Evaluating spatiotemporal fusion models for remote sensing monitoring of conservation tillage

Conservation tillage reduces soil disturbance and improves soil health. Due to its systematic, scalable, and cost-effective nature, remote sensing offers an efficient means to monitor conservation tillage. Reliable monitoring of conservation tillage remains a significant challenge because it requires remote sensing data with high spatial and temporal resolution, while optical observations can be hindered by cloudy and rainy weather conditions as well as by green vegetation. This study in central Indiana (USA) combined Landsat 8 OLI's high spatial resolution with Moderate Resolution Imaging Spectroradiometer (MODIS)'s high temporal resolution to estimate the normalized difference tillage index (NDTI) using three spatiotemporal fusion algorithms: spatial and temporal adaptive reflectance fusion model (STARFM), enhanced spatial and temporal adaptive reflectance fusion model (ESTARFM), and flexible spatiotemporal data fusion (FSDAF). Their performance was assessed in predicting NDTI and crop residue cover (CRC), validated with field data across subregions with different cropland ratios. Results showed that ESTARFM achieved the best overall performance for NDTI prediction, with coefficient of determination ( R 2 ) values of 0.56–0.70 and positive NSE values of 0.26–0.45, outperforming STARFM and FSDAF. The proportion of cultivated land was associated with model performance, but it was not the sole controlling factor. The 2015 spatial distribution of CRC revealed pronounced heterogeneity across croplands, with values ranging from 0% to 98% and higher values mainly concentrated in the northern part of the study area, demonstrating the method’s ability to characterize continuous residue-cover patterns. The integration of multi-source remote sensing and ESTARFM offers a practical approach for large-scale residue-cover monitoring.