Fill‐spill‐merge terrain analysis reveals topographical controls on Canadian river runoff

Abstract

AbstractDigital Elevation Models (DEMs) are a crucial tool for watershed analysis, offering valuable insights into landscape‐scale hydrology. Traditional watershed delineations are derived by filling a DEM to force flow paths through topographic depressions, thus creating a continuous drainage network throughout the domain. However, this approach is challenged in landscapes with abundant real‐world depression storage, intermittently flowing stream channels, and internally drained lake basin (endorheic basin) such as the Canadian Shield (CS). The CS landscape is characterized by “fill‐and‐spill” surface water hydrology, with runoff flow paths controlled by bedrock sills and rocky cascades that overtop when water levels are high but cease flowing when water levels are low. To better represent these intermittent drainage networks, we apply a non‐traditional, less‐aggressive DEM filling model (Fill‐Spill‐Merge or FSM) to a continental‐scale DEM (MERIT) all of Canada. To ensure adequate filling of DEM noise while also preserving real‐world topographic depressions, we propose a climatic method to initialize a key FSM parameter (“runoff depth”) that calibrates observed discharges from 1690 Environment and Climate Change Canada (ECCC) river gauges with climate model P‐ET (precipitation minus evapotranspiration) data. Our application of FSM to all 1690 gauged watersheds identifies 916 significant topographical control points controlling >20% and/or 1000 km2 of their respective areas. The Geikie, Snare, Kazan, Tazin, and Seal rivers may be particularly affected, with impacted watershed areas ranging from 12% to 64%. Extending this approach to ungauged parts of the CS reveals an additional 635 significant topographical control points. Ensemble climate model projections suggest that around 10% of these control points are currently dry but will become active by 2100. This research explicitly determines how CS watersheds are affected by fill‐and‐spill hydrology, and demonstrates the importance of accurate terrain modelling for delineating surface water flow paths in depressional landscapes.