A multi-objective Markov Chain Monte Carlo cellular automata model: Simulating multi-density urban expansion in NYC

Cellular automata (CA) models have increasingly been used to simulate land use/cover changes (LUCC). Metaheuristic optimization algorithms such as particle swarm (PSO) and genetic algorithm (GA) recently introduced into CA frameworks generate more accurate simulations. Although Markov Chain Monte Carlo (MCMC) is simpler than PSO GA, it rarely calibrate models. In this article, we introduce a novel multi-chain multi-objective MCMC (mc-MO-MCMC) model LUCC. Unlike the classical MCMC, proposed mc-MO-MCMC multiple chains method that imports crossover operation from evolutionary algorithms. each new chain, after initial one, operator generates solution. The selection of solutions be crossed over are made according their fitness score. paper, chose example New York City (USA) apply our three conflicting objectives non-urban low-, medium- or high-density urban between 2001 2016 using USA National Land Cover Database (NLCD). Elevation, slope, Euclidean distance highways local roads, population volume average household income LUCC causative factors. Furthermore, demonstrate efficiency model, compare with (MO-GA) standard single-chain (sc-MO-MCMC). Our results produces simulations use dynamics featured by faster convergence Pareto frontier comparing MO-GA sc-MO-MCMC. cellular should efficiently help trade-off among and, possibly, change at once.