Research On Simulation Of Land Use Change Based On Neural Cellular Automata

Land use change(LUC)is a fundamental process that reflects the interaction between human activities and the land resources on which they rely.Numerous studies have demonstrated that LUC has a significant impact on regional ecological changes,climate change,agricultural production,urban development,and other critical human-land relationship systems.The use of simulation models to study the evolutionary process and pattern of LUC not only helps to reveal the mechanism of human-land interaction and alleviate the contradiction of human-land relationships but also provides scientific decisions for regional land use and urban planning and plays an important role in promoting regional sustainable development.In recent decades,domestic and foreign scholars have widely used the cellular automata(CA)model in the simulation of LUC because its expression mechanism is very similar to that of the classical geographic process analysis theory,it has powerful and complex computational functions,and it can simulate spatio-temporal dynamic processes.In this research,using Guangzhou and Dongguan as the study areas,two models are constructed to simulate the land use change trends in these cities.The main contents and conclusions of this paper are as follows:1.Simulating urban land use change by integrating logistic regression and a convolutional neural network with cellular automata(LR-CNN-CA)To address the shortcomings of the LR-based cellular automata model in that the obtained transformation rules are heavily influenced by subjective factors and cannot adequately quantify the effects of spatial variables,CNN is integrated into the LR-based cellular automata model to effectively reflect the complex relationships among spatial variables and simulate complex land use changes.The case study shows that the Kappa coefficient and figure of merit(FOM)of the simulation of the proposed model are significantly improved,which verifies that the adoption of LR and CNN-based cellular automata models in land use change simulation can improve the shortcomings of the LR-CA model in land use change simulation and improve the simulation accuracy of land use change.2.Simulating urban land use change with an asynchronous neural cellular automata model(AN-CA)Existing cellular automata models synchronously update the state of land units and ignore the impact of newly developed land units on their surrounding neighbors,which is not consistent with the actual change pattern of urban land and thus affects the performance of CA models.To address these problems,we change the state of land units from synchronous to asynchronous updates based on the LR-CNN-CA model and take into account the impact of newly developed land units on their surrounding neighbors in the next moment to simulate complex land use changes.The case study shows that the simulation of the proposed model have significantly improved Kappa coefficients and FOM indices,verifying that the use of an asynchronous neural cellular automata model in land use change simulation can improve the deficiencies of the traditional cellular automata model,enhance the simulation accuracy of land use change,and reflect the real changes in urban land use patterns.