Modeling soil evaporation has been a notorious challenge due to the complexity of the phenomenon and the lack of data to constrain it. In this context, a parsimonious model is developed to estimate soil evaporative efficiency (SEE) defined as the ratio of actual to potential soil evaporation. It uses a soil resistance driven by surface (0-5 cm) soil moisture, meteorological forcing and time (hour) of day, and has the capability to be calibrated using the radiometric surface temperature derived from remotely sensed thermal data. The new approach is tested over a rainfed semi-arid site, which had been under bare soil conditions during a 9-month period in 2016. Three calibration strategies are adopted based on SEE time series derived from (1) eddy-covariance measurements , (2) thermal measurements, and (3) eddy-covariance measurements used only over separate drying periods between significant rainfall events. The correlation coefficients (and slopes of the linear regression) between simulated and observed (eddy-covariance-derived) SEE are 0.85, 0.86 and 0.87 (and 0.91, 0.87 and 0.91) for calibration strategies 1, 2 and 3, respectively. Moreover, the correlation coefficient (and slope of the linear regression) between simulated and observed SEE is improved from 0.80 to 0.85 (from 0.86 to 0.91) when including hour of day in the soil resistance. The reason is that, under non-energy-limited conditions, the receding evaporation front during daytime makes SEE decrease at the hourly time scale. The soil resistance formulation can be integrated into state-of-the-art dual-source surface models and has calibration capabilities across a range of spatial scales from spaceborne microwave and thermal data.