32 M.J.BLACKWELL 



crops which completely cover the ground, made in applications of Pen- 

 man's (1948) method for the calculation of irrigation needs. However, 

 quite apart from the short term microclimatic variations, it seems highly 

 probable that significant differences of surface roughness would also qualify 

 the assumption. Equally, depending on the roughness of a crop, evapo- 

 transpiration can often exceed evaporation from a smooth surface of open 

 water, a result which was not predicted by earher workers in the subject. 

 One last method, suggested by Deacon and Swinbank (1956), might 

 overcome some of the earlier difficulties arising from stabiUty dependence. 

 The surface drag coefficient Cn, defined by: 



T = pCnUn^ (30) 



where Un is the wind speed at the height of measurement Zn, can be deter- 

 mined from wind profJes obtained in neutral conditions for a given site. 

 Then, since Kip=Km, 



^=-i^Zii (31) 



T M2"""l 



and substituting from eq. 30: 



E=pCnUn\^^^^^' (32) 



For convenience, Zn and z^ can be made identical, and Z2 chosen to be high 

 enough to obtain the required accuracy in the measurements. Cn does vary 

 sHghtly with stabiHty, but much less so than the roughness parameter Zo- 

 The essential feature of the method is that no assumptions are made con- 

 cerning the logarithmic profde law, only that the profiles have the same 

 form. Provided that Zn can be made sufficiently low to be independent of 

 StabiHty effects, eq. 32 will be vaHd for all stabilities. 



6. DATA-PROCESSING 



In my introduction, it was pointed out that a programme involving a 

 combination of the intricacies of the aerodynamic method with the 

 continuous control of energy balance considerations would yield a vast 

 amount of somewhat intractable data. The change to semi-automatic data- 

 processing, however, is not one which many physicists take to Hghtly since 

 in the first place some loss of accuracy is almost inevitable, and secondly 

 there are recognised stages in most experimental investigations where 

 discrimination, selection and rejection thought-processes are required. 

 These processes require the integrated experience and insight of the 



