Seafloor Interpretation Through Applications of Remote Sensing - Morgan Erhardt (merhardt@coas.oregonstate.edu) - GEO565

1) Ph. Blondel and O. Gomez Sichi. 2008. Textural Analyses of multibeam sonar imagery from Stanton Banks, Northern Ireland Continental Shelf. Applied Acoustics 70:1288-1297

Annotation: This article was put together following a workshop held in Ireland to evaluate different methods of bathymetry processing and interpretation. While this workshop featured many methods, this paper focuses on the use of a proprietary software package known as TexAn, which was originally developed for side scan-analysis. Through this interpretation, the measurements that are collected initially are first order gray levels which look only at the intensity of the return in individual locations. Secondly, texture is interpreted through analysis of both grey levels and their proximity to other pixels with similar characteristic grey levels. These analyses are known as Grey-Level Co-occurrence Matrices and were referred to in Parnum et al, 2004 as a way to automate backscatter interpretation. In this study the authors cite that these second order interpretations of texture are indiscernible by visual interpretation but are of more importance than the first order grey level interpretations. The data used in this evaluation appears to be the same data use by Fonseca et al. (2009). As this paper is primarily focused on a proprietary software package it is not as interesting as open analysis.

2) B. M. Costa, T. A. Battista and S. J. Pittman. 2009. Comparative evaluation of airborne LiDAR and ship-based multibeam SoNAR bathymetry and intensity for mapping coral reef ecosystems. Remote Sensing of Environment 113:1082-1100

Annotation: This study took place in the vicinity of Puerto Rico. In this study the same bathymetric area was mapped using two different methods. In the end the two methods created very similar datasets for both bathymetry and backscatter intensity. The depth for these surveys was generally less than 50 meters and the LiDAR dataset was consistently shallower than the multibeam dataset. This depth discrepancy aside, the focus of this study was to evaluate these two methods in terms of their abilities to discern differing benthic habitat types. Additionally, these two methods were evaluated in terms of cost and time efficiency. At the end of this study, the authors determined that the LiDAR was a more cost effective, and less time consuming, method for mapping shallow waters of less than 50 meters depth. That said, they also identified that in these water depths multibeam bathymetry was significantly higher resolution than LiDAR but both methods were sufficient when general characterizations of benthic habitat was the goal. This conclusion was somewhat surprising to me as I was not aware that LiDAR was capable of producing comparable imagery to Multibeam Acoustics in water as deep as 50 meters.

3) Daniel C. Dunn and Patrick N. Halpin. 2009. Rugosity-based regional modeling of hard-bottom habitat. Marine Ecology Progress Series 377:1-11

Annotation: This paper argues that direct observation of seafloor habitat is costly and time consuming. They continue that an effort to characterize habitat types is warranted due to the increasing efforts toward marine conservation. In response to this need, the authors identify a proxy which can be used to better characterize seafloor habitat types without direct observation. They indicate that high correlations between hard-bottom substrate and biodiversity make the identification of such habitat essential to marine resource management and protection. This study used rugosity (the roughness of the physical structure of the seafloor) generated with a relatively low resolution bathymetric grid to build a predictive model for identification of hard-bottomed substrate. In the paper the authors used this model to provide a coarse resolution, remotely sensed, regional predictor of hard-bottomed habitat. Characteristics such as slope, depth, flow accumulation, bathymetric position index and rugosity were used in the resultant model. Following evaluation of several methods, the authors came to the conclusion that a stripped stepwise regression model was best for predicting hard bottom habitat when compared to empirical datasets.

4) Lucaino Fonseca, Craig Brown, Brian Calder, Larry Mayer and Yuri Rzhanov. 2009. Angular range analysis of acoustic themes from Stanton Banks Ireland: A link between visual interpretation and multibeam echosounder angular signatures. Applied Acoustics 70:1298-1304

Annotation: Fonseca and Mayer (2007) identified the process of angular range analysis as a method of characterizing the seafloor. One hiccup in this method was that the process worked with averages and values across an entire swath-width which allowed characterizations of sediment at higher resolution than some methods but not as high as would be preferred. In this paper, the authors used this same process but made a jump to better apply it on a more useful spatial scale. In so doing, the authors identified areas in the backscatter that had relatively uniform characteristics and applied the sediment characteristics derived from angular range analysis to those entire backscatter delineated regions. This allowed for a more realistic and accurate distribution of angular characteristics along the seafloor. The angular range analysis of the study area worked relatively well except where high variability within a swath handicapped the Angular Range Analysis' calculation process. These results were ground-truthed against camera photos from different sedimentary regions of the subject area and held up well.

