Landscape pattern prediction method based on ANN-CA-Markov coupling model

The simulation and prediction of landscape patterns can provide a reference for the formulation of ecological space security policies and management and control measures. Therefore, it is necessary to use objective and accurate scientific methods to make accurate simulation predictions. In this study, the quantitative prediction of the Markov model is combined with the advantages of the space-time simulation of the CA model, and the artificial neural network (ANN) module is used for calculation and coupling correction. The Kappa coefficient and the FoM (figure of merit) index are used to test the simulation accuracy. Based on this process, the changes of the landscape pattern in Mudanjiang City at three time points in 2000, 2010 and 2020 were analyzed, and the simulation prediction was made based on this. The experimental results show that the CA-Markov model using the ANN module for coupling correction can be tested for accuracy in the study area, and its Kappa coefficient is 0.834 and the FoM index is 0.001. The model meets the requirements of the landscape pattern for the accuracy of simulation and prediction, and the prediction and analysis of the study area is carried out, and the accurate prediction results are obtained. The results of the study were: Looking at the change in landscape pattern over the 20 years from 2000 to 2020, significant changes occurred in the areas of arable land and artificial surface. The total area of arable land decreased by nearly 130 km², while the area of artificial surface (i.e., built-up land) increased by about 167 km². The forest area initially decreased and then increased, with an overall increase of approximately 56 km², while the water and grassland areas first increased and then decreased. In terms of spatial distribution, the changes in agricultural and forest land are mainly concentrated in the central area where urban development is intensive and in the northern forest area. The results of this study are quite important for policy formulation, according to which it was found that human activities, especially urbanization, have led to a decrease in the area of forested land, and therefore the implementation of the natural forest protection policy can lead to a sustained increase in the area of forested land.