This paper presents a methodology for estimating bare soil evaporation from microwave radar data. The classical Richards' equation is solved numerically by using soil moisture values estimated from simulated radar data as the upper boundary condition. It is shown that a sampling period of 3 days for the radar data permits a good estimation of evaporation for a clay soil with relatively low hydraulic conductivity values. However, for a sandy soil with higher hydraulic conductivity values the sampling period must be shorter, particularly for wet conditions which occur after rain events. The possibility of extending the local transfer model to the field scale, matching the expected observed scale from airborne or satellite radar, is investigated. We show that this can be done when the spatial heterogeneities of the field can be scaled through similar media concept.
INTRODUCTIONActual evapotranspiration (and/or evaporation from bare soils) is an important Parameter, by itself or as part of the energy budget at the ground surface, which often needs to be known at a regional scale, compatible, for instance, with the grid scale of meteorological models [Manabe, 1969; Wiin Nielsen, 1980] or with the size of watershed models [Saxton et al., 1974]. On a local basis it can be obtained either by measuring the latent heat flux in the lower part of the atmosphere (aerodynamic, Bowen ratio methods, or any combination of those methods) or from soil water balance based on simultaneous measurements of volumetric water content and hydraulic head profiles in the soil.
At the regional scale, potential evapotranspiration (E v)can be evaluated from classical meteorological observations. Extension to actual evapotranspiration (ET) is more difficult and assumes the existence of a regional equilibrium regime [Bouchet, 1963;Priestley and Taylor, 1972;Brutsaert and Stricker, 1979] or includes soil and vegetation parameters which account for the more or less strong heterogeneities of the ground surface.With the development of remote sensing techniques welladapted to regional scale observations, many attempts have been made to use remote sensing data in place of or added to meteorological ones. Most of the attempts currently use thermal infrared data coupled with radiative energy budget [ldso et al., 1977;Soer, 1980;Price, 1980] to infer soil moisture and evapotranspiration. The surface soil moisture can also be used as additional variable to determine actual evapotranspiration, either through relationships between E v and ET [Deardorff, 1977;Barton, 1979] or as the upper boundary condition in a soil water transfer model as it is shown in this paper. Direct monitoring of surface soil moisture can be derived on a large scale from microwave measurements, active [Ulaby et al., 1978[Ulaby et al., , 1979 or passive