Testing the alpha and beta methods of estimating evaporation from bare andvegetated surfaces in CLASS

Surface soil wetness determines whether evaporation occurs at the potential rate or is limited by soil water supply Many land surface models calculate evaporation by parametrizing the relative humidity at the soil surface (al pha method) or the soil water diffusion resistance (beta method). The relat ionships of alpha and beta to the average moisture content of soil surface layer thicknesses ranging from 0.5 to 10 cm were examined using Bowen ratio /energy balance measurements of evaporation from a bare loam/silt-loam soil at two adjacent sites, one of which was culti-packed while the other was d isc-harrowed. It was found that the relation ships were sensitive to the su rface layer thickness and tillage treatment and became better defined with larger thicknesses. Simulations of evaporation by CLASS (Canadian Land Surface Scheme) using Ph ilip's relationship (currently used in CLASS to determine alpha with a 10-c m thick surface soil layer) and our proposed relationships of alpha and bet a were compared with measurements made during a 7-day drying period for the culti-packed soil. Evaporation was significantly overestimated by Philip's relationship when the soil was dry. The alpha and beta relationships gave much better estimates of evaporation than Philip's relationship; however ev aporation under conditions of high air humidity (e.g., night-time) was sign ificantly sllppressed by the a relationship. In long-term simulations, ther e were large differences between simulated and measured evaporation rates d ue to differences between actual and calculated soil moisture. When CLASS w as forced to Else measured soil moisture, evaporation rates calculated usin g the beta relationship agreed remarkably well with the measured values, wh ile those calculated using Philip's relationship were significantly overest imated. The effectiveness of the beta relationship was also tested with a 2 -cm thick surface layer using data collected from another bare silt loam so il and for a 10-cm thick surface layer using data from a crop and two fores ts with different soil types. It gave better estimates of evaporation from these such ces than Philip's relationship. In the simulation with CLASS, th e beta relationship appeared less sensitive to possible errors in modelled drainage than Philip's relationship, and was recommended for use in the mod el as replacement for Philip's relationship.