Field-scale monitoring of cover crops is vital for assessing soil organic carbon (SOC) sequestration and greenhouse gas (GHG) dynamics across agroecosystems, yet reliable wall-to-wall information at that scale remains largely unavailable. Existing products are limited by coarse resolution, sparse ground truth, and inconsistent revisit across sensors, so the timing of establishment and termination — the part that matters most for carbon and nitrogen outcomes — is rarely resolved. This project leverages deep learning architectures to harmonize multi-sensor satellite observations (HLS,ECOSTRESS, NISAR), enabling high-resolution tracking of cover crop establishment, biomass accumulation, and termination dates across broad geographic domains. The resulting maps and time series are designed to feed directly into biogeochemical and policy-relevant assessments of agricultural conservation practice.