J. R Hearst, P. W. Kasameyer, and L. A. Dreiling
Lawrence Livermore National Laboratory
Livermore, California
ABSTRACT
We have studied four sources of error (uncertainty)in water content obtained from neutron logs calibrated in partially saturated media for holes up to 3 m. For this calibration we built a special facility and developed an algorithm for a commercial epithermal neutron log that obtains water content from count rate, bulk density, and gap between the neutron sonde and the borehole wall. In our study we found errors in the facility itself: finite length of the cells making up the facility, inhomogeneous and nonrepresentative material within the cells, gaps between the cells, inappropriate borehole shape, and approximation of a large borehole by a flat face. The algorithm contained errors due to the calibration and lack of fit, while the field measurements included uncertainties in the count rate (caused by statistics and a short time constant), gap, and density. Finally, there can be inhomogeneity in the material surrounding the borehole. Under normal field conditions the hole-size-corrected water content obtained from such neutron logs can have an uncertainty as large as 15% of its value. Primary causes are uncertainty in gap (correctable by measuring at both ends of the sonde), uncertainty in count (reducible by averaging over depth), and lack of fit.