https://doi.org/10.1029/2025WR043074

Moghairib, Mohamed; Clark, Martyn P.; Pietroniro, Alain; Stadnyk, Tricia (2026) A Model-Agnostic Representation of Prairie Pothole Hydrology: Enhancing Generality and Implementation Across Hydrological Models, Water Resources Research, https://doi.org/10.1029/2025WR043074

Modeling streamflow in low-lying, flat, and pothole-dominated prairie or Arctic regions is challenging due to variable non-contributing areas that influence how runoff translates to streamflow. Several modeling approaches have been developed to represent these dynamics, but many (a) lump depressions and permit spill only after a fixed capacity is reached, (b) rely heavily on calibration, (c) are unsuitable for large basins, (d) do not account for non-pothole contributions, and/or (e) are not model-agnostic. Here we present HDSv2, a second-generation Hysteretic Depressional Storage (HDS) module that is open-source, model-agnostic, numerically robust, and grounded in long-established physical understanding of prairie potholes. HDSv2 represents dynamic contributing area and storage–discharge hysteresis, enabling realistic simulation of fill-and-spill behavior and cold-region processes. We couple HDSv2 with three hydrological and land-surface models of differing architectures: HYPE (Hydrological Predictions for the Environment), MESH (Modélisation Environnementale communautaire—Surface and Hydrology), and SUMMA (Structure for Unifying Multiple Modeling Alternatives), applied in the Smith Creek River Basin, Canada. Results show that HDSv2 improves numerical stability and process fidelity relative to the original HDS model, which exhibited instabilities affecting contributing-area simulation within HYPE. Across all host models, integrating HDSv2 produces more robust hydrographs than the original configurations and better reproduces observed relationships between depressional storage and contributing area. Although hydrograph improvements vary by host, additional performance metrics show consistent gains in both high and low flow conditions. These findings demonstrate that HDSv2 provides a transferable and scalable pathway for incorporating depressional-storage dynamics into diverse hydrological models and regions.

Predicting water movement across landscapes is especially challenging in flat, low-lying regions such as the North American prairies and Arctic areas, where numerous small depressions (potholes) temporarily store water and disrupt how runoff reaches streams. Most hydrological models simplify these depressions as a single storage that overflows only when full, which limits their ability to represent real prairie wetland behavior. These models also often require extensive calibration and may not transfer well across regions or modeling systems. In this study, we develop an improved, open-source version of the Hysteretic Depressional Storage model (HDSv2), grounded in decades of prairie hydrology research. HDSv2 more realistically simulates how potholes fill, spill, and connect to streams by dynamically linking water storage with the area of the landscape that contributes runoff. This approach is numerically stable, physically realistic, and can be added to a wide range of hydrological models without major structural changes. We test HDSv2 in three different modeling systems and show substantial improvements in streamflow prediction for a prairie watershed in Canada. These advances support better water-resource management and forecasting in regions where small depressions play an important hydrological role, with potential applications in flood mitigation, agriculture, and climate-resilience planning.

https://doi.org/10.1029/2025WR043074