The Bovine Slurry Index (BSI): An integrated framework for manure pond mapping using GeoFM-derived cattle-operation detection and Sentinel-2 spectral formulation

Quantifying greenhouse gas (GHG) emissions from livestock operations is a prerequisite for reconciling bottom-up inventories with atmospheric observations. However, global GHG reporting remains limited by a lack of spatially explicit data on animal waste management systems (AWMS), forcing reliance on coarse regional assumptions rather than operation-level observations. Existing geospatial foundation models partly address this spatial gap by mapping cattle-operation footprints that provide upstream spatial context, but contextual detection alone cannot identify the manure-management features within them. To address this gap, we introduce the Bovine Slurry Index (BSI), an empirical Sentinel-2 spectral index for detecting liquid/slurry manure ponds within known cattle-operation footprints. The primary formulation, BSI-E, uses calibrated narrow-NIR (B8A) and SWIR-1 (B11) reflectance bounds derived from manure pond samples in the United States and China. The index targets the empirically observed intermediate NIR-SWIR response of manure ponds relative to clear water, vegetation, dry holding pens, and exposed soil. Calibration used 2,730 pond samples from the U.S. and China, and segmentation was evaluated against 945 reference polygons across all five geographic regions. BSI-E produced stronger pond-to-background contrast and higher segmentation accuracy than evaluated water indices, with dataset-level intersection-over-union (IoU) ranging from 0.26 in Brazil to 0.68 in the United States and precision reaching 0.91. Optimal thresholds varied geographically from 0.25 to 1.00, and performance was strongest in dryland feedlot systems and lower in humid or spectrally mixed settings. BSI provides a target-specific spectral method for generating operation-scale evidence that could inform AWMS activity data within these footprints.