5) Luciano Fonseca and Larry Mayer. 2007. Remote estimation of surficial seafloor properties through the application of Angular Range Analysis to multibeam sonar data. Marine Geophysical Research 28:119-126

Annotation: The bathymetry of the seafloor can be observed through multibeam sonar systems, though this only provides information on depths and not composition. Through analysis of the beam's intensity upon return to the sonar head we can get an idea of the hardness of the seafloor. Beyond this characterization, an additional analysis can be completed which takes into account the angle of incidence of the sound wave interacting with the seafloor sediment. This process is called Angular Range Analysis and is used to determine diagnostic characteristics of the sediment on the seafloor. The premise of this method is as follows: If one is able to establish a mathematical model, or relationship, that links the seafloor's acoustic physical properties to the observed acoustic return at the sonar transducer they could hypothetically invert this relationship to make inferences about seafloor composition on a larger scale. Like the information presented by Dunn and Halpin (2009), this data could be used to characterize seafloor sediment on a large spatial scale. Unlike Dunn and Halpin (2009), this method is of finer resolution and would be looking at changes in angular response of sediment rather than roughness of the seafloor on a more broad scale.

6) Gary C. Guenther, Mark W. Brooks, and Paul E. LaRocque. 2000. New Capabilities of the "SHOALS" Airborne Lidar Bathymeter. Remote Sensing of Environment 73(2):247-255

Annotation: In bathymetric survey conducted from a boat a person must find a way to deal with tidal flux if they want to present an accurate depiction of the water's depth relative to something other than the boat. In most cases this depth is eventually referenced to some vertical datum specific to the ocean such as Mean Low Water (MLW) or Mean Lower Low Water (MLLW). From this frame of reference one can make extrapolations to other vertical ellipsoidal or geoidal datums that are more globally referenced. As this paper points out, the use of a LiDAR system mounted to an aircraft allows for seamless transitions of bathymetric mapping up onto land. When the land elevations are known at some of the mapped locations, this allows a tidal independent reference point for bathymetry which can be immediately extrapolated to a terrestrial vertical datum eliminating several steps that would have otherwise been necessary. This is an interesting selling point for bathymetric LiDAR.

7) J. E. Hewitt, S. F. Thrush, P. Legendre, G. A. Funnell, J. Ellis, and M. Morrison. 2004. Mapping of Marine Soft-Sediment Communities: Integrated Sampling for Ecological Interpretation. Ecological Applications 14(4):1203-1216

Annotation: This study incorporated side-scan, single beam and video data to evaluate sediment types. This study seems to focus on ground-truthing; they use remote sensing data, such as the above mentioned side-scan and single beam technology, to gather large quantities of physical data, then use video to evaluate this data in accordance with associated ecology. Their goal, as with Fonseca and Mayer (2007) is to find characteristics of the remote sensing data that can be linked to ecology of one type or another. Using this technique the authors determined that under certain circumstances remote sensing data could be used to interpolate between areas of video-supported data. Some of the circumstances that helped to justify these interpolations were the planning of video surveys in such a way that they were trying to identify heterogeneity of benthos with the video tows. Additionally, both the multibeam and the side-scan data were found to be useful in these interpolations.

