A case study at Starnberger See for hyperspectral bathymetry mapping using inverse modeling

Abstract

In coastal regions, hyperspectral remote sensing is becoming an established method to map water depth. For inland waters however, only few studies based on empirical methods have been published so far. This paper presents a study for the German lake Starnberger See using a physically based approach. Hyperspectral data were acquired from airplane using a HySpex VNIR-1600 sensor. They were processed by inverse modeling using a radiative-transfer based analytical model. In situ measurements were taken to decide which model parameters to fit and which to keep constant during data analysis, and they were used to initialize the parameters. Bottom reflectance was determined from the image itself. An echo sounding survey was undertaken for validation. For the studied area, bathymetry could be mapped up to a depth of 8 m with a rms error of 37 cm. Accuracies of 10-25 cm from 0-4 m and 35-65 cm from 4-8 m seem possible if the remaining systematic errors can be further reduced, e.g. by accounting for changes of bottom reflectance.