Bathymetric retrieval from manifold coordinate representations of hyperspectral imagery

Abstract

In this paper we examine the accuracy of manifold coordinate representations as a reduced representation of a hyperspectral look-up table for bathymetry retrieval. The approach is based on the extraction of a reduced dimensionality representation in manifold coordinates of a sufficiently large representative set of hyperspectral imagery [4]. The manifold coordinates are derived from a scalable version [4] of the isometric mapping (ISOMAP) algorithm [12] [11]. In the present work and in [5], these coordinates are used to establish an interpolating look-up table for bathymet- ric retrieval by associating the representative data with ground truth data, in this case from a LIDAR estimate in the representative area. The compression of look-up tables could also readily be applied to look-up tables generated by forward radiative transfer models [9]. In this paper, we analyze the approach using data acquired by the PHILLS [6] hyperspectral camera over the Indian River lagoon Florida in 2004. Within a few months of the PHILLS overflights, SHOALS LIDAR data was obtained for a portion of this lagoon, principally covering the beach zone and in some in stances portions of contiguous river channels.