Modeling and forecasting water quality responses to intensifying drought

ABSTRACT

This study evaluates drought in the Kor River, southern Iran, for 2000, 2010, and 2020 using remote sensing-based drought indices and assesses its impact on water quality (WQ). WQ was analyzed at 30 sampling sites using Ca, Cl, EC, HCO3, K, Na, Mg, SO4, TDS, and TH, with Kriging interpolation and the water quality index. Regression identified the most influential drought index for each parameter, and concentrations were predicted using multilayer perceptron (MLP) neural networks along with Markov and CA-Markov models. Results indicated increasing drought severity from 2000 to 2020, with southern regions more affected. WQ declined from 2015 to 2020 due to rising dissolved element concentrations. Correlation analysis showed strong links: Ca with temperature vegetation dryness index (R = 0.820), Cl, EC, K, Na, Mg, TDS, and TH with precipitation condition index (R > 0.80), and SO4 with temperature condition index (R = 0.815), and HCO3 with normalized difference vegetation index (R = 0.831). CA-Markov projections suggest worsening drought in southern areas by 2040. The MLP model accurately predicted WQ, with R2 values of 0.85 (Cl), 0.72 (EC), 0.71 (K, Na, Mg), and 0.89 (HCO3). The study warns that ongoing and intensifying droughts will likely worsen water pollution in the Kor River in the coming decades.