Most models of evaporation ( E ) provide estimates at one rather than many locations and thus cannot be used to describe the spatial variability of evaporation. An energy balance model (EBM) that estimates E at many locations was tested, improved, and validated, using daily evaporation measurements made with microlysimeters, giving an r2 value of 0.82 for regression of actual vs. estimated evaporation. The model is based on the surface energy balances of dry and drying soil. Data needed include only wind speed and soil surface temperature measurements obtained at a suitably small time interval (e.g., 0.5 h) with an automated weather station and reference dry soil at one location, and measurements of predawn and midday soil surface temperature made with a hand‐held infrared thermometer at as many locations as desired for evaporation prediction. The reference dry soil was established in a plastic bucket buried in the soil and protected from rain and irrigation. Model improvements included an easy method of accurately estimating continuous soil surface temperature at many points in a field. Also, an empirically fitted transfer coefficient function for the sensible heat flux from the reference dry soil showed that sensible heat flux from the relatively hot reference dry soil was dominated by free convection. Soil heat flux and reflected shortwave radiation terms are omitted in the EBM and this was shown to reduce model accuracy by as much as 9.2% of the measured evaporation. The model may prove useful for prediction of spatial variability of evaporation based on soil surface temperatures.