IKONOS Imagery to Estimate Surface Soil Property Variability in Two Alabama Physiographies

c-means clustering of remotely sensed (RS) data, and multiple linear regression relating spectral response to Knowledge of surface soil properties is used to assess past erosion soil attributes may be used to evaluate variability in and predict erodibility, determine nutrient requirements, and assess surface texture for soil survey applications. This study was designed to surface soil properties. evaluate high resolution IKONOS multispectral data as a soil-mapThe basic relationships between spectral response and ping tool. Imagery was acquired over conventionally tilled fields in soil properties have been well researched. Early studies the Coastal Plain and Tennessee Valley physiographic regions of Alahave shown a negative correlation exists between surbama. Acquisitions were designed to assess the impact of surface crustface TC and reflectance in the visible (VIS) and near infraing, roughness, and tillage on our ability to depict soil property variabilred (NIR) (Baumgardner et al., 1970; Sudduth and Humity. Soils consisted mostly of fine-loamy, kaolinitic, thermic Plinthic mel, 1991; Henderson et al., 1992). Increasing amounts Kandiudults at the Coastal Plain site and fine, kaolinitic, thermic of TC have a darkening effect, consequently reducing Rhodic Paleudults at the Tennessee Valley site. Soils were sampled the amount of energy reflected. Similarly, Coleman and in 0.20-ha grids to a depth of 15 cm and analyzed for percentages of Montgomery (1987) found a strong negative correlation sand (0.05–2 mm), silt (0.002–0.05 mm), clay ( 0.002 mm), citratedithionite extractable Fe, and total C (TC). Four methods of evaluat(r 0.58) between TC and NIR (0.76–0.90 m) reflecing variability in soil attributes were evaluated: (i) kriging of soil attance in Vertisols and Alfisols in Alabama’s Blackbelt tributes, (ii) cokriging with soil attributes and reflectance data, (iii) region. These authors noted increasing soil water conmultivariate regression based on the relationship between reflectance tent, coincident with increasing TC, tended to depress and soil properties, and (iv) fuzzy c-means clustering of reflectance surface reflectance and mask spectral features of interdata. Results indicate that cokriging with remotely sensed (RS) data est (Johnson et al., 1998). improved field scale estimates of surface TC and clay content comSoil texture also impacts soil spectral response curves. pared with kriging and regression methods. Fuzzy c-means worked In highly weathered and eroded soil systems of the best using remotely sensed data acquired over freshly tilled fields, Southeastern Coastal Plain and Tennessee Valley physreducing soil property variability within soil zones compared with iographic regions, the sand (0.05–2 mm) fraction is prifield scale soil property variability. marily composed of quartz with lesser quantities of mica, and clay ( 0.002 mm) particles consist of kaolinite, with lesser quantities of hydroxy–interlayered vermiculite, Fe S soil properties are often used to assess soil oxides, and gibbsite (Shaw et al., 2002, 2003). In eroded quality, establish soil survey map units, and detersoils, increasing clay quantities at the surface attenuate mine agrochemical application rates. Current soil samreflectance as finer particles cause scattering of energy pling methods designed to capture field scale variability (Mathews et al., 1973; Stoner and Baumgardner, 1981; include grid-sampling and directed sampling using manSalisbury and D’Aria 1992). Barnes and Baker (2000) agement zones. In a grid-sampling approach, grids are used multispectral airborne and satellite data to create created in an attempt to assess spatial variability (Fransurface soil texture maps for two sites at the Maricopa zen and Peck, 1995). Depending on field size and variResearch Farm in Arizona. Due to variability in soil ability, an accurate assessment of soil properties is best water content, surface roughness, and residue cover beachieved through a densely sampled grid, making spatially tween sites, RS-derived soil texture maps were most representative estimates cost-prohibitive. While manageaccurate when generated on a site-by-site basis. Thomasment zone (directed) sampling shows promise, represenson et al. (2001) reported similar findings, which showed tative zones are best developed over time using a comthe relationship between spectral response and soil texbination of data layers such as yield, topography, and ture was highly variable between two farm sites in Missoil maps (Franzen et al., 1998). Newly available highsissippi. However, using spectra within the 0.40to 0.80resolution satellite imagery may discriminate among difand 0.95to 1.05m ranges 50% of the variability in ferences in surface soil attributes. Methods such as cosoil texture was explained. kriging soil samples with highly correlated spectra, fuzzy Soil spectral response relative to the amount of mineral, organic, and water content has been well estabD.G. Sullivan, USDA ARS Southeast Watershed Research Lab., P.O. lished. However, extrapolation to field conditions is conBox 748, Tifton, GA 31794; J.N. Shaw, Dep. of Agronomy and Soils, Auburn University, 202 Funchess Hall, Auburn, AL 36849; D. Rickfounded by variability in surface roughness, crop residue man, Global Hydrology and Climate Center, NSSTC/MSFC/NASA, cover, crusting, and soil water content. Several methods 320 Sparkman Drive, Huntsville, AL 35805. Received 7 Mar. 2005. show potential for improving our ability to discriminate *Corresponding author ([email protected]). among changes in surface soil attributes. Two of these Published in Soil Sci. Soc. Am. J. 69:1789–1798 (2005). Soil & Water Management & Conservation doi:10.2136/sssaj2005.0071 Abbreviations: NIR, near infrared; PAN, panchromatic; RMSE, root mean square error; RS, remote sensing/sensed; TC, total carbon; VIS, © Soil Science Society of America 677 S. Segoe Rd., Madison, WI 53711 USA visible. 1789 Published online September 29, 2005