal analysis grain size value is50 m.(3)At the level of landscape type,from 2005 to 2021,the shape of grassland tends to be complex,the degree of fragmentation increases,and the connectivity is weakened;the edge shape of cultivated land is more complex,and the internal connectivity is higher;the shape complexity of woodland increases first and then decreases,and the fragmentation The shape of construction land and water area tends to be regular and agglomerated;the connectivity of transportation land is enhanced;the fragmentation of other land tends to be balanced,and the shape becomes more and more regular.At the landscape level,the fragmentation of the overall landscape increased,and the degree of fragmentation intensified after 2016.The distribution of each landscape type became more and more uniform,the landscape richness increased,and the dominance decreased.(4)From 2005 to 2021,the spatial distribution characteristics of the landscape ecological risk level in the study area are high in the middle,gradually decreasing in the north-south direction,and the low-risk level area is widely distributed.Lower risk class level.In terms of area change,the areas of low and high ecological risk areas showed a decreasing trend as a whole,while the areas of low,medium and high ecological risk areas showed an overall increasing trend.(5)The global Moran’s I values of the landscape ecological risk values at the four time nodes in the study area are all positive,indicating that the ecological risk values in the study area are positively correlated in the spatial distribution,and the risk values show the characteristics of agglomeration in space.The spatial aggregation patterns of ecological risks in the study area are mainly high-high aggregation and low-low aggregation. |