ONR Capturing Uncertainty DRI

Seabed Variability and its Influence on Acoustic Prediction Uncertainty

Principal Investigators


Area of Expertise


Brain Calder

Bathymetric Uncertainty

U. New Hampshire

John Goff

Statistical characterization of surface/sub-surface properties/morphology


Chris Harrison

Multi-static modeling

NATO Undersea Research Centre, harrison@nurc.nato.int

Charles Holland;
team leader

Acoustic measurements/geoacoustic inversion

ARL/Penn State

Barb Kraft

In-situ velocity measurements

U. New Hampshire

Kevin LePage

Reverberation modeling


Larry Mayer

Geoacoustic/bathymetry measurements

U. New Hampshire

Bob Odom

Acoustic propagation modeling (forward/inverse)


Irina Overeem

Predictive geophysical modeling


Lincoln Pratson

Predictive geoacoustic modeling; lab-generated 3D strata

Duke Un.; lincoln.pratson@duke.edu

James Syvitski

Predictive geophysical modeling



1. Introduction

The weakest link in performance prediction for naval systems operating in coastal regions is the environmental data that drive the models.  In shallow water downward refracting environments, the seabed properties and morphology often are the controlling environmental factors. There are two important acoustic parameters for predicting acoustic interaction with the seabed: bottom reflection (sometimes called bottom loss) and bottom scattering strength. Both parameters are controlled by sediment properties down to depths of order several tens of meters, depending on the operating characteristics of the sonar, and the properties of the sediment.


At present, the seabed environmental data comes from static databases that provide the bottom reflection coefficient itself or geoacoustic properties.  These databases treat spatial variability of order ~10 m in the vertical and 10 km in the horizontal.  The error associated with the databases comes from variety of sources, but a crucial limitation of the current databases arises from the fact that they grossly under-sample the important scales of variability of the seabed.  Due to the under-sampling of the seabed properties, the acoustic predictions have significant, but at present unknown, uncertainty.

Long Term Goals

In order to address the problem of unknown uncertainty, we propose an effort focused on two overarching goals:

1.      Assess and characterize the seafloor variability

2.      Determine the impact of the seafloor variability on acoustic prediction uncertainty


2. Overview of Technical Approach

In order to address the principal issues mentioned above, we propose an approach that blends the key disciplines of geology/geophysics (G&G) and acoustics, linked by the field of geoacoustics. The field of geoacoustics was born out of the recognition that in order to predict the effect of the seafloor on acoustic systems, a physics-based approach was required, in which the physical properties of the seabed were identified and measured.  The “art” of geoacoustics is to identify the key properties that control acoustic interaction and parameterize them. Thus, a geoacoustic model is not an exact description of the seafloor, but an approximation that includes the important physics of the problem being addressed.  The current generation of seafloor databases is believed to over-simplify the important physics hence geoacoustics in some shallow water environments.

For many years, the underwater acoustics community has probed the sediment geoacoustics through inverse modeling of acoustic measurements (e.g., propagation, reverberation) or via empirical relations (e.g., Hamilton) or models (e.g., Biot-Stoll). From the other end, the G&G community has developed its strategies to obtain sediment geoacoustics such as advanced coring and in-situ sampling devices as well as models (e.g., sediment deposition, transport).


We propose to bring together these two disciplines, G&G and acoustics, at the intersection of geoacoustics in order to bring to bear the very best tools of both disciplines and the concomitant synergy on the thorny problem of Uncertainty.  In fact, we believe that a combination of G&G and acoustics, is not the only the best but the only way to significantly advance the understanding of seafloor variability and its effect on acoustic predictions.  

We have deliberately stopped far short of an end-to-end approach to the Uncertainty problem in order to focus on this crucial aspect of the entire problem.  We see our team transitioning and providing seafloor variability products to other teams addressing oceanographic variability, signal processing, and sensing/information dominance issues.


In addition to the fertile union of the G&G and acoustics, we also bring several new advances together.  For example, we offer very high resolution seabed reflection and scattering data sets/measurement techniques that are uniquely suited to expanding the understanding of seafloor variability. In addition we bring the state-of-the-art G&G sediment deposition modeling and tools to map its predictions into geoacoustic quantities.  We also bring powerful ideas using Frechet derivatives for determining acoustic uncertainty due to seafloor variability. Finally, we bring the very high fidelity reverberation modeling that includes many aspects of the scattering process ignored by other models which will allow us to estimate errors inherent in the modeling of active system performance.  The details of each of these areas are described in the following sections addressing each PIs approach.

Figure 1 shows our teamed approach. The research to the left largely focuses on goal #1, i.e., determining and characterizing the seafloor variability.  From the G&G side, Syvitski and Pratson will model the variability of deterministic geoacoustic properties, and Goff will develop stochastic geoacoustic descriptions based on field data.  From the acoustics side, Holland will develop deterministic and stochastic properties from inversion of reflection and scattering measurements.  An important aspect of the research will be to compare the geoacoustic results from the G&G and acoustics approaches to determine errors/weaknesses that are inherent in the two approaches.  Moreover, we will explore new techniques for reducing uncertainty by a combination of G&G and acoustic methods.  The research in the bottom right corner largely focuses on goal #2, i.e., acoustic prediction uncertainty resulting from the seafloor variability with Odom addressing propagation uncertainty and LePage the reverberation uncertainty.

The accomplishment of our primary goals will provide other teams within the DRI the information required to carry the effects seafloor variability all the way through to an estimation of uncertainty in Fleet Prediction Products (e.g., tactical decision aids).  In addition, we anticipate that our products will also provide the scientific foundation for reducing the impact of the seafloor variability on acoustic prediction uncertainty.  For example, one of the products coming out of the seafloor variability data analysis and modeling will be the identification of weaknesses in the current Navy seabed databases with specific recommendations concerning their improvement.



Figure 1. Simple diagram showing major task areas and responsible PIs.  See text for discussion. The key elements of our total approach include measurements, models, and data, which are represented by the black squares, trapezoids, and ellipses respectively.