8) Vladimir E. Kostylev, Brian J. Todd, Gordon B. J. Fader, R. C. Courtney, Gordon D. M. Cameron, Richard A. Pickrill. 2001. Benthic habitat mapping on the Scotian Shelf based on multibeam bathymetry, surficial geology and sea floor photographs. Marine Ecology Progress Series 219:212-137

Annotation: This paper discusses the use of multibeam bathymetry, surficial geology, sea floor photographs and mega-faunal statistics to interpret habitat regimes. Bathymetry processing was accomplished using Caris HIPS and SIPS (which is what our lab uses ). Following bathymetry, backscatter and photographic interpretation, this study makes some cool maps of habitat. In reading through their process there are striking similarities between this work and the process occurring along the Oregon Coast in the Oregon Territorial Seas Mapping Project. Early in the paper the authors claim this to be a new technique which I believe dates the paper to some degree as this process is now widely practiced. In Figure 6 of this paper they have used GIS with varying sized symbols to indicate the distribution of species taxonomy throughout the region of Browns Bank. We could possibly use their habitat maps as templates for how to present data in the territorial seas mapping project.

9) T. Missiaen, S. Murphy, L. Loncke and J. -P. Henriet. 2002. Very high-resolution seismic mapping of shallow gas in the Belgian coastal zone. Continental Shelf Research 22:2291-2301

Annotation: While bathymetric mapping, in conjunction with backscatter, is often very useful for mapping potential resources along the ocean floor it lacks any resolution below the upper layer of sediment. Where bathymetric sonar falls off, seismic reflection survey picks up. In this paper the authors describe how seismic surveys in shallow water along the Belgian coast have possibly lead to the discovery of a large area of biogenic gas residing below the sea-floor. They theorize that this gas is the result of bacterial degradation on a layer of peat in the subsurface. Indications of this layer include acoustic turbidity, multiple reflections and phase reversals of the acoustic wave. Similar interpretation of seismic surveys could be, and I believe have been, completed along the northwest coast of the USA as a way to identify gas hydrates which are a potential energy source. As our lab has worked with some seismic data I was hoping to find reference of a seismic processing software package used for this work though nothing of substance was evident.

10) Daniel L. Orange, Janet Yun, Norman Maher, James Barry and Gary Greene. 2002. Tracking California seafloor seeps with bathymetry, backscatter and ROVs. Continental Shelf Research 22:2273-2290

Annotation: As our lab uses collects bathymetric data along the California margin this article was rather interesting. In this paper the authors use bathymetric data and backscatter to identify surficial expressions of marine seeps. They identify that anomalous backscatter could be indicative of authigenic carbonate of cold seep clams sustaining themselves on seep gasses. Following the identification of suspected seep sites through bathymetric interpretation, ROVs were sent down as a means of ground-truthing the remotely sensed data. These ROV dives discovered carpets of chemoautotrophic bacteria, seep communities and bubbling gas. As our lab has done work in the vicinity of Eel Canyon, this was an interesting prospect though the majority of our work has been in water depths greater than 3000m so the impacts of this study are only of cursory interest to our work.

11) Parnum, I.M., Siwabessy, P.J.W. and Gavrilov, A.N. 2004. Identification of seafloor habitats in coastal shelf waters using a multibeam echosounder. 2004. Proceedings of Acoustics. Nov. 3-5

Annotation: In this paper the authors reference a Reson SeaBat 8125 which was used to map areas of the Australian Coastal Shelf in 2003-2004. This project sounds similar to the Oregon Territorial Seas Mapping Project which was also focusing on habitat characterization. Like the Oregon Territorial Seas Mapping Project, this project used both bathymetry and backscatter to delineate habitat regimes. In this paper they discuss the different elements of backscatter interpretation for the characterization of seafloor cover. They identify issues such as angle varying gain and ultimately, come to the conclusion that they believe the primary characteristic of importance for habitat characterization from multibeam data is Backscatter Intensity. In the paper they mention an automated process that might be used to interpret backscatter in a more objective fashion. I have my doubts that it would be as accurate as human interpretation. The analysis they referenced is called a grey-level co-occurrence matrices analysis (GLCM). They don't go into great detail. The bathy and backscatter illustrated in this paper is less than ideal.

