Using artificial neural networks to model the impacts of climate change on dust phenomenon in the Zanjan region, north-west Iran

The effects of climate change on the dust phenomenon was simulated in this study using an artificial neural network (ANN) until 2050. Hourly Particulate matter concentrations and daily meteorological data were analyzed from 2007 to 2018 and 1988 to 2018, respectively, in Zanjan city. The outputs of HadGM2-ES (Hadley Centre Global Environmental Model, version 2-Earth System) models of atmospheric circulation were used to generate future climatic patterns under two scenarios of Representative Concentration Pathway (RCP2.6 and RCP8.5). The Long Ashton Research Station Weather Generator (LARS-WG 6.0) software was utilized for statistical down scaling and production of climate-related datasets in artificial high-resolution time series. The observed climatic variables, including maximum and minimum temperature and precipitation, were determined as predictors in the artificial neural network. The highest surge of PM10 levels was in May and July, and the lowest increase of PM10 was observed in December with a monthly average of 84.85 and 50.54 mu g/m(3), respectively. The highest amount of PM10 was estimated for the year 2043, with a concentration of 74.26 mu g/m(3). Minimum and maximum temperature and wind speed had a significant relationship with PM10 concentrations; further, this pollutant level increased by boosting each atmospheric variable. The minimum and maximum temperatures in both scenarios were rising till 2050, and the highest temperature growth was obtained under the worst situation of RCP8.5.