Assessment of climate change uncertainty effects on groundwater level prediction using Bayesian analysis

This work investigates the effect of climate-change uncertainty on groundwater level prediction using the Bayesian approach. This work’s methodology was applied to the Qom-Kahak aquifer, in Iran. Five Atmospheric Ocean General Circulation Models (AOGCMs) and a new Hybrid model are implemented under A2 and B2 Greenhouse Gases (GHGs) emissions scenarios during a current period (2001–2017) and a future period (2054–2069) to simulate monthly precipitation in the region encompassing the Qom-Kahak aquifer to assess the effect of climate change on groundwater resources. AOGCMs’ climate projections are weighed using the Mean Observed Precipitation (MOP) method. Next, a Bayesian-based hybrid model is developed. Subsequently, the effect of climate change uncertainty on the prediction of the groundwater level of Qom-Kahak aquifer is assessed using Monte-Carlo simulation and sampling from the Probability Distribution Function (PDF) of monthly downscaled precipitation. The largest and smallest precipitation reductions under the A2 scenario correspond to CCSR-NIES (by -20.2%) and hybrid model (by -4.3%). The largest and smallest precipitation increases under the A2 scenario correspond to GFDL R30 (by + 26.8%) and CGCM2 (by + 5.7%). The largest and smallest precipitation reductions under the B2 scenario correspond to CCSR-NIES (by -4.3%) and HadCM3 (by -1.4%). The largest and smallest precipitation increases under the B2 scenario correspond to GFDL R30 and CGCM2 (by + 1.4%). The projected future precipitations by the AOGCMs and Hybrid model are introduced to the Groundwater Modeling System (GMS) to project the future groundwater level. GMS simulations indicate the largest declines in groundwater level compared to the current climate are associated with CCSR-NIES and CSIRO-MK2 by 20.2 and 14.4%, respectively. GFDL R30 shows the largest increase (26.8%) in groundwater level under A2. Also, the largest decline (4.3%) corresponds to CCSR-NIES under B2, and the largest increase (1.4%) is projected to be equal by CGCM2 and GFDL R30.