View Abstract


The past two decades has seen the development of several shallow water inversion models that can simultaneously retrieve the water column's inherent optical properties (IOPs), geometric depth and bottom reflectance from above-water remote sensing reflectance. However, the accuracy of these retrievals is partly dependent on the spectral shape of the benthic substrate(s) used as input to the inversion scheme. Paradoxically, knowledge of this spectral shape is unknown prior to the inversion and is often a crucial product for the optical classification of benthic substrates. Current approaches bridge this gap by iterating through unique combinations of benthic endmembers to optimize for the IOPs, depth and benthic spectral shape. This is a computationally expensive procedure requiring multiple inversions per-pixel to find the benthic endmember combination that yields the lowest error of fit between the modeled and measured remote sensing reflectance. To improve the efficiency of shallow water inversion models and to break this paradox, a look up table (LUT) approach was developed to predetermine the likely benthic endmembers directly from remote sensing reflectance. These benthic endmembers are subsequently implemented in a physics-based inversion model to retrieve IOPs and depth and to further refine the spectral shape of the derived bottom reflectance. We compare this coupled LUT-inversion based approach against the standard physics inversion algorithms using both in-situ radiometry and imagery from NASA's Portable Remote Sensing Imaging SpectroMeter (PRISM) of Lizard Island and Heron Island, Great Barrier Reef, Australia, captured during the recent CORAL airborne campaign


Garcia, R. A., University of Massachusetts, Boston, USA,

Lee, Z. P., University of Massachusetts, Boston, USA,


Oral presentation

Session #:096
Date: 02/27/2017
Time: 12:00
Location: 302 A/B

Presentation is given by student: No