We conducted an electromagnetic (EM31) survey of our target paddock located near Birchip, Victoria, Australia.  The apparent electrical conductivity (ECa) data were converted to sowing water content and lower limit (LL) via a calibration equation and then kriged to a 10 m grid.  Water use was derived by assuming the LL would be the harvest water content and subtracting the LL from the sowing water content and adding the seasonal rainfall.

The French and Schultz potential yield model was applied with different assumptions than originally proposed.  That is, the transpiration efficiency was assumed to be constant while the soil evaporation was assumed to vary from one part of the paddock to another reflecting changing soil types.  We assumed a potential transpiration efficiency of 22 kg/ha/mm for barley and varied soil evaporation from 20 to 200 mm depending on the soil type.  We used the ECa map to estimate soil evaporation in a linear way. 

Without calibration yields were overestimated by about 1.6 t/ha over the whole paddock. However, the overall pattern of simulating the high and low yielding parts of the paddock was simulated well.

A less data intensive but sound yield model adapted from an old point-source potential yield model, applied in a spatial context appears to offer a simple alternative method of explaining spatial variance in crop performance where water supply is the major determinant of yield.  This should also apply where subsoil constraints like salt are involved.  Further testing over a range of environments is necessary.