Heuristic, fuzzy landscape modeling has become a standard technique for digital soil mapping (MacMillan et al., 2000; Zhu et al., 2001).  However, most resultant soil-landscape models have been constructed using field classification of soil profiles with scant soil characterization data.  Recent studies have demonstrated the potential of proximal visible and near infrared (VNIR) diffuse reflectance spectroscopy for rapid, nondestructive, and inexpensive soil characterization (Dunn et al., 2002; Shepherd and Walsh, 2002).  Our goal in this study was to evaluate the potential application of proximal VNIR spectroscopy to soil-landscape modeling by comparing soil-landscape models constructed using (i) conventional lab data; and (ii) VNIR diffuse reflectance spectroscopy.

The Rock Creek Watershed within the Beartooth Mountains, Montana is a 47500-ha area with steep slopes and forested terrain underlain by granitic gneiss and glacial deposits in valley bottoms.  Landform segmentation was applied to a 10-m USGS digital elevation model (DEM), resampled to 15 m, following the fuzzy, heuristic approach of MacMillan et al. (2000).  The heuristic fuzzy rules of MacMillan et al. (2000) were revised and adapted to steep, mountainous terrain.

Surface (1 to 10 cm) and subsurface (20 to 30 cm) soil samples were collected from 238 spatially distributed locations.  Conventional laboratory techniques were used to analyze soil pH and texture for all samples and total carbon by combustion for surface samples.  Clay mineralogy was determined by x-ray diffraction for a subset of samples.  Visible and near infrared spectral reflectance for all samples was measured using an ASD “Fieldspec® Pro FR” VNIR spectroradiometer (Analytical Spectral Devices, Boulder, CO) with a spectral range of 350 to 2500 nm.  A 4000-sample global spectral library (Brown et al., 2005) and boosted regression trees were used to develop spectral calibrations for all targeted soil properties.

Soil properties were estimated for landscape elements using mixed-effects ANOVA models.  Results are presented and compared for landscape model parameter estimation using soil properties obtained from both conventional laboratory techniques and VNIR spectroscopy.

-

References

Brown, D.J., Shepherd, K.D., Walsh, M.G., Mays, M.D. and Reinsch, T.G., 2005. Global soil characterization using a VNIR diffuse reflectactance spectroscopy. Geoderma (in press).

Dunn, B.W., Beecher, H.G., Batten, G.D. and Ciavarella, S., 2002. The potential of near-infrared reflectance spectroscopy for soil analysis - a case study from the Riverine Plain of south-eastern Australia. Australian Journal of Experimental Agriculture, 42(5): 607-614.

MacMillan, R.A., Pettapiece, W.W., Nolan, S.C. and Goddard, T.W., 2000. A generic procedure for automatically segmenting landforms into landform elements using DEMs, heuristic rules and fuzzy logic. Fuzzy Sets and Systems, 113(1): 81-109.

Shepherd, K.D. and Walsh, M.G., 2002. Development of reflectance spectral libraries for characterization of soil properties. Soil Science Society of America Journal, 66(3): 988-998.

Zhu, A.X., Hudson, B., Burt, J., Lubich, K. and Simonson, D., 2001. Soil mapping using GIS, expert knowledge, and fuzzy logic. Soil Science Society of America Journal, 65(5): 1463-1472.