Seafloor sediment characterization using multibeam echosounders without grab sampling

Abstract

In the last two decades, the use of multibeam echosounders has been growing for seafloor mapping and characterization. The former uses bathymetry data whereas the latter makes use of backscatter data. The use of backscatter data has been the subject of intensive research to gain insight into seafloor composition using either empirical or model-based methods. Model-based methods employ the available physical models for predicting the backscatter strength and determine the seafloor geoacoustic parameters in an inversion algorithm using optimization methods. These methods allow for direct coupling between the backscatter curve and sediment characteristics. But the methods usually suffer from a shortcoming associated to uncalibrated sonars, which is referred to as calibration curve. Grab samples at reference areas are required to estimate the calibration curve. A question may arise as to whether, or to what extent, the calibration curve can be estimated without grab sampling. Knowing that the calibration curve is an unknown function of incident angle, in principle, one can approximate it using the available estimation and optimization theories. This is elaborated in this paper and its opportunities and challenges will be addressed. The potential benefit is twofold. 1) The huge amount of MBES backscatter currently available in many hydrographic organizations can directly be used for seafloor characterization. 2) The available multiple-frequency MBESs can further improve the performance of the inversion process. There are also challenges to be addressed. 1) Estimation of the calibration curve is an unstable process because it is merely based on observed backscatter data without using grab samples. 2) The physical models, and component parts thereof, are not usually wellbehaved functions, possibly due to their discontinuities or discontinuity of their derivatives. These issues will be elaborated in this paper.