Estimates of conservation tillage practices using Landsat archive

Authors: BEESON, P. - GLOBAL CONSERVATION INSTITUTE; DAUGHTRY, CRAIG; WALLANDER, S. - ECONOMIC RESEARCH SERIVCE (ERS, USDA)

Submitted to: Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/15/2020
Publication Date: 8/18/2020
Citation: Beeson, P.C., Daughtry, C.S., Wallander, S.A. 2020. Estimates of conservation tillage practices using Landsat archive. Remote Sensing. 12(16):2665. https://doi.org/10.3390/rs12162665.
DOI: https://doi.org/10.3390/rs12162665

Interpretive Summary: The USDA Environmental Quality Incentives Program (EQIP) provides technical and financial assistance to encourage producers to adopt conservation practices. Historically, one of the most common practices is conservation tillage, which protects the soil with more crop residue cover than intensive tillage. This research identified agricultural fields with crop residue cover in the 58,000 square mile study area in South Dakota, North Dakota, and Minnesota using 10 years of Landsat Thematic Mapper data. The results were validated against field-level survey data. This study demonstrated that researchers can implement retrospective estimates of conservation tillage with sufficient accuracy using the Landsat Archive which is available at no cost.

Technical Abstract: The USDA Environmental Quality Incentives Program (EQIP) provides financial assistance to encourage producers to adopt conservation practices. Historically, one of the most common practices is conservation tillage, primarily the use of no-till planting. The objectives of this research were to determine crop residue using remote sensing, an indicator of tillage intensity, without using training data and examine its performance at the field level. The Landsat Thematic Mapper Series platforms can provide global temporal and spatial coverage beginning in the mid-1980s. In this study, we used the Normalized Difference Tillage Index (NDTI), which has proved to be robust and accurate in studies built upon training datasets. We completed 10 years of residue maps for the 150,000 km2 study area in South Dakota, North Dakota, and Minnesota and validated the results against field-level survey data. The overall accuracy was between 64%–78% with additional improvement when survey points with suspect geolocation and satellite tillage estimates with fewer than four dates of Landsat images were excluded. This study demonstrates that, with Landsat Archive available at no cost, researchers can implement retrospective, untrained estimates of conservation tillage with sufficient accuracy for some applications.