Simulation Study On The Dynamic Change Of Wetland Landscape Pattern In Xiaoxing'an Mountains

With the growth of population and the rapid development of economy and the acceleration of urbanization process,the way of land use has led to changes in the natural ecological landscape of the surface.The contradiction between people and land has caused a series of environmental problems,and at the same time it affects our life and health.In order to build a resource-saving and environment-friendly society,we have paid more attention to the degradation and protection of wetland resources in recent years.Based on the analysis of the spatial distribution characteristics of wetland landscape pattern and the driving factors affecting the evolution of wetland landscape pattern,the future wetland landscape pattern in the study area was predicted and simulated,and its temporal and spatial evolution law was analyzed,which provided a scientific basis for the construction of wetland ecological protection.In this paper,Xiao Xing'an Mountains area is taken as the research area.Landsat TM data of 1995,2005 and 2015 are used to interpret and extract the classification information of wetland landscape pattern,DEM,slope and other topographic data,meteorological data such as annual precipitation and annual average temperature,raster data such as population density,per capita GDP and basic geographic information of the study area.The data source is equal vector data.The spatial distribution characteristics of wetland landscape types were analyzed by single variable local spatial autocorrelation method,and the correlation between wetland landscape types and driving factors was analyzed by bivariate spatial autocorrelation method.Using FLUS model,the conversion probability of wetland landscape types was extracted based on 2005 wetland landscape classification data and driving factors based on ANN,and the wetland landscape in 2015 was simulated.Based on the landscape classification data in 2015,the changes of wetland landscape pattern in2025 and 2035 were simulated and predicted,and the changing trend of wetland landscape pattern in 1995-2035 was analyzed.It provides a theoretical reference for the future ecological protection and construction in Xiaoxing'an Mountains.The main conclusions are as follows:(1)The spatial distribution characteristics of different types of wetland landscapepattern in Xiaoxing'an Mountains were studied by spatial autocorrelation and single variable local autocorrelation method,and the correlation between the driving factors and the change of wetland landscape pattern was analyzed by bivariate spatial autocorrelation.The results show that all landscape types in Xiao Xing'an Mountains are positively correlated in spatial position,and the degree of correlation is forest land,wetland,arable land,urban and rural industrial and mining residential land,unused land and water.The spatial distribution of landscape types is mainly forest land,arable land is concentrated in the South and middle of Xiao Xing'an Mountains,and the rest are scattered.Wetlands are mostly distributed in the middle of Xiao Xing'an Mountains.Precipitation in climatic factors is the most direct,followed by annual average temperature;slope and DEM in topographic factors are positively correlated in space,and indirectly affect wetland area change by affecting local microclimate;population density and per capita GDP are negatively correlated in space.With the growth of population and economy,wetland area increases rapidly.There was a negative trend of growth,and the distance from road and from town was positively correlated with traffic location,which directly affected the distribution of wetland landscape pattern.(2)Based on the correlation analysis of wetland landscape pattern by the above driving factors,using ANN probability calculation module in FLUS model and starting data in 2005,input driving factor data,and calculate the probability of each landscape type appearing on each pixel in the study area.Then,the distribution of landscape types in 2015 is simulated by cellular automata with adaptive inertia mechanism.By inputting suitability probability data and protected area data with no change in constraints,and setting and adjusting simulation parameters through many experiments,the Kappa coefficient and Fom index reach 0.78 and 8.99%,and the accuracy is verified by comparing pixels.The results show that the reliability of the simulated and real raster maps in the study area is very high.(3)Based on Markov chain to predict the number of grid of wetland landscape pattern in the future,the trained FLUS model was used to simulate the landscape types in 2025 and 2035,and the simulated grid results were converted into vectors to calculate the change trend of wetland from 1995 to 2035.The results show that the development trend of wetlands in the future is slow growth and tends to be stable,which is closely related to global warming,reduction of precipitation,regionaleconomic development planning and increasing awareness of wetlands protection.