A linked global model of terrestrial hydrologic processes: Simulation of modern rivers, lakes, and wetlands

Abstract

A terrain‐based hydrologic model is developed to simulate rivers, lakes, and wetlands on the continental scale as a linked dynamical system. This surface water area model (SWAM) is an extension of earlier work [Coe, 1997] and is based on a linear reservoir model. The model requires, as input, estimates of runoff, precipitation, and evaporation from either observations or climate simulations. The model develops its own river transport directions based on elevation. The river discharge is calculated at each grid cell as the accumulated flow of water across the land surface. Lake and wetland area and volume are calculated as a function of the local precipitation minus evaporation, the streamflow into and out of the grid cells, and the potential for storage of water on the surface as derived from 5′ × 5′ resolution digital terrain models. SWAM is applied on a 5′ × 5′ spatial resolution to simulate present‐day surface waters and river transport for all continents (except Antarctica and Greenland). The model simulates the discharge of the 27 major rivers of the world in fair agreement with the observations; 12 of the rivers have simulated discharge within ±20% of the observational estimates. The discharge from rivers in arid and semi‐arid climates is generally overestimated probably due to underestimated evaporation from wetlands, irrigated croplands, and reservoirs within the river basins. The modern simulated lake area (about 3% of the land area) is larger than the observed area (about 2% of the land area). Wetlands are poorly simulated by the model due to the coarse vertical and horizontal resolution of the digital terrain models. The model simulates the potential for increased surface water (up to 7.5% of the global land area) and river basin area in semi‐arid and arid regions where closed basins exist in the present day climate which illustrates the utility of SWAM for climate change experiments. While higher resolution and more accurate digital terrain models are needed to improve these surface water simulations (particularly for wetlands), these preliminary results indicate that current digital terrain models are adequate for including lakes and rivers as interactive components in the surface hydrology parameterizations of climate models and for studying feedbacks between lakes and climate.