Simulating Study Of Population Dynamics Considering Future Industrial Land Changes

Population data can reflect the geospatial distribution of population within a country or region,and is also an important data source for reflecting the urban development and socioeconomic status.Traditional population data is usually obtained through census and sampling survey,which has limitations such as poor timeliness and difficulty in integrating with other geospatial data.Population spatialization is an effective method for modelling the spatial distribution of population,allowing for a higher temporal and spatial resolution of population data.However,this kind of methods usually used a ‘top-down’ modeling idea when modelling the spatial distribution of the population and the simulation results do not uncover the microlevel individual behavioral process,which is a "bottom-up" process.Agent-Based Model(ABM)can simulate macro phenomena of human population spatial distribution from "bottom-up" by simulating the decision-making behaviors of agents at the micro level.In terms of the ABM model,the determination of the parameters of agent’ decision behavior is essential.Agent’ decision behavior parameters obtained based on intelligent optimization algorithms can improve model accuracy,however,few studies have employed this method to population spatial distribution simulation.In addition,geospatial big data provides a new perspective for ABMbased population simulation studies,but none of the existing studies employs the geospatial big data to characterize the model environment when using ABM model to simulate population residential decision behavior.Therefore,to fill these gaps,this study constructed an ABM model to simulate the spatial distribution of population in Dongguan City,an important manufacturing metropolis in Guangdong Province.In the constructed model,based on the labor economics,a labor market was constructed,and points of interest(POI)and building data obtained from the third national land survey of Dongguan City were used to finely characterize the ABM model environment.During the simulation,the equilibrium of labor supply and labor demand of labor market in the real world could be simulated.In addition,the actual population survey big data and genetic algorithm(GA)were integrated to automatically calibrated the parameters of agent’s residential decision behaviors.After that,the GA calibrated ABM(GA-ABM)and the expert experience based ABM(EEB-ABM)models were used to simulate the spatial distribution of population in Dongguan in 2019.Secondly,individual employment-residential decision behavior has important impacts on the spatial distribution of population.In manufacturing metropolises,the development of industrial land affects the individual employment choice,and individual employment choice have important impact on residential decision behavior,which ultimately influence the spatial distribution of population.However,none of the existing studies took the impact of future industrial land changes on individual decision behavior into consideration when simulating spatial distribution of future population.Therefore,this study coupled the FLUS model,Markov model,and SD model to simulate industrial land changes of Dongguan in 2030 under the historical development scenario(HD),rail transit-leading scenario(RTL)and innovationdriven scenario(ID)and analyzed the spatial expansion of different types of industrial land under different development scenarios,so that to provide an environmental data source for predicting spatial distribution of population in Dongguan City under different development scenarios.Finally,this study combined the simulation result of the future industrial land and the ABM model to carry out the multi-scenario simulation of the spatial distribution of population considering the future industrial land use changes.The main findings of this study are as follows:(1)Compared with the agent’ residential decision behavior parameters before calibration,the residential decision behavior parameters calibrated by genetic algorithm can better reflect the heterogeneity of agent’s residential decision-making preferences.The GAABM calibrated through genetic algorithms has higher simulation accuracy than the EBB-ABM at different scales.With the decrease of the spatial scale,the relative error gap between the simulation results of GA-ABM and EBB-ABM gradually increases.(2)The FLUS model and SD model constructed in this study have high accuracy.According to the analysis results of the driving factors of land use changes,the development of primary industry land in Dongguan is mainly influenced by natural elements,the expansion of secondary industry land is related to natural elements and location elements,and the expansion of tertiary industry land is closely related to location elements and human social and economic activities.In the future simulation of industrial land under different scenarios,the primary industry land under the historical development scenario will be greatly reduced,and the expansion of secondary and tertiary industry land will be around existing industrial land.Besides,the planning of railway stations under the RTL scenario will promote the expansion of tertiary industry land;the planning of innovation centers under the ID scenario will promote the expansion of secondary industry land.(3)The simulation results of the future spatial distribution of population show that the future population size of Dongguan under the RTL scenario and the ID scenario will significantly increase compared with the HD scenario.In terms of spatial distribution of population,there will be a significant expansion of the population spatial distribution in the eastern part of Dongguan in 2030,but the central and western region of Dongguan will still be areas of concentrated population distribution.At the same time,this study found that the planning of future rail transit stations and innovation centers can boost the growth of regional population,while considering the effect of future industrial land use changes is helpful to better reveal the changing pattern of population spatial distribution under different scenarios.