Simulation Study On Impervious Surface Expansion In Dianchi Basin Based On Multi-agent And Cellular Automata Model

Land resource is one of the most basic and important resources in nature world and human life,the study of LUCC has long been an important part of the field of global environmental change and climate change,and it is also a hot spot in the current research.At the same time,land use planning and decision-making also play a great role in guiding and restricting the dynamic evolution of land resources.Since the 21 st century,China's has became the second largest economy in the world,both comprehensive strength of national economy,international status and influence have been significantly improved,and the number and scale of cities are also increasing rapidly.China has a large cardinal number of population and a lack of land resources,and the change of land use structure will cause many changes in the ecological process.Therefore,under the guidance of the idea that "Lucid waters and lush mountains are invaluable assets",the research on the dynamic change of land resources is more realistic.To explore the spatial law of land use change and predict the future land use situation is not only conducive to the rational use and optimization of land resources,but also conducive to the formulation of regional land use planning,which has an important impact on the high-quality and sustainable development of China's social economy and ecological environment.In this paper,the simulation and prediction of land use in Dianchi valley is taken as an example.First of all,the five phases of remote sensing image information extraction and classification in 2000-2016 in the study area are carried out by classification and stratification method,and then obtain the land use status in 2000,2006,2009,2013 and 2016.Analysis of the dilation and drive factors of land use and impermeable surfaces.Secondly,using artificial neural network algorithm to construct a cell automaton model based on partition and asynchronous evolution rules to explore the law of land use change.Kappa coefficient is used to verify the accuracy of model simulation results and remote sensing technology information extraction.Meanwhile,the cellular scale and asynchronous speed grid scale that affect the accuracy of the model are discussed and advantages of subarea asynchrony over traditional cellular automata.Finally,under the rule of ant colony algorithm and the decision of land use planning,the scenario of future land use is predicted based on the verified cellular automata model.This study can provide scientific and reasonable reference for adjusting the land use structure,coordinating the relationship between urban construction and environmental protection,and planning and decision-making of land use.On the basis of consulting and collecting a large number of domestic and foreign land resources and environment relations,cellular automata model and biological intelligence algorithm,with the help of ArcGIS,Envi,MatLab and Visual Studio software,combined with the actual land use situation of Dianchi valley,the ANN-CA model suitable for the land use change process of Dianchi valley is constructed by using artificial neural network and cellular automata model.On the basis of the model validation,through ant colony intelligent algorithm and regional planning decisionmaking rules,the land use situation of Dianchi valley from 2000 to 2016 is simulated in four stages,and the land use situation under different scenarios is predicted.Here are the main conclusions :(1)From 2000 to 2016,the area of impervious surface area in Dianchi valley increased year by year,the growth rate was first fast then slow,while the cultivated land decreased year by year,the decrease range was first large then small;The impermeable surface is mainly distributed in Wuhua District and Xishan District in the north of the Dian Lake.It expands around the Dian Lake in a radial way,especially in Guandu District and Chenggong District.The expansion speed of Jinning District and Songming District is slow,with the potential of "Extending to the north and south" and "One lake and Four areas".The areas with rapid land change are concentrated in the areas with low terrain and small slope.It is also a region with relatively developed population,economy and transportation.The trend of land transfer is mainly concentrated in cultivated land,forest land and impermeable surface,and the largest transfer out area is cultivated land,which is mainly transferred to impermeable surface.The land type shift to impervious surface the most,followed by forest land..(2)The traditional cellular automata model based on neural network can well simulate the land use change of Dianchi valley,while the spatial clustering partition and asynchronous cellular automata model based on spatial domain and non spatial domain can significantly improve the simulation accuracy of land use in Dianchi valley.Compared with the traditional spatial clustering algorithm for cell space partition,the spatial domain and non spatial domain clustering partition method not only ensures the cell's compactness in the spatial domain,but also ensures the similarity in the non spatial domain.In addition,the asynchronous evolution rate is used to replace the synchronous rate,and different cells are treated differently,which also highlights the spatial differentiation law of cells.Therefore,the model has higher reliability and applicability in the simulation of land use.(3)Cell scale and spatial scale of asynchronous rate grid have obvious influence on the model.With the increase of cell scale,the simulation accuracy shows a downward trend.The smaller the cell scale is,the more accurate the description of the land type is,and the more rules can be mined by the neural network algorithm to express the actual laws.However,the smaller the cell scale is,the more the number of cells is,and the power function is increased,which reduces the efficiency of model operation.Similarly,the accuracy of simulation results decreases with the increase of spatial scale of rate grid,but the degree of decrease is smaller than that of cell scale.In the practical research,we should according to the actual needs to choose the appropriate cell scale and the spatial scale of the rate grid.(4)Ant colony algorithm rules can better simulate the behavior rules of agents.Under the constraints of land planning,the land use data in 2016 is applied to the model of asynchronous cellular automata.The paper uses the subarea asynchronous cellular automata model to predict the future land use scenarios,which can provide the corresponding scientific basis for the future land use planning and decision-making.