Infrared Measurements of Pristine and Disturbed Soils 1. Spectral Contrast Differences between Field and Laboratory Data

Abstract

Comparison of emissivity spectra (8–13 μm) of pristine soils in the field with laboratory reflectance spectra of the same soils showed that laboratory spectra tend to have less spectral contrast than field spectra (see following article). We investigated this phenomenon by measuring emission spectra of both undisturbed (in situ) and disturbed soils (prepared as if for transport to the laboratory). The disturbed soils had much less spectral contrast than the undisturbed soils in the reststrahlen region near 9 μm. While the increased porosity of a disturbed soil can decrease spectral contrast due to multiple scattering, we hypothesize that the effect is dominantly the result of a difference in grain-size distribution of the optically active layer (i.e., fine particle coatings). This concept was proposed by Salisbury et al. 1994to explain their observations that soils washed free of small particles adhering to larger grains exhibited greater spectral contrast than unwashed soils. Our laboratory reflectance spectra of wet- and dry-sieved soils returned from field sites also show greater spectral contrast for wet-sieved (washed) soils. We therefore propose that undisturbed soils in the field can be characterized as “clean” soils (washed free of fine particles at the surface due to rain and wind action) and that disturbed soils represent “dirty” soils (contaminated with fine particle coatings). The effect of packing soils in the field and laboratory also increases spectral contrast but not to the magnitude of that observed for undisturbed and wet-sieved soils. Since it is a common practice to use laboratory spectra of field samples to interpret spectra obtained remotely, we suggest that the influence of fine particle coatings on disturbed soils, if unrecognized, could influence interpretations of remote sensing data.Published by

Introduction

One of the assumptions often made in remote sensing is that the spectral character of geologic materials observed in the field will be similar for the same samples when returned and analyzed in the laboratory. However, slight spectral differences between field and laboratory measurements are often observed. For example, emittance spectra derived from infrared (7–14 μm) observations using the Thermal Infrared Multispectral Scanner (TIMS) often exhibit lower spectral contrast in the reststrahlen region than laboratory spectra of rocks and minerals acquired from the same site (e.g.,Rivard et al. 1993; Hook et al. 1994; Collins 1991). While this may result from differences in instrumentation, observing conditions (atmospheric, geometric), or spatial/spectral resolution between lab and field data, it is more likely that the lower spectral contrast of field data is due to the occurrence and physical state of soils within the scene. If laboratory analyses of soils are not included with those of rocks and minerals, spectral differences between field and lab data should be expected.

Further, in regions dominated by soil cover the physical condition of the surface soils will influence the observed spectral character. We have observed greater spectral contrast in the reststrahlen region for infrared field emission spectra of bare, pristine soils than for laboratory reflectance spectra of soil samples obtained from the same site. Considering that the sampled soils are no longer in a pristine state when analyzed in the laboratory, we have investigated this phenomenon further by measuring emission spectra of both pristine and disturbed soils in the field. In nearly all cases, the spectral contrast of disturbed soils is less than that of the undisturbed soils (cf.Winter et al. 1996).

Salisbury and Wald 1992and Salisbury et al. 1994showed that laboratory spectra of soils cleaned in an ultrasonic bath show greatly enhanced spectral contrast relative to “natural” soils. Following this idea, we wet-sieved soil samples returned from the field to remove silt- and clay-sized fines and found that most of the dried, wet-sieved soils also showed greatly enhanced reflectance contrast relative to soils dry-sieved. We hypothesize that this difference in contrast between undisturbed and disturbed soils in the field is mainly due to the influence of fine-particle coatings in the optically active layer of disturbed soils. Photomicrographs of soils show that very fine grains (e.g., <10–20 μm) cling to all larger particles in natural soils (e.g.,Salisbury et al. 1994), but that cleaned soils have few or no fine particles. Since spectral contrast is lower for smaller grain sizes in the reststrahlen region, the effect of these fine-particle coatings is to shield the underlying larger grains, resulting in lower spectral contrast more characteristic of the fine grains themselves (e.g.,Salisbury et al. 1994).

We show examples of this phenomenon using field emission spectra and laboratory reflectance spectra of returned soils from several sites across the United States (Nevada Test Site, Nevada; Ft. Huachuca, Arizona; Camp Lejeune, North Carolina). The results suggest that bare soils exposed in the field for periods of weeks or months are cleaned of fine particles, probably due to rain impact and wind, exposing surfaces of optically large grains. When the soils are disturbed, the fine material which is abundant in the subsurface coats the larger grains, reducing the spectral contrast substantially. We also show that the increased spectral contrast of packed soils in the field and laboratory does not fully explain these observations.