Prediction of water-level variations in an alluvial aquifer surrounded by karstic formations using a combined time series-geostatistical model: a case study from Iran

Understanding groundwater responses to recharge is important in alluvial aquifers which are surrounded by karstic formations in semi-arid and arid regions, such as the high Zagros region of western Iran. Evaluation of input and output time series provides comprehensive information on the hydrodynamic behavior of these aquifers. Time-series models were used to predict water levels using 20 years of monthly data from 29 observation wells in the Nahavand Plain aquifer. To test the trend (mean instability), water-level data were linearly fitted using Minitab 17 software. Different models were fitted to the time series for each well and the best model was selected based on the lowest value of the Akaike information criterion (AIC), as well as root mean square error (RMSE), coefficient of determination ( R 2 ) and modified Nash–Sutcliffe efficiency coefficient (MNSE). Water levels were interpolated using conventional kriging and inverse-distance weighting and show consistent trends. Water-level change maps were prepared for annual intervals from 2012 to 2019 and for the period 2018–2022. Most of the area shows a trend of rising water levels, which appears to reflect recharge through karstic formations around the margins of the plain.