Soil Reflectance Properties: Ground-based Versus Remotely Sensed Measurements

Soil samples were collected on a 0.4-ha (1.0-ac) grid from two agricultural fields in northeastern Mississippi. The samples were measured with a spectrophotometer for diffuse reflectance between 250 and 2500 nm. Remote-sensing and topographic data of the fields were also collected. The objective was to determine whether remote-sensing data, along with other easily obtainable data like topography, could be used to predict the reflectance spectra of soils from small sample sites in agricultural fields. From the raw reflectance spectra, 50-nm-band averages were calculated. Significant correlations existed between laboratory soil-reflectance spectra and models comprised of remote-sensing and topographic data. The highest coefficient of determination was around 0.5. Inclusion of an extra multispectral image improved the correlation between selected models and spectral variables. Coarser resolutions of remote-sensing data appeared to be better than finer resolutions at predicting the laboratory reflectance spectra of soil samples. Site-specific field elevation data added some information to predictions. Remote-sensing bands were predictive of spectral variables in the same spectral region.