Towards Estimation of Uncertainty in the Measurement of the Seafloor Acoustic Backscatter

Mashkoor Malik
Ph.D. Candidate
Natural Resources & Earth System Science University of New Hampshire
Friday, Feb. 4, 2011, 3:00pm

Seafloor backscatter data derived from Multibeam Sonar Echo Sounders (MBES) have gained widespread use in a host of acoustic remote sensing applications. The evaluation of measurement uncertainty in the MBES-derived backscatter, however, has received relatively little attention. During this study an empirical approach was developed, based on the sonar equation, to identify and differentiate between expected sources of uncertainty: device-, environment- and the seafloor-related uncertainty. To assess the device-related uncertainty, an EM 3002 (Kongsberg, Inc 300 kHz) MBES was installed in the UNH acoustic test tank. Observed tank wall backscatter showed a random uncertainty of <0.5 dB. To assess the systematic uncertainty of the device, a complete assessment of transmit and receive characteristics of the MBES is required which is challenging considering limited access to proprietary information about the device configuration and inability to access raw elemental (non-beam formed) data. Despite in-tank experiments were successful to evaluate the transmit characteristics including transmit source level and transmit beam pattern. A systematic bias of ~0.61 dB was found between the source level measured in tank and the value provided by the manufacturer.

For the study of environment- and seafloor-related uncertainties, the EM 3002 was rigidly mounted to the side of the UNH marine pier from May 21–June 2, 2008. A 35.6 cm diameter stainless steel sphere, filled with distilled water, was placed above the seafloor to isolate the effect of environment-related changes on the backscatter measurements during 28-31 May, 2008. The pooled standard deviations from all the one minute ensembles resulted in a mean standard deviation of 0.6 dB which is considered the combined effect of random device- and environment-related uncertainty. A maximum variation of ~3dB was observed in the sphere backscatter that is strongly correlated to the observed temperature variations. This relationship is thought to be caused by changes in the sphere backscatter characteristics under the influence of temperature. The backscatter from the seafloor (21 – 26 May, 2008) was found to be highly variable (up to 20 dB) at time scales of few minutes to days.  This variability is thought to be the result of changes in seafloor backscatter due to biological activity and/or movement of sediments / shells under currents. Over shorter time scales (< 1 minute), the standard deviation obtained from the seafloor backscatter data was found to be on order of 0.5 - 3 dB. This value is considered to be the inherent variability of the seafloor backscatter measurement resulting from the combined effects of the device as well as the environmental variability and the instantaneous variability of the seafloor. 

This study is considered to be a first step towards establishing the uncertainty of MBES-derived backscatter measurements. We have used controlled experiments and limited but realistic, environmental variability for a specific MBES to isolate and quantify device-, environment- and seafloor- related random uncertainty. The results indicate that EM 3002 is a stable backscatter measurement device capable of providing results within 0.5 dB. However, biases due to a specific sonar head, time scale of variations in the seafloor properties, specific data processing techniques, heterogeneity of the seafloor, movement of the vessel and environmental variability will have to be considered to assess the overall uncertainty.


Mashkoor Malik is a UNH / CCOM alum (MS Ocean Mapping 2005). Since August 2008 he has been working as Physical Scientist for NOAA Office of Ocean Exploration and Research. He is also currently a part time PhD student. His research interests are application of mapping techniques towards ocean exploration specifically use and interpretation of seafloor backscatter data.