Parameterization of a Non-Linear Groundwater Storage-Discharge Relationship for PCR-GLOBWB

Abstract

Groundwater is an essential major water resource that is utilized for human water needs and is an integral part of the intricate water system. To grasp its complexity and challenges, a global groundwater model has been developed. This model predicts future groundwater patterns, especially under the influence of climate change, and serves as a risk mitigation tool in groundwater management. PCR-GLOBWB 2 is one such global hydrological model that can be coupled with the groundwater model. In this groundwater model, the linear reservoir concept is applied. However, some previous studies have indicated a non-linear relationship between baseflow and groundwater storage represents a better reality compared to the linear reservoir. Therefore, we aim to identify and apply this non-linearity using the recent groundwater model, GLOBGM. Additionally, we explore the multiple drainages and various conductance concepts to explore potential model improvements. The result of this recent advancement concept does not yet show an entire improvement. However, improvements are observed partially, especially in regions with wet or high-precipitation climates based on their Kling-Gupta Efficiency (KGE) values. In contrast, the model performs poorly in most dry climate areas, sometimes failing to simulate discharge, as seen in the River Niger, Mali, where it only shows a zero value for some periods. A very low variability and bias in the simulated data indicate that the discharge from the new model is still often underestimated when compared to observational GRDC data. The temporal dynamic of this new model, however, is better than the PCR-GLOBWB default version, as indicated by the correlation and determination coefficient (R2), which shows a notable improvement. In subsequent research, a novel regression approach may be taken into consideration to improve model performance. The very low recession coefficient found in this study can be replaced with the default recession coefficient from the PCR-GLOBWB 2 dataset to find a new baseflow exponent.