Modeling future PM10 concentrations under climate change scenarios

Abstract Climate change poses significant challenges to air quality, particularly in arid regions prone to dust pollution. This study assesses future trends in particulate matter (PM 10 ) concentrations in Ahvaz, Iran, under climate change scenarios defined by the Sixth Phase of the Coupled Model Intercomparison Project (CMIP6). Historical climate data (1998–2014) and observed PM 10 records (2013–2022) were used to establish a baseline. Future climate variables were statistically downscaled using the LARS-WG 6.0 model, with projections from the MIROC6 model under three Shared Socioeconomic Pathways (SSP1-2.6, SSP2-4.5, SSP3-7.0). A Nonlinear Autoregressive Neural Network with Exogenous Inputs (ANN-NARX) was developed to forecast PM 10 concentrations for the period 2023–2042, using temperature, precipitation, and solar radiation as predictors. The ANN-NARX model showed strong performance with RMSE values of 8.66 µg/m 3 , 22.18 µg/m 3 , and 16.83 µg/m 3 , and correlation coefficients of 0.95, 0.96, and 0.92 for SSP1-2.6, SSP2-4.5, and SSP3-7.0, respectively. All scenarios indicate an increase in PM 10 levels, particularly under the high-emission SSP3-7.0 pathway, with the most pronounced rises during the summer months. Sensitivity analysis identified maximum temperature as the most influential predictor. These findings highlight the urgent need for proactive air quality management and integrated climate adaptation policies to mitigate health risks in dust-prone urban environments.