Modelling Evaporation from a Subarctic Sedge Wetland

Abstract

Evapotranspiration is a major constituent of both the energy and water balances of wetland tundra environments. Reliable estimates of evapotranspiration are required in the analysis of specific climatological and hydrological problems occurring within a wetland. As a result, where direct measurements are unavailable, models designed to accurately predict evapotranspiration for a particular wetland are highly desirable. This paper evaluates the limitations, sensitivity and performance of four physically-based one-dimensional models in the simulation of evaporation from a subarctic tundra sedge wetland in the Hudson Bay Lowland near Churchill, Manitoba (58°45'N, 94°04'W). The surface of the study site consists of near-saturated peat soil with a sparse sedge canopy and a constantly varying coverage of open water. Measured evaporation was determined using the Bowen ratio approach to which the results of the models were compared. The comparisons were conducted with hourly and daily simulations over dry, wet and moderately wet surface conditions. The four models comprised two previously developed and tested models and two modified versions of these models. All four are based on the well-known Penman-Monteith combination formula. The first two are the Penman-Monteith model and the Shuttleworth-Wallace sparse canopy model. The third is an extension of the Penman-Monteith model which is weighted for surface area of the evaporation sources. The fourth is a modified version of the Shuttleworth-Wallace model which includes open water as an additional component to sparse canopy and bare soil as a contributor to the evaporation stream. Results from the study suggest that the weighted Penman-Monteith model has the highest potential for use as a predictive tool. In all four cases, the importance of accurately measuring the surface area of each evaporation source is recognized. The difficulty in determining a representative surface resistance for each source and the associated problems in modelling without it is also stressed. An analysis of the role and impact of feedbacks within the models is recommended as an important direction for future research.