Mapping Soil Swell Potential by Reflectance Spectroscopy

The steeply dipping Pierre Shale along the Front Range of Colorado provides a major challenge to geotechnical engineers and builders because of the heterogeneous nature of the expansive soil characteristics. Differential heaving is the common result of the failure of standard soil testing to detect the rapid lateral changes in swell potential. Smectite clay structures attract water into the interlayers causing changes in volume by as much as a factor of two.

We are developing a method to use reflectance spectroscopy in the 1800-2400 nm wavelength region to identify clay composition and abundance as well as water content that could be used in the field to make determinations in seconds and hence augment standard techniques for better area coverage in construction sites. This spectral region contains overtone absorption features for water and Al, Mg, Fe-OH in the octahedral sheet of phyllosilicates. XRD and the PVP-55 surface area method are used to determine smectite content. Partial least squares analysis is used to develop regressions between reflectance spectra and smectite and water content as well as standard geotechnical soil tests. We opened a 5 m deep, 30 m long trench perpendicular to the strike of the nearly vertical beds of the Pierre Shale adjacent to Roxbourough Park East, SW of Denver.

Approximately 29,000 reflectance spectra in 5cm intervals were acquired in the 400-2500 nm region, in situ along the trench walls. The spectra are recorded in 1 nm intervals, 2100 in all for each point. The data were combined into an image to facilitate data manipulation with existing hyperspectral image analysis software. We have identified smectite, illite, jarosite, gypsum and calcite from visual interpretation of the reflectance spectra. 250 core samples were acquired from the trench. Of the 100 that have been tested, the Chen's swell index exhibits the best correlation with reflectance spectroscopy. A new trench was opened up near Golden, CO in July 2003.