12) Juanita C. Sandidge and Ronald J. Hoyler. 1998. Coastal Bathymetry from Hyperspectral Observations of Water Radiance. Remote Sensing of Environment 65(3):341-352

Annotation: This paper presents a very cool idea. They identify several properties of a water column that combine to create the overall spectral radiance of the water. In this case, several of those characteristics are Water Depth, Bottom Reflectance, Inherent Optical Properties of the Water Column and Illumination Conditions. This spectral radiance can be observed and recorded by passive sensors rather than active sensors such as LiDAR or Acoustics. In this case, the authors use a hyperspectral sensor to better characterize the spectral radiance of the water. Through a process that utilizes a neural network this analysis attempts to remove signals generated by factors other than depth. When successful, this process should allow for a calculation of depth based on spectral analysis. In this study, the process employed by the authors successfully created a grid of bathymetric data that was within about 0.5 meters vertical accuracy when compared to known bathymetric data in the study area. I think this is a cool concept though it is likely only applicable in relatively shallow depths.

13) Thierry Schmitt, Niel C. Mitchell and A. Tony S. Ramsay. 2008. Characterizing uncertainties for quantifying bathymetry change between time-separated multibeam echo-sounder surveys. Continental Shelf Research 28:1166-1176

Annotation: Bathymetric data, when collected over a period of time, can be used to identify changes in seafloor character in terms of events such as submarine volcanic eruptions, advances of river deltas, submarine landslides or slope failures, or migration of sand layers along the seafloor. While all of these processes are important the process of mapping them over time is not without its issues. Inconsistency of bathymetric data quality can, in some cases, lead to the incorrect interpretation of temporal changes. In this paper the authors discuss several methods to quantify errors in bathymetric data that can be removed to better interpret observed temporal changes in bathymetry. The primary method presented here involves repeated surveys of a single immobile object and analysis of the multiple passes over that object. The analysis will allow the production of a histogram to show vertical offsets between the multiple passes. The behavior of this histogram allows for the isolation of variability associated with relatively-depth independent factors such as heave verses depth-dependent factors which have greater effect on outer beams which must travel through more water to reach the sea-floor and return.

14) Chi-Kuei Wang, and William D. Philpot. 2007. Using airborne bathymetric lidar to detect bottom type variation in shallow waters. Remote Sensing of Environment 106:123-135

Annotation: As with the Territorial Seas Mapping Project here in Oregon, there are difficulties associated with bathymetric mapping in nearshore waters. Water turbulence and breaking waves make the extreme nearshore a potentially dangerous place to work often resulting in gaps in data coverage in what we know as the "White Zone". This study evaluates shallow-water bathymetry collected by Airborne Bathymetric Lidar in a study location near Egmont Key, Florida. In the study they evaluate two potential sources of error associated with this method of data acquisition. The first was the effect of surface waves and the second was the effect of view orientation on the bottom return from a homogeneous bottom type. Surface waves are determined to be the most influential sources of error in determining accurate bathymetry through lidar. As discussed in this paper, this is the result of moving refraction as the aircraft flies over the study area. In multibeam bathymetric survey we experience similar issues when a sound velocity profile is not available for a region where acoustic survey is being performed. In multibeam we try to conduct CTD casts to account for stratification of the water column and associated refraction of sound waves. For LiDAR, this issue is amplified when looking through surface waves as the ocean surface changes on a much shorter timescale than a sound velocity profile.

15) Lisa M. Wedding, Alan M. Friedlander, Matthew McGranaghan, Russell S. Yost, Mark E. Monaco. 2008. Using bathymetric lidar to define nearshore benthic habitat complexity: Implications for management of reef fish assemblages in Hawaii. Remote Sensing of Environment 112:4159-4165

Annotation: In this study the authors use bathymetric LiDAR to evaluate shallow reef habitat in a bay on the south shore of Oahu in the Hawaiian island chain. This study shows that the use of bathymetric LiDAR can be an effective method of collecting rugosity data in shallow waters (4 meter grid size in this case). Truthing is conducted to compare in situ rugosity to that observed through LiDAR interpretation and a strong correlation was observed. The authors go further and use this rugosity as a proxy for fish biomass. Several different metrics were evaluated as measures of biomass and in the end a least-squares linear regression analysis was used to evaluate the relationship between LiDAR derived rugosity and fish assemblages. This analysis suggested that LiDAR derived rugosity was a statistically significant predictor of fish biomass with a R2 value of 0.64 and a P-value of less than 0.001. This indicates a surprising ability to predict fish biomass in tropical coral reef environments using bathymetric LiDAR. Kinda cool.