Evaporation from bare soil in south-western Australia - a test of two models using lysimetry

Abstract

Physical models of the soil water balance components are useful where direct methods of determination are costly and time consuming. Information from such models is useful in developing more efficient soil and crop management practices. The simplicity of such models sometimes encourages their use in new regions without proper testing. In this study, two models of soil evaporation (Es), developed by Ritchie (Water Resour. Res., 1972, 8, 1204-13) and Boesten and Stroosnijder (Netherlands J. Agric. Sci., 1986, 34, 75-90) were evaluated in the dry mediterranean environment of the cereal belt of south-western Australia under a wide range of evaporative demand by using lysimetry. Both models treated soil evaporation as a two-stage process, but in the Ritchie model, second-stage evaporation varied with the square root of time, whereas in the Boesten and Stroosnijder model, Es varied with the square root of the potential evaporation (Eow). In the Boesten-Stroosnijder model, only the parameter â was stable over a wide range of potential evaporation Eow, whereas in the Ritchie model, both the parameters U and C varied with E(ow). However, interdependence between U and C was conservative, and higher values of U under low Eo were compensated by lower values of C, and vice versa. For the soil used in this study, the following values were found to be appropriate for the parameters: U, 4.7 mm; C, 4.0 mm1/2 ; and â, 1 . 5 mm1/2. For dry environments where seasonal Eo fluctuates widely, low U values and high C values are recommended. The Es estimated by the models agreed poorly with Es measured by the lysimeter on a daily basis. However, when the daily values were accumulated over periods of at least 6 days, good agreement was obtained, indicating that the errors associated with the daily estimates cancel out each other over time. Hence, both models gave satisfactory short-term and long-term predictions of Es, but in both cases variation between predicted and observed values was least in the Ritchie model. This is consistent with the conclusion from other studies in the region, which show that second stage evaporation dominates Es, and hence should depend more on soil moisture supply than on Eo.