Computing water flow and storage in complex landscapes

Abstract

Depressions in the landscape can hold water, forming lakes or, on a smaller scale, puddles. Some of these features form long-term water stores, while others are ephemeral, filling and emptying as local weather conditions change. They impact hydrologic connectivity as they either store incoming water, or allow it to pass through them to another portion of the hydrologic system. Surface-water contained in these depressions also interacts with local groundwater. Despite these interactions, many flow-routing algorithms require the complete removal of depressions from a landscape by filling or carving. Resultant flow-routing surfaces can host a continuous, integrated drainage network, but lose information about possible surface-water storage and interruptions to the flow network. I first present an initial answer to this problem: FlowFill, an algorithm that routes a prescribed amount of runoff across the surface in order to flood depressions, but only if enough water is available. FlowFill is useful both in determination of flow-routing surfaces, and in visualisation of changing hydrology through time. I then present Fill-Spill-Merge, a method that rapidly processes and distributes runoff to and through depressions. Fill-Spill-Merge makes use of a depression hierarchy data structure which records information about depressions and their relationships to one another. Fill-Spill-Merge produces appropriate flow-routing surfaces up to 2,600 times faster than FlowFill. Finally, I couple Fill-Spill-Merge with a groundwater module in the Water Table Model. The Water Table Model is capable of computing changes in water table elevation at large spatial scales and over long temporal scales. It incorporates lakes into water table elevation estimates and enables assessments of changing terrestrial water storage through time. I demonstrate an application of the Water Table Model in North America’s Great Basin, where a suite of model runs reveals past hydroclimate.