Satellite mapping of conservation tillage adoption in the Little River experimental watershed, Georgia

Abstract

Conservation tillage is a commonly adopted best management practice for improving soil quality and reducing erosion. However, there are currently no methods in place to monitor conservation tillage adoption at the watershed scale. The primary objective of this study was to evaluate the usefulness of Landsat TM data as a tool to depict conservation tillage in a small Coastal Plain watershed. Satellite imagery was used to calculate four commonly used indices: Normalized Difference Vegetation Index, Crop Residue Cover Index, Normalized Difference Tillage Index, and the Simple Tillage Index. Ground truth data consisted of a windshield survey, assigning each site a tillage regime (conventional or conservation tillage) at 138 locations throughout the watershed and surrounding areas. A logistical regression approach was used on two subsets of the data set (n = 20 or n = 44) to determine the influence of the number of ground control points on the success of modeling the occurrence of conservation tillage. The most accurate model was re-applied to the satellite image and evaluated using an independent sample of 94 survey sites. Results indicate that the normalized difference tillage and simple tillage indices performed best, with an overall accuracy of 71% and 78% for models developed using n = 20 and n = 44 sample locations, respectively. Errors were typically in the form of commission. Results are encouraging and suggest that currently available satellite imagery can be used for rapid assessment of conservation tillage adoption using minimal a priori information.