What Can We Learn from a Reduced-Dimensional Groundwater Representation for Diagnosing Groundwater–Land Interactions? Insights from the Water Table Ratio

Aoqi Sun, Zeyu Tang, Chunmiao Zheng, Wei Shangguan, Chen Yang

Groundwater exerts an important control on land–atmosphere interactions, yet its explicit representation in Earth system models remains computationally prohibitive. Reduced-dimensional metrics, such as the Water Table Ratio (WTR), have been proposed to represent groundwater influences on land–atmosphere interactions. However, their uncertainty, classification stability, and correspondence with process-based representations remain largely unexplored. Here, because groundwater influences on the atmosphere are mediated through land surface processes, we compare WTR-based diagnoses of groundwater–land interactions with those derived from ParFlow–CLM simulations over a 34,000 km² catchment. Groundwater-sensitive regions are first identified by comparing paired ParFlow–CLM simulations with and without groundwater feedback and then used as a physically explicit reference for comparison with WTR. A total of 63 WTR configurations were examined, with Monte Carlo analyses performed for each to quantify uncertainty and classification stability.
Hydraulic conductivity (K) is the dominant source of uncertainty in WTR and exerts the strongest destabilizing effect on its threshold-based classification. Although the domain-wide proportions of correctly classified sensitive and insensitive areas remain nearly unchanged across alternative configurations and Monte Carlo realizations, substantial switching occurs among individual grid cells. Across the 63 WTR configurations, the mean correctly classified fraction was 0.40. In comparison, the critical-water-table-depth (WTD) approach yielded fractions of 0.59, 0.70, and 0.64 using WTD estimates from ParFlow–CLM, PCR-GLOBWB, and Fan et al., respectively. WTR errors are dominated by overestimation rather than underestimation, indicating a conservative screening tendency. Combining observed parameter contrasts with analytical WTR elasticities further identifies K as the dominant contributor to classification discrepancies, with characteristic length (L) providing a secondary contribution. The dominant role of K reveals a key consequence of reducing a heterogeneous three-dimensional groundwater system to a one-dimensional metric governed by a few effective parameters: the resulting classification becomes disproportionately dependent on a single effective hydraulic conductivity field. WTR can therefore support precautionary regional screening, but its local classifications should be interpreted cautiously, particularly where K is poorly constrained or subsurface heterogeneity is strong.