Estimation of crop residue cover utilizing multiple ground truth survey techniques and multi-satellite regression models

Soil erosion within agricultural landscapes has a significant impact on environmental and economic systems and is strongly driven by the lack of residue cover in agricultural fields. Soil erosion models such as the Water and Erosion Prediction Project (WEPP) model and its large area implementation, the Daily Erosion Project, are important tools for understanding patterns of soil erosion, but they rely on the accurate estimation of crop residue cover over large regions to infer tillage practices. Remote sensing analyses are now becoming accepted as reliable way to estimate crop residue cover, but most use localized training datasets that are not practical outside of small study areas. An alternative source of training data may be commonly conducted tillage surveys that capture information via rapid 'windshield' surveys. In this study, we utilized Google Earth Engine to assess the utility of three crop residue survey types (windshield tillage surveys, windshield binned residue surveys, photo analysis surveys) and one synthetic survey (retroactively binned photo analysis data) as sources of training data for crop residue cover regressions. Overall, we found that Google Earth Engine was a viable platform for creating crop residue cover models, and the top performing models had a tillage classification accuracy of 64.5% with a Cohen’s Kappa value of 0.47. We also found that neither windshield-based survey dataset was able to produce reliable regressions, but windshield binned residue survey data produced a reliable distinction between low residue and high residue fields. Finally, we found that retroactively binned photo analysis survey data performed equally as well as the original dataset, which suggests that less labor-intensive methods of retrieving crop residue cover from images should also produce reliable training data.