A statistical model of land use/cover change integrating logistic and linear models: An application to agricultural abandonment

Abstract

Several land use/cover change (LUCC) models have been developed to simulate future LUCC. However, current models work with the assumption that the input non-spatial variables are significant to the LUCC in hand and there is still a lack of model that could identify which non-spatial variables are significant drivers of LUCC. This paper presents a statistical model of LUCC that integrates a logistic model based on spatial drivers and a linear model based on non-spatial drivers. The logistic model produces a probability map that represents local probabilities of LUCC while the linear model produces a global probability threshold that represents a global probability of LUCC, and by comparing the two variables, LUCC is mapped. The statistical model was utilized to model agricultural abandonment in the Ifugao rice terraces, Philippines. Statistical modeling identified the significant spatial and non-spatial drivers of agricultural abandonment in the terraces. Accuracy assessment showed that simulated maps achieved accuracies suitable for LUCC simulation, demonstrating that the statistical model can be a potential tool for prediction of future LUCC.

Keywords

• GIS
• Land-use simulation
• Logistic regression
• Genetic algorithm
• Fuzzy Kappa
• Ifugao rice terraces