on neutron measurements made at a given 

 depth was the difference, measured horizontal- 

 ly, between the two curves. This difference is 

 referred to as measurement error and is ex- 

 pressed in percent moisture by volume. 



Measurement errors were determined for 

 each inch of depth for each graph and were 

 plotted on total error graphs as the horizontal 

 deviation from a vertical axis (figs. 7 through 

 10). The area under the error curve is the sum- 

 mation of the products of each measurement 

 error and the inch of depth which each repre- 

 sents. Total error can be thought of as the total 

 error in measurement (in inches of water) if 

 neutron measurements had been made at each 

 inch through the zone of influence around each 

 void. The purpose of the computation of total 

 error is to quantify the relative difference in the 

 effect of a void as the water level changes. 



Small Void-Coarse Sand 



Figure 7A shows the results of measure- 

 ments from the two tanks when the water levels 

 were near the tops of the tanks; i.e., the area 

 around the small void is saturated completely. 

 The error in measurement at 24 inches below 

 the surface was +17.5 percent by volume cal- 

 culated by subtracting 45.0 percent for the 

 control tank from 62.5 percent for the sample 

 tank. From the surface of the sand down to a 

 depth of 13 inches there is no measurement 

 error due to the void. If readings were to be 

 taken at each inch from 13 inches to 33 inches, 

 the total error would be +1 .86 inches. 



Figure 7B shows the two curves when the 

 water level had dropped to 18.75 inches below 

 the surface. The void was still saturated and 



1.4 1 1 1 1 1 1 1 1 1 r 



qI I I 1 I I I I I 



10 20 30 40 50 60 70 80 90 100 



SOIL MOISTURE-PERCENT BY VOLUME 



Figure 6. — Troxler calibration curve for a model 104 probe. 



7 



