TY - JOUR
T1 - Multi-sensor assessment of phenology-based field-level cover cropping detection using satellite vegetation time series from Harmonized Landsat-8 and Sentinel-2, MODIS, and PlanetScope
AU - Zhou, Qu
AU - Guan, Kaiyu
AU - Wang, Sheng
AU - Wu, Xiaocui
AU - Stroebel, Samuel
AU - Hipple, James
N1 - This study was primarily supported by the USDA National Institute of Food and Agriculture (NIFA) Foundational Program awards, US Department of Energy 's Advanced Research Projects Agency -Energy ( ARPA-E ) SMARTFARM ( MBC Lab and SYMFONI) projects, and NSF CAREER Award by Environmental Sustainability Program. This study was also supported in part by the C3.ai Digital Transformation Institute and NASA Early Career Investigatior Program in Earth Science (80NSSC24K1057). K.G. and Q.Z. acknowledge the support from the NASA FINESST award. Q.Z. and S.W. acknowledge the support from the ESA NoR (Network of Resources) Sponsorship. Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the views of the above U.S. government agencies.
PY - 2026/2
Y1 - 2026/2
N2 - Remote sensing-detected phenological differences among cover crops, cash crops, and soil backgrounds are crucial for detecting cover cropping practices. However, variations in detection performance with satellite data of different spatial, temporal, and radiometric characteristics and the potential factors affecting the performance remain unclear. Building on the previously developed framework, we aimed to evaluate the performance of multiple satellite sensors, leverage multi-scale ground truth data, and identify potential factors for improving field-level cover cropping detection. Specifically, we (1) compared widely-used NDVI time series from Harmonized Landsat-8 and Sentinel-2 (HLS, ∼3-day and 30 m), MODIS (daily and 250 m), and PlanetScope (near-daily and 3 m, requiring radiometric calibration), (2) quantified discrepancies between county-level data from National Agricultural Statistics Service (NASS) and field-level data from Indiana State Department of Agriculture (ISDA), and (3) analyzed the impacts of cover cropping field sizes, regional adoption rates, and cover crop species on cover cropping detection. We found that (1) HLS outperformed MODIS and MODIS-calibrated PlanetScope (CPS) with an average accuracy of 76.2 % ± 3.0 % across Indiana in 2017. (2) Significant discrepancies between ISDA and NASS data were found in 9 % of counties and were negatively correlated with detection accuracies in counties where ISDA-reported adoption was substantially higher than that reported by NASS (r = -0.54, p < 0.005). (3) Detection accuracy was higher in larger cover cropping fields, positively correlated with regional adoption rates (r = 0.42 and p < 0.001), with the highest accuracy for wheat (88.95 %), followed by winter grains (77.73 %), ryegrass (75.52 %), barley (75.35 %), and cereal rye (68.59 %). Our study offers valuable insights into selecting satellite sensors, reconciling multi-scale ground truth data, and identifying potential factors for improving phenology-based field-level cover cropping detection.
AB - Remote sensing-detected phenological differences among cover crops, cash crops, and soil backgrounds are crucial for detecting cover cropping practices. However, variations in detection performance with satellite data of different spatial, temporal, and radiometric characteristics and the potential factors affecting the performance remain unclear. Building on the previously developed framework, we aimed to evaluate the performance of multiple satellite sensors, leverage multi-scale ground truth data, and identify potential factors for improving field-level cover cropping detection. Specifically, we (1) compared widely-used NDVI time series from Harmonized Landsat-8 and Sentinel-2 (HLS, ∼3-day and 30 m), MODIS (daily and 250 m), and PlanetScope (near-daily and 3 m, requiring radiometric calibration), (2) quantified discrepancies between county-level data from National Agricultural Statistics Service (NASS) and field-level data from Indiana State Department of Agriculture (ISDA), and (3) analyzed the impacts of cover cropping field sizes, regional adoption rates, and cover crop species on cover cropping detection. We found that (1) HLS outperformed MODIS and MODIS-calibrated PlanetScope (CPS) with an average accuracy of 76.2 % ± 3.0 % across Indiana in 2017. (2) Significant discrepancies between ISDA and NASS data were found in 9 % of counties and were negatively correlated with detection accuracies in counties where ISDA-reported adoption was substantially higher than that reported by NASS (r = -0.54, p < 0.005). (3) Detection accuracy was higher in larger cover cropping fields, positively correlated with regional adoption rates (r = 0.42 and p < 0.001), with the highest accuracy for wheat (88.95 %), followed by winter grains (77.73 %), ryegrass (75.52 %), barley (75.35 %), and cereal rye (68.59 %). Our study offers valuable insights into selecting satellite sensors, reconciling multi-scale ground truth data, and identifying potential factors for improving phenology-based field-level cover cropping detection.
KW - Cover cropping
KW - HLS
KW - MODIS
KW - Phenology
KW - PlanetScope
KW - Time series
UR - https://www.scopus.com/pages/publications/105023645904
UR - https://www.scopus.com/pages/publications/105023645904#tab=citedBy
U2 - 10.1016/j.jag.2025.105004
DO - 10.1016/j.jag.2025.105004
M3 - Article
AN - SCOPUS:105023645904
SN - 1569-8432
VL - 146
JO - International Journal of Applied Earth Observation and Geoinformation
JF - International Journal of Applied Earth Observation and Geoinformation
M1 - 105004
ER -