Global Correction for Field Reconstruction From Incomplete Data

This method addresses field reconstruction from fragmented observations by using a locally constructed, globally revisable estimator. A coordinate-conditioned predictor learns from incomplete patches, allowing well-observed neighborhoods to provide supervision to correct local estimates across the full domain.