Abstract: The quantitative relationships among wheat surface temperature (TL), air temperature (Tɑ) and siol-surface temperature (TS) at noon in the wet and dry wheat fields were investigated based on field data. The data collected was divided into a training set and a validation set and three types of weather were categorized as clean, cloud and both. The results show that for the wet wheat field in clear days, the best results could be achieved using Tɑ to estimate TL; based on the final model, the average estimated error is 1.03oC and the RMS of estimated TL is 1.26 oC. For the dry wheat field in the clear or cloud days, TL could be well estimated with TS. The average estimated errors of the final model are 1.64 and 1.54 oC respectively and the RMS of estimated TL are 2.05 and 1.89 oC respectively. The RMS errors achieved are generally lower than those derived from NOAA digital images (2—3 oC). The results of this research show that it is feasible to forecast wheat surface temperature with air temperature and soil-surface temperature, too. It provides a new cheap effective means to get wheat surface temperature; at the same time, it is also possible to combine satellite image data with measured data from ground weather stations to predict the water requirement of winter over a larger region.