Multitemporal remote sensing of crop residue cover and tillage practices: A validation of the minNDTI strategy in the United States

Abstract

Accurate, site-specific tillage information forms an important dimension for development of effective agricultural management practices and policies. Landsat Thematic Mapper (TM) imagery provides the opportunity for systematic mapping of tillage practices via crop residue (plant litter or senescent or nonphotosynthetic vegetation) cover (CRC) estimation at broad scales because of its repetitive coverage of the Earth's land areas over several decades. This study evaluated the effectiveness of a multitemporal approach using the minimum values of Normalized Difference Tillage Index (minNDTI) for assessing CRC at multiple locations over several years. Local models were generated for each dataset. In addition, we tested the feasibility of a regional model in mapping CRC. Results show that the minNDTI method was able to estimate CRC, and a regional model is possible. We found that in addition to the known impact of emergent green vegetation, soil moisture and organic carbon (C) can also confound the NDTI signal, thereby underestimating CRC for low-lying wet and dark areas. Accuracy of the minNDTI technique is comparable to the hyperspectral Cellulose Absorption Index (CAI) and the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Shortwave Infrared Normalized Difference Residue Index (SINDRI) for tillage classification. The minNDTI technique is currently the best for monitoring CRC and tillage practices from space, opening the door for generating field-level tillage maps at broad spatial and temporal scales.