See more from this Session: Sensor-Driven Digital Soil Mapping: II
Digital soil mapping techniques have shown much promise to reduce soil mapping costs and produce high-resolution soil maps covering large areas. Numerous pedometrical methods have been developed that use Geographic Information Systems (GIS), Global Positioning Systems (GPS), advanced statistical and geostatistical methods, and field data. For environmental protection, cost reduction and optimization of crop yield farmers need to know the spatial distribution of soil nutrient deficiencies and abundances. The objectives of this study were to assess the usefulness and efficacy of different spatial resolutions of remote sensing images (Landsat ETM+ and IKONOS) to predict soil phosphorus (P). The study was conducted in the Santa Fe River Ranch Beef Unit (SFRRBU), which is approximately 650 hectares in size and located in the northern part of Alachua County, Florida. A stratified random sampling design based on land use and soil order combinations was used to collect soil samples in four layers (0 to 30, 30 to 60, 60 to 120 and 120 to 180cm). This study compared an univariate method (Ordinary Kriging) with multivariate methods (Regression Kriging and Co-kriging) to predict and map geospatial distributions of soil P for each soil layer using remote sensing images and ancillary spatial environmental datasets. Results showed that multivariate methods with finer resolution of remote sensing produced better predictions of soil P. The predictive power of spectral data in the upper layers was higher than in lower layers. The soil P distribution and variation maps will be helpful for management of fields in the SFRRBU to optimize nutrient status and minimize adverse impacts on the environment.