2003 Seattle Annual Meeting (November 2–5, 2003)

Paper No. 4
Presentation Time: 8:55 AM

A COMPARISON OF THE USE OF RADAR IMAGES AND NEUTRON PROBE DATA TO DETERMINE SPATIAL VARIABILITY IN WATER CONTENT AT HANFORD


KNIGHT, Rosemary1, IRVING, James1, TERCIER, Paulette2, FREEMAN, Gene3, MURRAY, Chris4 and ROCKHOLD, Mark3, (1)Geophysics, Stanford Univ, Mitchell Building, Room 360, Stanford, CA 94305, (2)7764 Elford Ave, Burnaby, BC V3N-4B7, Canada, (3)Pacific Northwest National Lab, P.O. Box 999, MS K9-36, Richland, WA 99352, (4)Pacific Northwest National Lab, P.O. Box 999, MS K6-81, Richland, WA 99352, rknight@stanford.edu

Surface-based ground penetrating radar (GPR) data were collected at the Sisson and Lu Injection Test Site in the 200 East Area at Hanford to assess the use of radar images as a means of quantifying the subsurface distribution in water content (qw). Available at the test site were two sets of water content data derived from neutron probe measurements that had been made to a depth of ~18 m in 32 wells.

The first question we asked: Do the locations and amplitudes of the reflections in the radar data correspond to changes in qw? In order to address this issue, the probe-derived water content data were used to generate synthetic radar data, i.e. the radar data that would be obtained if qw alone were controlling the radar response. Many of the features in the synthetic radar data were found to be present in the acquired 30 m radar section indicating that the amplitudes and locations of the radar reflections do correspond to changes in subsurface qw. This observed correspondence led to the second part of the study. We tested the idea that quantifying the spatial distribution of the radar reflections, in terms of the horizontal correlation structure, could be used to quantify the correlation structure in subsurface qw.

Geostatistical analysis was conducted of the two sets of qw- values and of the amplitudes of the reflections in the radar section. The geostatistical analysis of the radar reflection image yielded a correlation structure similar to that obtained from neutron probe-derived qw values, with a correlation length on the order of 10 to 14 m seen in both data sets. The limited horizontal extent of both the neutron probe data (~16 m) and radar data (30 m), relative to the correlation length, suggests that the true correlation length of the subsurface water content may be underestimated in modeling the semivariograms of both data sets. The radar data, with a sample spacing of 20 cm (compared to ~2 m for the probe data) provided additional evidence of structure at the sub-meter scale. We conclude that surface-based radar should be further explored for use at Hanford and at other sites where measurements obtained in wells are insufficient to provide the required detailed information about spatial variability in qw.