Using remote sensing data to evaluate surface soil properties in Alabama ultisols

Evaluation of surface soil properties via remote sensing could facilitate soil survey mapping, erosion prediction, and allocation of agrochemicals for precision management. The objective of this study was to evaluate the relationship between soil spectral signature and surface soil properties in conventionally managed row crop systems. High-resolution remote sensing data were acquired over bare fields in the Coastal Plain, Appalachian Plateau, and Ridge and Valley provinces of Alabama using the Airborne Terrestrial Applications Sensor multispectral. scanner. Soils ranged from sandy Kandiudults to fine textured Rhodudults. Surface soil samples (0 to 1 cm) were collected from 161 sampling points for gravimetric soil water content, soil organic carbon, particle size distribution, and citrate dithionite extractable iron content. Surface roughness and crusting were also measured during sampling. Two methods of analysis were evaluated: (1) multiple linear regression using common spectral band ratios and (2) partial least-squares regression. Our data show that thermal infrared spectra are highly, linearly related to soil organic carbon, sand and clay content. Soil organic carbon content was the most difficult to quantify in these highly weathered systems, where soil organic carbon was generally < 1.2%. Estimates of sand and clay content were best using partial least-squares regression at the Valley site, explaining 42 to 59% of the variability. In the Coastal Plain, sandy surfaces prone to crusting limited estimates of sand and clay content via partial least-squares and regression with common band ratios. Estimates of iron oxide content were a function of mineralogy and best accomplished using specific band ratios, with regression explaining 36 to 65% of the variability at the Valley and Coastal Plain sites, respectively.