EVAPORATION IN THE SURFACE ENERGY BALANCE 23 



tive measurement. The plates should be near the surface to minimise storage 

 errors arising from undetected changes in the surface layers, but deep 

 enough to permit proper growth of grass or other cover. For many of the 

 larger-rooted crops it is, of course, necessary to revert to temperature 

 profiles, though soil flux-plates can still be used with advantage to measure 

 the flux at a level below the roots. In addition warning should be given 

 concerning the use of flux-plates near the surface in very dry conditions. 

 Drying out or cracking of the surface layers is then Hable to cause imperfect 

 thermal contact with the flux-plate, and evaporation may occur at depths 

 below that of the detecting instrument, giving false indications of the lack 

 of an energy balance. 



For specialised short-term appHcations, it may be desirable to correct for 

 the storage term. This could be done quite accurately by invoking the 

 well-known exponential decrease of the ampHtude of the diurnal tempera- 

 ture variation with depth, and also the linear dependence of the phase lag 

 of temperature with depth. By comparing graphically the results of two or 

 three flux-plates buried at suitable intervals of depth, values of the flux at 

 the surface could be derived by extrapolating first for zero phase lag and 

 then adjusting the ampHtude in accordance with the exponential law. 

 Strictly, this assumes that the physical parameters for the soil are constant 

 with depth, but only a second-order approximation is involved here. After 

 initial trials have been carried out for a particular site and arrangement of 

 instruments, it might be possible to allow for phase lag by altering the time 

 scale, and for ampHtude by incorporating a correction in the caHbration 

 constant. 



The original method of caHbration used was one described by Hatfield 

 and Wilkins (1950) using layers of felt as the surrounding medium. To 

 obtain more reaHstic conditions, the felt was replaced in turn by dry sand, 

 oven dried soil and finally by sHghtly moist soil. Though S is often as Httle 

 as a tenth o£ R, and typicaUy of the order of a fifth of i^, it becomes of 

 similar importance during the night when systematic errors must be avoided 

 if insight is to be gained into the variation of K^ under stable conditions. 

 The typical order of accuracy of soil flux-plate caHbrations has been about 

 ± 20%, due largely to variations in thermal conductivity of the soil com- 

 pared with that of the flux-plate. To bring about a significant improvement 

 it has been found necessary to caHbrate in situ. For this, a caHbrating box is 

 placed over the site of the flux-plate, for various soil conditions, and a 

 constant flux is directed downwards until a steady reading is obtained. 

 This takes about two days to achieve and the apparatus is therefore made 

 self-balancing. By suitably varying the heat flux, the phase lag and ampHtude 



