LAND USE LAND COVER (LULC) CHANGE ANALYSIS IN BAGUIO CITY USING GIS-BASED TRANSITION POTENTIAL MODELING

Abstract

Baguio City experienced rapid LULC changes in the past decades due to rising population resulting in increasing demand in residential and commercial areas. The LULC change from 2003 to 2011 was quantified through a transition potential model using Artificial Neural Network and Cellular Automata. Potential driving factors considered in this study were barangay population and population density, land category, elevation, slope, soil type, and distances from the Central Business District, roads, and sinkholes. Multilayer Perceptron with Backpropagation technique was employed in the modeling simulations where various value combinations for the five hyperparameters were tested. Among the combinations of hyperparameter values tested, the combination that achieved the highest simulation accuracy was 1, 0.001, 1000, 10, and 0.01 for Neighborhood, Learning Rate, Iterations, Hidden Layers, and Momentum respectively. The prediction results for the year 2035 show that built-up areas in Baguio City will increase by 760.79 hectares while vegetation and bare soil will decrease by 347.64 hectares and 413.26 hectares, respectively. Built-up areas are expected to form mostly in Alienable and Disposable lands, barangays with high population density, and areas near roads and pre-existing development. On the other hand, minimal built-up expansion is expected in vacant forests, forest reserves, and areas near sinkholes. These findings are in line with the city government's expectations. With an accuracy of 72.80% and a Kappa statistic of 0.61, it can be concluded that the model is capable of predicting future LULC change and may serve as a viable guide for future land use development plans.