Land use change model based on bee colony optimization, Markov chain and a neighborhood decay cellular automata

This study provides a model for the assessment of land use/cover change (LUCC) process in the district 9th of Mashhad, Iran, during 1986-1996. The problem is of high importance due to the rapid development and growth of the city both in size and population. However, the model is applicable for other cities, especially in developing countries. In this model, bee colony optimization (BCO) and cellular automata (CA) are used to manage the spatial distribution of land uses and Markov chain (MC) is applied to model the amount of land use changes. In addition, due to the importance of the cellular neighborhood in CA, a function was added to the model to take the role of neighborhood into consideration. By comparing the simulated map of 1996 with the actual one and by measuring the overall accuracy and kappa coefficient, it is concluded that the model is suitable to simulate LUCC. The proposed model is compared with the BCO-CA model. The results of comparisons illustrated that the proposed BCO-MC-NDCA model outperforms the BCO-CA model in forecasting the spatial distribution and the amount of land use changes in the studied area. The overall accuracy and kappa coefficient, compared with those of the BCO-CA model, were improved by 7.64% and 0.11, respectively. Recommendations for Resource Managers BCO algorithm is presented in this paper for rule learning in the CA framework to model LUCC. In addition, Markov chain is applied to measure and control the amount of land use conversions. Furthermore, a function of neighboring effect is added to take advantage of the distance between cells in LUCC modeling. The experiments showed that the proposed approach is effective in predicting urban changes. The following implications could be realized based on the observations: While CA is a good model for cell conversion in urban maps, integrating that with MC could result in a more precise prediction of land use conversion. Since the neighboring cells are in close relation with each other, use of a neighboring decay effect, as a key factor for managers, is suggested to improve the predictions and decisions, significantly.