Two thermal channels of NOAA can be used to retrieve land surface temperature, which can be used to monitor soil water content. Nine kinds of split-window algorithms were used to retrieve surface temperature of Handan during the winter wheat growing stage in 2002, and the results were validated by ground experiment. It indicates that UL92 method is the best one for land surface temperature retrieval in plain area of Handan with less average error and standard error. For the mountainous areas and highland, if digital elevation model was combined, split-window algorithm could retrieve land surface temperature more precisely. Based on the field experiment, functions between surface temperature of winter wheat and soil water content were established. The functions can be used along with UL92 method to monitor soil water content of different soil layers. That was a new method of using thermal remote sensing to monitor regional sectional soil water content. It was concluded that regional sectional water soil content can be monitored by combining the experiment function with remote sensing data.