Selecting Unsaturated Flow Model Complexity for Groundwater Recharge Estimation Under Climate Change

Groundwater, the world’s largest accessible freshwater resource, supports billions of people but is increasingly threatened by excessive abstraction rates that exceed natural groundwater recharge (GWR). Sustainable groundwater management requires establishing an accurate balance between extraction and recharge. Hydrological models are commonly used to estimate GWR; these models are typically calibrated to historical data and then assumed to remain valid for future projections under the notion of stationarity, that is, the assumption that the conditions used for model training will remain representative in the future. However, under projected climate change, this assumption is likely to be violated in many regions. In this study, soil water content observations from the International Soil Moisture Network (ISMN) were used to calibrate both bucket-type and Richards’ equation models for estimating GWR across multiple sites, using the DREAM algorithm. For each location, the most appropriate model structure was selected based on the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC). Future climate projections from GCMs under the SSP5–8.5 scenario were then summarized as annual rainfall totals for all sites. Subsequently, site pairs were identified in which historical annual rainfall totals at one location resemble future projected totals at another location, while also exhibiting similar soil hydraulic properties. This framework enables testing whether the model method selected under historical conditions remains valid under future climates. Overall, the proposed approach offers a systematic method for determining the complexity of unsaturated flow models and is expected to reduce uncertainty in GWR estimates.