THROUGHFALL IN FOREST STANDS 131 



The importance of total throughfall in a stand is that it often represents 

 all the water available for transpiration and drainage. Its accurate measvire- 

 ment is therefore of considerable hydrological significance. The presence 

 of a pattern of throughfall distribution, however, introduces certain 

 complications which have not always been appreciated by workers in this 

 field. 



MEASURING TOTAL THROUGHFALL 



Because of the spatial variability of throughfall, a sampling technique must 

 be selected which provides an estimate with an appropriate statistic of this 

 variability ; in the present case we will employ the standard error expressed 

 as a percentage of the mean. One object of the present investigation was to 

 compare the efficiencies of various measuring techniques. 



Some difficulty has been experienced in the past in keeping the standard 

 error of throughfall estimates, exclusive of stem flow, sufficiently low and 

 yet maintain a manageable set up. 



Wilm and Niederhof (1941) found that estimates based on stationary 

 gauges in forest stands were subject to large standard errors. To improve 

 the samphng procedure, Wilm (1943) employed a single 8-in, gauge in 

 each experimental plot, moving it after every storm, to one of twelve 

 random positions. However, this experimental design did not achieve the 

 desired precision, and calculations suggested that the number of alternative 

 positions in each plot should have been increased to forty. 



In Yorkshire, Law (1957) adopted a somewhat similar approach, but 

 used a rather restricted randomisation of ten 5-in. gauges beneath a stand 

 of Sitka spruce, moving each of these after approximately equal amounts 

 of rain to one of thirty-six alternative positions. Over a period of a whole 

 year, though not for shorter periods, he obtained a standard error of almost 

 12%: for a strictly random (but less practicable) design, the corresponding 

 figure was 6-4%. 



In the present investigation with twenty 5-in. gauges sited at random in 

 the plot, large standard errors were also found (simple S.E., Table 3). 



An attempt has been made to increase the precision of the estimates from 

 stationary gauges by taking into account the relation between catch and 

 the distance from the stem, as presented in Table i. From the appropriate 

 regression equation, the catch was calculated for the mean position under 

 the canopy (estimated for a large number of random locations), and the 

 standard error of this value determined, taking into account the variance 

 of the estimate of mean position. Niederhof and Wilm (1943) used a some- 

 what similar approach but over a much greater range of canopy conditions. 



