Estimation of evaporation in rainfall-runoff models

The first part of this paper compares two algorithms relating daily evaporation to potential evaporation and simulated soil water content. The first algorithm assumes that the ratio of actual to potential evaporation is a single-valued function of water content. The second algorithm, which appears to be conceptually superior, determines the actual evaporation as the smaller of the potential evaporation and a rate determined by the soil water content. The comparison is based on evaporation data from a deep weighing lysimeter, and soil water data obtained by a neutron moisture meter, the vegetation being a deep-rooted pasture grass. Although there is considerable scatter in the results, the second algorithm performs slightly better than the first. The second part of the paper examines the hypothesis that runoff is only marginally affected by day-to-day variations in evaporation, and thus average monthly potential evaporation data can be used in place of daily data as a model input. Results from 4 models applied to data for 15 Australian benchmark catchments suggest this hypothesis is correct, and furthermore that for some catchments the best fit of all models to runoff data is obtained with unrealistically high values of potential evaporation.