Sensitivity of Annual Evaporation to Soil and Root Properties in Two Models of Contrasting Complexity

Simulations of soil water and evapotranspiration with physically based models at broad scales vary in both complexity of processes modeled and in parameterization of soil and root properties. Sensitivity of annual evaporation Eann to some of these processes and parameters was tested with both a model allowing multiple soil layers (BROOK90) and a single-layered water balance model (WBM). For nine widely scattered locations in North America Eann was controlled primarily by climate and cover type, but within a location–type combination, Eann was controlled primarily by the available water capacity Wac, which is the product of available water fraction and effective root depth. The definition of the upper limit of available water is important; it is precisely defined here as the water volume fraction at 30-cm depth after 48 h of drainage from an initially saturated, homogeneous profile with a fixed gravity potential gradient at 2-m depth. Specification of root depth was as important as specification of available water fraction in determining Wac. In climates of intermediate wetness a 100-mm change in Wac caused a similar change in Eann at low Wac, but little change in Eann at high Wac. WBM responded similarly to BROOK90. In BROOK90, texture-dependent hydraulic properties caused additional effects of less than ±50 mm in Eann for short covers and even less for tall covers. Effective root depth interacted with both distribution of infiltration and upward movement of water in the soil profile, but the effects were also on the order of only 50 mm. A multilayered model does not seem necessary for simulating Eann at the global scale if the primary objective is budget closure. Improvement in estimating Eann with WBM or similar global water budget models is not likely to result from making the model more complicated with respect to soil and root properties in the context of much larger uncertainties in atmospheric forcings.