WASI-2D: A free software for inverse modeling of multi- and hyperspectral water images

Abstract

A software tool is presented which allows quantitative data analysis of optical remote sensing data from oceanic, coastal and inland waters. It can be easily adapted by the user to different sensors and to optical properties of the studied area. The software can import atmospherically corrected images from multi- and hyperspectral sensors in a number of common data formats and units like remote sensing reflectance or radiance, above or below the water surface. Data analysis is done by inverse modeling using established analytical models. The bio-optical model of the water column accounts for coloured dissolved organic matter (CDOM), detritus, and mixtures of up to 6 phytoplankton groups and 2 spectrally different types of suspended matter. The reflectance of the sea floor is modeled as sum of up to 6 substrate types. An analytic model of downwelling irradiance allows wavelength dependent modeling of sun glint and sky glint at the water surface. The provided database, which has been used successfully in inland lakes, covers the spectral range from 350 to 1000 nm in 1 nm intervals. It can be exchanged easily to represent the optical properties of water constituents, bottom types and the atmosphere of the studied area. The usage of the software is demonstrated at the example of a hyperspectral image from Lake Starnberg in southern Germany. The image has been acquired using a HySpex VNIR-1600 sensor with 160 bands from 416 to 992 nm. Goal of data analysis was to derive water depth of each image pixel in the shallow areas. The best results were achieved by treating water depth, sun glint, 2 substrate types, CDOM concentration and CDOM spectral slope as unknows (fit parameters) during inverse modeling, and keeping all other model parameters constant.