12:15–12:30 | ABSTRACT 1359
ADVANCED SHADOW CORRECTION FOR COASTAL HYPERSPECTRAL-LIDAR FUSION PRODUCTS: ADDRESSING RADIOMETRIC CHALLENGES IN BELOW-CLOUD OCEAN OPTICS
Low-altitude airborne hyperspectral imaging below cloud cover presents unique radiometric challenges for coastal ocean optics applications. Shadow contamination from clouds and atmospheric variability introduces spectral distortions that compromise water-leaving remote sensing reflectance (Rrs) retrievals, inherent optical property inversions, benthic albedo estimation, and bottom type classifications critical for shallow water characterization. These challenges are amplified in bathymetric lidar-centric operations where hyperspectral acquisition is secondary to lidar operations under variable atmospheric conditions. We present an automated dual-surface shadow correction framework for 48-band Itres CASI 1500 visible-NIR imagery collected at ~400m altitude during coastal surveys. The system employs ensemble detection combining K-means spectral clustering, PCA-based brightness analysis, spectral derivative variance metrics, and multi-index shadow detection (chromatic indices C1/C2, Normalized Shadow Index, Shadow Ratio Index, Modified Normalized Difference Index) with NDWI water masking, optimized separately for land and water surfaces. Water surface corrections preserve ocean optical properties by computing band-specific multiplicative correction factors from non-shadowed boundary reference spectra, applying Gaussian-weighted spatial transitions to maintain spectral shape fidelity while restoring radiometric intensity. Corrected imagery fused with lidar bathymetry supports navigation channel assessment, benthic habitat classification, sediment optical characterization for dredge material management, and coastal vulnerability analysis. Validation across coastal zone datasets demonstrate successful radiometric restoration with mean intensity enhancement of 1.5-2.2× in shadowed water areas while maintaining spectral shape integrity (r>0.85 correlation between normalized pre- and post-correction spectra), ensuring band-ratio-dependent algorithms for water quality and benthic classification remain viable in corrected imagery.
Joseph Harwood, USACE, [email protected]
Nicholas Johnson, USACE, [email protected]
