Neutron Probe Calibration in Unsaturated Tuff A. N. Elder and Todd C. Rasmussen* ABSTRACT The measurement of water content in unsaturated media using neutron moisture probes requires a calibration relationship between neutron counts and water contents. Techniques for obtaining the relationship in unconsolidated geologic media may be unsuitable for consolidated media due to the difficulty of extracting undisturbed samples for water content analysis. The calibration relationship, 9 = a, + (01 + o 2 p b )C + ospb, with C = neutron counts, and p b = dry bulk density, provided a good predictor of volumetric water content, 6, for consolidated media at the Apache Leap Tuff Site. Four techniques were employed to obtain data necessary to construct this relationship. Two methods employed a neutron transport code to predict neutron counts from scattering and absorption neutron cross sections. The first method calculated cross sections from elemental compositions, while the second used cross sections obtained from rock samples placed in a graphite pile containing a neutron source. The third method used in situ neutron counts measured in two access holes constructed in unsaturated rock with known water contents estimated from rock fragments excavated surrounding the access holes using a pneumatic hammer. The fourth technique used neutron counts from crushed rock packed into containers and maintained at prescribed water contents. Neutron cross sections obtained from elemental analyses were smaller than graphite-pile values, although the differences were insignificant at the 95% confidence level and the difference in rock water content predictions was small. The neutron transport model provided neutron count estimates consistent with in situ counts and counts from saturated crushed rock in containers. K NOWLEDGE of the in situ water content of unsaturated geologic media provides important characterization information for use in understanding and predicting fluid flow and solute transport in such media. Neutron attenua- tion techniques are commonly employed for measuring water contents and the change in water content with time. Converting neutron attenuation measurements to water contents requires calibration that may be unique to the site of interest. The calibration is a function of the neutron probe design, the properties of the geologic material, and the access hole geometry. The probe design affects the calibration relationship due to differences in the strength of the neutron source, the size and composi- tion of the neutron detector, and the position of the detector relative to the source (Schmugge et al., 1980). The calibration relationship is also affected by the access hole geometry such as the access hole diameter, the type and geometry of the casing material, and the type and density of backfill material (Zuber and Cameron, 1966). The material properties that may affect the calibration ' relationship include the amount of H in forms other than free water (e.g., hydrocarbons), the presence of neutron-absorbing materials such as B, and the bulk A.N. Elder, Tucson Water, 310 W. Alameda St., Tucson, AZ 85701; and T.C. Rasmussen, Warnell School of Forest Resources, University of Georgia, Athens, GA 30602. Work was performed at Dep. of Hydrology and Water Resources, Univ. of Arizona, Tucson. Received 21 June 1993. "Corresponding author (trasmuss@uga.cc.uga.edu). Published in Soil Sci. Soc. Am. J. 58:1301-1307 (1994). density and temperature of the medium (Visvalingam and Tandy, 1972). Wilson (1988a,b) has shown that variations in bulk density and elemental composition can significantly affect the calibration relationship. Calibra- tion of water contents to neutron counts (cf. Carneiro and De Jong, 1985) in rock is often difficult due to problems associated with extracting undisturbed rock samples. Existing augering and drilling methods (cf. Ruprecht and Schofield, 1990) may be inappropriate for correlating field estimates of water contents due to the large thermal and hydraulic disturbances induced using these methods. One method for obtaining the calibration relationship is to use rock chips placed in containers. The thermalization properties are then determined for a range of water contents. Because me size of the container may be smaller man the size required to thermalize the energetic neutrons produced by the source to lower energy levels measurable by the neutron probe, fast neutrons may escape the container and bias the calibration relationship. Vachaud et al. (1977) recommended the use of an alternate calibration relationship that employs calculated or measured nuclear cross sections for calibration of heavy soils or for soils where the bulk density varies rapidly with depth. Wilson and Ritchie (1986) demon- strated the utility of this method for various soils. No application to unsaturated rock has been made, however. It was the purpose of this study to develop a calibration relationship between neutron counts and water contents for unsaturated rock. This relationship was estimated using four methods that have previously been applied to soil materials, but not to unsaturated rock. A neutron transport model (Olgaard, 1965) was modified to predict neutron counts employing neutron cross sections obtained from elemental analyses of Apache Leap Tuff. Neutron transport cross sections were also estimated from rock samples placed near a 111-Bq (3-Ci) neutron source in a graphite pile. A third method consisted of neutron probe measurements obtained from crushed Apache Leap Tuff packed into containers and maintained at known water contents. The final method used in situ neutron probe measurements in shallow access holes placed in unsaturated rock. THEORY Neutron probes provide a source of fast, high-energy neu- trons that move radially outward from the source. As fast neutrons pass through matter, the neutrons interact with atomic nuclei and are randomly scattered. Each collision between a neutron and a nucleus results in a transfer of energy from the neutron to the nucleus. An average index of the energy reduc- tion per collision that accounts for all impact angles is (Cam- eron, 1970): 1; * 2/(A + 2/3) [1] where 4 is the average logarithmic energy loss per collision and A is the atomic mass of the impacted nucleus. Of the 1301