|INTRODUCTION - Importance of marine habitat studies|
The study of marine habitats is a key component for understanding and managing ocean resources, in addition to its intrinsic scientific interest. Over the last years there has been a strong interest in developing techniques and technology for habitat mapping and monitoring of temporal changes. One of the reasons for such interest is the need that coastal managers have for detailed and accurate habitat information in the decision-making process. Habitat information can be used for assessing sites that are important from a conservation point of view, determining areas sensitive to disturbance and pollution, development of coastal and marine protected areas management plans, and providing valuable indicators for environment impact assessment and monitoring environmental change. A second reason stems from recent technical advances that provide scientists with the tools that they needed to acquire, process, and merge vast amounts of ocean data with the objective of producing quantitative and qualitative information on marine habitats.
The rapid decline of some habitats (for example coral reef communities worldwide), is motivating the design of innovative assessment tools to rapidly and effectively document the distribution and condition of endangered benthic organisms (Solan et al. 2003, Fisher et al. 2005). With the development of better and more affordable photography and videography equipment, the use of digital imagery in benthic monitoring has increased dramatically in the last decade and video surveys are now routinely conducted as complements to diver-based measurements (Carleton and Done 1995, Ninio et al. 2003). Moreover, several large-scale monitoring programs are now based almost exclusively on the analysis of video imagery. One such example is the Coral Reef Monitoring Program of the Florida Reef Tract where permanent belt transects are surveyed annually and video frames are sub-sampled to obtained estimates of coral cover and condition (Porter et al. 2002).
These imaging tools improve survey efficiency by: (1) reducing the time that divers need to spend underwater by shifting data analysis away from the field and into the lab; and (2) providing a permanent visual record of reef condition.
While diver-based video is helpful in estimating a number of of important ecological indices, it provides only limited size-estimation capabilities. Sizes can only be measured for structures that fall completely within an image. This is a strong limitation, for example in reef habitats with large corals and high topographical relief, since individual colonies will rarely be totally captured within a single frame or a video transects. These surveys are also limited as georeferencing of the acquired images is poor, and repeatability of surveys difficult for large areas.
The development of sensors and platforms for the study of coastal and deep sea habitats is on the rise. Such technological progress goes in hand with the notion that advances in marine habitat mapping require multidisciplinary approaches which are only possible by combining efforts from different fields of research. Some research institutions have undertaken important steps for the study of the spatial pattern of benthic habitats following a "decrease scale, increase resolution" strategy (e.g., Instituto Superior de Robotica in Lisbon, IFREMER, University of Miami, CNRS/IPGP, etc.). In this context they have pushed programs whereby the ocean floor is first surveyed on a large scale using side-scan sonar, increasing the resolution and improving quantitative refinement as observations are made at closer ranges by resorting to multibeam systems, remote and diver-held video, diver written records, and sampling of the seabed. The problem of overlaying the different datasets and their scales (for example, multibeam data and grab samples), so as to generate composites of the benthic ecosystem and is then resolved by resorting to commercially available Geographic Information Systems (GIS).
Within the context of coral reef mapping, mosaicing techniques has been used to map areas of few hundreds of square meters, while having enough ground resolution to allow the identification of colonies of a few centimetres size (Lirman et al. 2006). Over areas of low topographic relief, such techniques allow creating mosaic from image information alone. However, when attempting to cover areas of more than 500 sq meters or with high relief, the use of additional positioning information becomes crucial to properly place the images in a geographic framework. Such information can be provided by acoustic positioning systems, inclinometers and rate gyros. Recent technologic developments have significantly reduced the price and size of these sensors thus allowing for their use even in small frame underwater robots (Singh et al. 2004).
From the benthic mapping point-of-view, the analysis of mosaics constructed over two or three spatial dimensions presents several advantages over strip mosaics constructed along a single spatial dimension (video transects). Not only the sizes and spatial arrangements of benthic structures can be more accurately estimated, but more importantly, can also serve as the basis for change-detection from repeated surveys. This contrasts with surveys from linear transects which are part of the current state-of-the-art for monitoring, and are exceedingly difficult to duplicate precisely over time.
In addition to imagery, relief (microbathymetry) is a key parameter to understand and properly interpret the observed structures, and to ultimately properly merge and combine these two datasets. Furthermore, many geological and biological processes involve either movement or growth of structures, and subsequent combined 3D structure and imagery are key for their understanding.
Mosaics can provide landscape-level views for benthic data collected by remotely operated or autonomous platforms. Image registration techniques, complemented with navigational data from positioning sensors have the potential to allow the accurate and repetitive mapping of study areas. Furthermore, a platform equipped with these sensing capabilities will require minimal end-user intervention during mission execution, thus potentially benefiting a large group of marine scientist.
Image based mapping techniques can provide unique opportunities to study the spatial arrangement, condition, and sizes of benthic structures at locations not easily accessible to scientific divers, thus providing a crucial set of tools for the study of deeper benthic communities.
Accessibility constraints are another important limiting factor in habitat characterization. Detailed studies have been restricted to shallow waters areas, within the reach of trained divers who traditionally perform in situ observation and analysis. ROVs are typically operated at depths greater than 100-200 m. In addition to the danger that human diving represents and its limitations, shallow water AUVs can bridge the gap between these depth ranges, allowing access to an important part of the marine ecosystem that it is not studied in a systematic manner at the present time.