Analysis and prediction of land use/land cover change in the Llanganates-Sangay Connectivity Corridor by 2030

Abstract

This paper analyses Land Use and Land Cover (LULC) change trends in the Llanganates-Sangay Connectivity Corridor (CELS) from 2018 to 2022 and predicts trends through 2030. MapBiomas LULC maps reveals annual change rates (2018–2022) of -0.37 %/year (-1147.33 ha) for Forest Formation, -1.17 %/year (-30.01 ha) for Non-Forest Natural Formation, 2.21 %/year (906.19 ha) for Agriculture and Livestock Areas, 8.50 %/year (250.84 ha) for Non-Vegetated Areas, and 0.17 %/year (30.31 ha) for Water Bodies. The higher annual change rate inside Forest Formation is -0.58 %/year (-990.35 ha) occurring in areas not designated under any conservation status. Projections for 2030 were made using the MOLUSCE tool, combining an Artificial Neural Network (ANN) model with Cellular Automata simulations. The ANN model was trained on five explanatory variables and LULC maps from 2018 and 2020, achieving a training error of 8.46 %. Predictive accuracy was assessed by comparing the simulated 2022 LULC map with the 2022 MapBiomas map, resulting in a Kappa coefficient of 0.95, indicating excellent predictive accuracy. Additionally, LULC simulations from 2022 to 2030 predict annual rates of change of –0.27 %/year (-1628.97 ha) for Forest Formation, -1.39 %/year (-63.49 ha) for Non-Forest Natural Formation, 1.92 %/year (1778.26 ha) for Agriculture and Livestock Areas, 0.97 %/year (30.38 ha) for Non-Vegetated Areas, and 0.63 %/year (-146.18 ha) for Water Bodies. The findings show that annual rates of deforestation will remain low and protected areas will have less deforestation than nonprotected areas.