Urban growth scenario projection using heuristic cellular automata in arid areas considering the drought impact

摘要: Arid areas with low precipitation and sparse vegetation typically yield compact urban pattern, and drought directly impacts urban site selection, growth processes, and future scenarios. Spatial simulation and projection based on cellular automata (CA) models is important to achieve sustainable urban development in arid areas. We developed a new CA model using bat algorithm (BA) named bat algorithm-probability-of-occurrence-cellular automata (BA-POO-CA) model by considering drought constraint to accurately delineate urban growth patterns and project future scenarios of Urumqi City and its surrounding areas, located in Xinjiang Uygur Autonomous Region, China. We calibrated the BA-POO-CA model for the drought-prone study area with 2000 and 2010 data and validated the model with 2010 and 2020 data, and finally projected its urban scenarios in 2030. The results showed that BA-POO-CA model yielded overall accuracy of 97.70% and figure-of-merits (FOMs) of 35.50% in 2010, and 97.70% and 26.70% in 2020, respectively. The inclusion of drought intensity factor improved the performance of BA-POO-CA model in terms of FOMs, with increases of 5.50% in 2010 and 7.90% in 2020 than the model excluding drought intensity factor. This suggested that the urban growth of Urumqi City was affected by drought, and therefore taking drought intensity factor into account would contribute to simulation accuracy. The BA-POO-CA model including drought intensity factor was used to project two possible scenarios (i.e., business-as-usual (BAU) scenario and ecological scenario) in 2030. In the BAU scenario, the urban growth dominated mainly in urban fringe areas, especially in the northern part of Toutunhe District, Xinshi District, and Midong District. Using exceptional and extreme drought areas as a spatial constraint, the urban growth was mainly concentrated in the "main urban areas-Changji-Hutubi" corridor urban pattern in the ecological scenario. The results of this research can help to adjust urban planning and development policies. Our model is readily applicable to simulating urban growth and future scenarios in global arid areas such as Northwest China and Africa.