Abstract: Abstract: Soil evaporation consumes a large part of evapotranspiration during the crop growth season, especially during the seedling or sparse crop growth stage. It has been reported that soil evaporation makes little contribution to crop yield, and thus it has been seen as invalid water consumption. Separate determination of soil evaporation and transpiration is required in many irrigation management programs or yield analysis models. However, it is quite difficult to directly measure soil evaporation and transpiration separately. To achieve this purpose, a soil evaporation model was developed using a new defined soil moisture function based on the actual measurement of meteorological data (air temperature, relative humidity, and wind speed), soil surface moisture and soil evaporation data. The model combined two processes of water vapor transfer: one is the vapor transport in air while the other is molecular diffusion of vapor in the surface soil pore with the vapor being carried from the interior soil pore to the land surface. For the field observation, air temperature and relative humidity were measured in three different heights above the buckwheat canopy in order to determine the actual evapotranspiration with Bowen ratio energy balance method. Leaf area index and plant height was measured regularly, with the maximum values of 2.25 and 62.7 cm, respectively. The variation of surface soil water content (5 cm) was from 11.2% to 30.9%. An important parameter, surface moisture availability, in the proposed model was decided by surface soil moisture and wind speed. It was shown that surface soil water content was the main factor affecting surface moisture availability, and wind speed had slight influence on it. The modeled surface moisture availability with soil content and constant wind speed was compared to calculated value with varied wind speed. By assuming surface moisture availability to be 1 in the model, another important parameter, bulk transfer coefficient, could be calculated. It has been reported that the bulk transfer coefficient for bare field is mainly influenced by soil texture and atmospheric stability. In this study, average value of bulk transfer coefficient was applied for three different leaf area stages based on the analysis of its actual variation. Actual evapotranspiration and soil evaporation beneath the buckwheat canopy respectively measured by Bowen ratio energy balance method and micro-lysimeter were compared and the soil evaporation measured by micro-lysimeter was applied to validate the accuracy of the model. It was shown that the soil evaporation beneath the buckwheat canopy during seedling stage was quite close to actual evapotranspiration measured by Bowen ratio energy balance. The average hourly soil evaporation measured by Bowen ratio energy balance and micro-lysimeter were 0.16 and 0.17 mm, respectively; while the average relative error between two methods was 12%, root mean square error was 0.077, and correlation coefficient was 0.89. It was also shown that the soil evaporation beneath the buckwheat canopy could be reproduced using the constructed surface moisture availability model with average relative error of 13.5%, root mean square error of 0.249, and correlation coefficient of 0.95. The study is very important in separately estimating soil surface evaporation beneath the canopy and crop transpiration, and in decreasing invalid water consumption through soil surface beneath the canopy.