Efficient techniques for incorporating standing surface water into hydrological models

Abstract:

Depressions – inwardly draining regions of landscapes – present difficulties for terrain analysis and hydrological modeling. In the past, keeping depressions in models was either impossible because models required a depression-free landscape or meant that computation would be inefficient because iterative methods would be required. This often meant that depressions were removed as a preprocessing step prior to modeling. However, depressions serve important roles in forming wetlands, impacting subglacial hydrology, controlling terrestrial water storage, and influencing soil water retention and flood extent. Lakes and wetlands also host biodiversity and provide ecosystem services including denitrification and recreation, in addition to impacting as well as drainage network integration and realignment. Therefore, retaining depressions in models is important.

Here, we demonstrate new techniques which allow depressions and standing water to be treated efficiently in models. These techniques leverage simple data structures to dramatically reduce the computational load of hydrological models. In a case study, we show that a global model coupling surface- and groundwater runs 90-2000x faster using the new techniques. We discuss how this improvement enables new science by expanding the fidelity of models and their ability to represent complex interactions among natural systems.