The flow-based multi-model ensemble strategy for projecting future runoff with CMIP6 GCMs

Abstract

Abstract The dramatic rise in the number of global climate models (GCMs) caused a rise in the uncertainty of future runoff projection. A multi-model ensemble (MME) of suitable GCMs selected based on their performances has been proposed to solve this problem. This study used three MME generation methods which are climate-based, mixed climate-flow-based and flow-based approaches, coupled with two GCM selection methods (all GCMs and five best-performing GCMs), and two weight assignment methods (equal and unequal) to prepare the best MME to assess their relative performances in simulating historical runoff and reducing uncertainty in future runoff projections. The GCMs were selected from 20 coupled model intercomparison project phase 6 (CMIP6) models, while Storm Water Management Model (SWMM) was used for long-term runoff simulation based on MMEs for four shared socioeconomic pathway scenarios (SSPs). Four evaluation metrics were used to verify the performance of each method, and the uncertainty of future runoff simulation was quantified using the reliability ensemble averaging (REA) method. The flow-based MME approach provided a better simulation of historical runoff and also lowered the uncertainty in future runoff simulation compared to the other MMEs. The selection of an efficient GCMs subset and assigning unequal weights to GCMs showed more effective than considering all GCMs and equal weight. The results of this study can provide meaningful information to researchers in future runoff projections using GCMs.