Spatio-temporal analysis and cellular automata-based simulations of biophysical indicators under the scenario of climate change and urbanization using artificial neural network

Abstract

Rapid urbanization is an indicator of infrastructure and economic development. Changes in the urbanization pattern contribute significantly to land use land cover (LULC) change, precipitation pattern, and vegetation cover (VC). These changes are intensified by climate change and the increasing population. Understanding the responses of the biophysical indicators such as precipitation accumulation (PA), trends, and vegetation cover to climate change and rapid urbanization is the key to predicting future scenarios. This study aims to monitor and simulate the scenario of the biophysical indicators in response to urbanization and climate change in the Rangpur district, Bangladesh. Landsat 4–5 TM and 8 OLI satellite images, data regarding meteorological conditions, and biophysical indicators from 2001 to 2020 were used to quantify the LULC, PA, and VC changes and predict future scenarios for 2025 and 2030 using machine learning algorithms such as cellular automata (CA) and artificial neural network (ANN). Between 2001 and 2020, urban areas and vegetation have increased by 768% and 27.9% respectively, followed by redactions in water bodies and barren land by 65% and 77.5% respectively. Simulation results indicate significant growth in urban areas by 17.4% and 25.1% in 2030 and 2040, flowed by decreased vegetation cover (1.2% and 1.8%), water bodies (12.9% and 17.2%), and barren lands (22.9% and 31.5%). Mann Kendall trend test also shows no pattern for eight months in one year, suggesting changing precipitation pattern. These changes are indicative of climate change at a micro-level and threaten sustainable development and climate resilience.