An adding/deleting approach to improve land change modeling: a case study in Qeshm Island, Iran

Land use/cover change (LUCC) simulation models are helpful tools for decision makers because of their capacity of predicting the landscape dynamics under various scenarios and thereby developing countermeasures. Developing LUCC models with high reliability still remains challenging due to complicated influencing of natural and anthropogenic factors. An adding/deleting approach is proposed in this study to explore whether and to what extent it can improve the accuracy of a hybrid LUCC model involved with cellular automata, Markov chain, and artificial neural network in the Qeshm Island, the biggest island in the Persian Gulf. The accuracy assessment was conducted by comparing the simulation results obtained from the models with the maps derived from Landsat image in 2014. The results revealed that the adding/deleting approach could improve the prediction accuracy of the model for the majority of land use classes as the area of the correctly predicted classes increased to 7.2 km2, which is greater than 6.09 km2 without using the approach. We further compared the results derived from the proposed approach with those from cellular automata-Markov chain-artificial neural network, Markov chain-artificial neural network, and cellular automata-Markov chain-logistic regression, resulting in the Figure of Merit index of 7.8 with this approach, compared to 6.7, 5.1, and 4.5 with the other three models mentioned above. This study demonstrates that the proposed approach is effective for improving the performance of LUCC modeling.