Characterization of Soil Spatial Variability for Site-Specific Management Using Soil Electrical Conductivity and Other Remotely Sensed Data

Abstract

Field-scale characterization of soil spatial variability using remote sensing technology has a great potential for achieving the successful implementation of site-specific management (SSM). The objectives of this study were to: (i) examine the spatial relationships between apparent soil electrical conductivity (ECa) and soil chemical and physical properties to determine if ECa could be useful to characterize soil properties related to crop productivity in the Coastal Plain and Piedmont of North Carolina; (ii) evaluate the effects of in-situ soil moisture variation on ECa mapping as a basis for characterization of soil spatial variability and as a data layer in cluster analysis as a means of delineating sampling zones; (iii) evaluate clustering approaches using different variable sets for management zone delineation to characterize spatial variability in soil nutrient levels and crop yields. Field studies were conducted in two physiographic regions of North Carolina, the Piedmont and Coastal Plain. Spatial measurements of ECa via electromagnetic induction (EMI) were compared with soil chemical parameters (extractable P, K, and micronutrients; pH, cation exchange capacity (CEC), humic matter or soil organic matter; and physical parameters (percentage sand, silt, and clay; and plant-available water (PAW) content; bulk density; cone index; saturated hydraulic conductivity [Ksat] in one of the coastal plain fields) using correlation analysis across fields. We also collected ECa measurements in one coastal plain study field on four days with significantly different naturally occurring soil moisture conditions measured in five increments to 0.75 m using profiling time-domain reflectometry probes to evaluate the temporal variability of ECa associated with changes in in situ soil moisture content. Nonhierarchical k-means cluster analysis using sensor-based field attributes including vertical ECa, near-infrared (NIR) radiance of bare-soil from an aerial color infrared (CIR) image, elevation, slope, and their combinations was performed to delineate management zones.

The strengths and signs of the correlations between ECa and measured soil properties varied among fields. Few strong direct correlations were found between ECa and the soil chemical and physical properties studied (r2 < 0.50), but correlations improved considerably when zone mean ECa and zone means of selected soil properties among ECa zones were compared. The results suggested that field-scale ECa survey is not able to directly predict soil nutrient levels at any specific location, but could delimit distinct zones of soil condition among which soil nutrient levels differ, providing an effective basis for soil sampling on a zone basis.

The strengths of the correlations of ECa with measured soil properties varied depending on soil moisture conditions. In general, the strongest correlations were observed when ECa was measured under relatively dry conditions. The results suggest that the spatial and temporal ECa variability measured under different soil moisture conditions could be a critical factor when evaluating the ability of ECa to predict soil chemical and physical characteristics important to soil and crop productivity and management.

The relationships of sensor-based field attributes as clustering variables to group soil test parameters and crop yields varied among fields. Most of the clustering data combinations effectively captured the within-zone variability of soil test parameters and crop yield, but the use of different variables for cluster analysis resulted in different results in terms of capturing soil test and yield variability by management zones. The results indicate that zones created by cluster analysis could provide a way to group and manage spatial variability of soil nutrients within fields; however, appropriate selection of clustering variables for each individual field could be critical to implement cluster analysis for developing management zones.