Benthic mapping of the Bluefields Bay fish sanctuary, Jamaica

Abstract

Small island states, such as those in the Caribbean, are dependent on the nearshore marine ecosystem complex and its resources; the goods and services provided by seagrass and coral reef for example, are particularly indispensable to the tourism and fishing industries. In recognition of their valuable contributions and in an effort to promote sustainable use of marine resources, some nearshore areas have been designated as fish sanctuaries, as well as marine parks and protected areas. In order to effectively manage these coastal zones, a spatial basis is vital to understanding the ecological dynamics and ultimately inform management practices. However, the current extent of habitats within designated sanctuaries across Jamaica are currently unknown and owing to this, the Government of Jamaica is desirous of mapping the benthic features in these areas.
Given the several habitat mapping methodologies that exist, it was deemed necessary to test the practicality of applying two remote sensing methods - optical and acoustic - at a pilot site in western Jamaica, the Bluefields Bay fish sanctuary. The optical remote sensing method involved a pixel-based supervised classification of two available multispectral images (WorldView-2 and GeoEye-1), whilst the acoustic method comprised a sonar survey using a BioSonics DT-X Portable Echosounder and subsequent indicator kriging interpolation in order to create continuous benthic surfaces.
Image classification resulted in the mapping of three benthic classes, namely submerged vegetation, bare substrate and coral reef, with an overall map accuracy of 89.9% for WorldView-2 and 86.8% for GeoEye-1 imagery. These accuracies surpassed those of the acoustic classification method, which attained 76.6% accuracy for vegetation presence, and 53.5% for bottom substrate (silt, sand and coral reef/ hard bottom). Both approaches confirmed that the Bluefields Bay is dominated by submerged aquatic vegetation, with contrastingly smaller areas of bare sediment and coral reef patches. Additionally, the sonar revealed that silty substrate exists along the shoreline, whilst sand is found further offshore.
Ultimately, the methods employed in this study were compared and although it was found that satellite image classification was perhaps the most cost-effective and well-suited for Jamaica given current available equipment and expertise, it is acknowledged that acoustic technology offers greater thematic detail required by a number of stakeholders and is capable of operating in turbid waters and cloud covered environments ill-suited for image classification. On the contrary, a major consideration for the acoustic classification process is the interpolation of processed data; this step gives rise to a number of potential limitations, such as those associated with the choice of interpolation algorithm, available software and expertise. The choice in mapping approach, as well as the survey design and processing steps is not an easy task; however the results of this study highlight the various benefits and shortcomings of implementing optical and acoustic classification approaches in Jamaica.

Popular Abstract

Persons automatically associate tropical waters with spectacular views of coral reefs and colourful fish; however many are perhaps not aware that these coral reefs, as well as other living organisms inhabiting the seabed are in fact extremely valuable to our existence. Healthy coral reefs and seagrass assist in maintaining the sand on our beaches and fish populations and are thereby crucial to the tourism and fishing industries in the Caribbean. For this reason, a number of areas are protected by law and have been designated fish sanctuaries or marine protected areas. In order to understand the functioning of theses areas and effectively inform management strategy, the configuration of what exists on the seafloor is crucial. In the same vein that a motorist needs a road map to navigate unknown areas, coastal stakeholders require maps of the seafloor in order to understand what is happening beneath the water’s surface.
The location of seafloor habitats within fish sanctuaries in Jamaica are currently unknown and the Government is interested in mapping them. However a myriad of methods exist that could be employed to achieve this goal. Remote sensing is a broad grouping of methods that involve collecting information about an object without being in direct physical contact with it. Many researchers have successfully mapped marine areas using these techniques and it was believed crucial to test the practicality of two such methods, specifically optical and acoustic remote sensing. The main question to be answered from this study was therefore:
Which mapping approach is better for benthic habitat mapping in Jamaica and possibly the wider Caribbean?
Optical remote sensing relates to the interaction of energy with the Earth’s surface. A digital photograph is taken from a satellite and subsequently interpreted. Acoustic/ sonar technology involves the recording of waveforms reflected from the seabed. Both methods were employed at a pilot site, the Bluefields Bay fish sanctuary, situated in western Jamaica. The optical remote sensing method involved the classification of two satellite images (named WorldView-2 and GeoEye-1) and this process was informed using known positions of seafloor features, this being known as supervised image classification. With regard to the acoustic method, a field survey utilising sonar equipment (BioSonics DT-X Portable Echosounder) was undertaken in order to collect the necessary sonar data. The processed field data was modelled in order to convert lines of field point data to one continuous map of the sanctuary, a process known as interpolation. The accuracy of each method was then tested using field knowledge of what exists in the sanctuary.
The map resulting from the image classification revealed three seafloor types, namely submerged vegetation, coral reef and bare seafloor. The overall map accuracy was 89.9% for the WorldView-2 image and 86.8% for GeoEye-1 imagery. These accuracies surpassed those attained from the acoustic classification method (76.6% for vegetation presence and 53.5% for bottom type - silt, sand and coral reef/ hard bottom). Similar to previous studies undertaken, it was shown that the seabed of Bluefields Bay is primarily inhabited by submerged aquatic vegetation (including seagrass and algae), with contrastingly smaller areas of bare sediment and coral reef.
Ultimately, the methods employed in this study were compared and the pros and cons of each were weighed in order to deem one method more suitable in Jamaica. Often, the presence of cloud and suspended matter in the water block the view of the seafloor making image classification difficult. On the contrary, acoustic surveys are capable of operating throughout cloudy conditions and attaining more detailed information of the ocean floor, otherwise not possible with optical remote sensing. A major step in the acoustic classification process however, was the interpolation of processed data, which may introduce additional limitations if careful consideration is not given to the intricacies of the process. Lastly, the acoustic survey certainly required greater financial resources than satellite image classification.
In answer to the main question of this study, the most cost effective and feasible mapping method for Jamaica is satellite image classification (based on the results attained). It must be stressed however that the effective implementation of any method will depend on a number of factors, such as available software, equipment, expertise and user needs, that must be weighed in order to select the most feasible mapping method for a particular site.

• author: Mcintyre, Karen
• supervisor: Ulrik Mårtensson
• organization: Dept of Physical Geography and Ecosystem Science
• alternative title: Mapping the seafloor at Bluefields Bay, Jamaica
• course: GISM01 20142
• year: 2015
• type: H2 - Master's Degree (Two Years)
• subject: Earth and Environmental Sciences
• keywords: remote sensing, image classification, GIS, acoustic survey, benthic habitat mapping
• publication/series: LUMA-GIS Thesis
• report number: 40
• language: English
• additional info: External Supervisor: Kurt McLaren, Lecturer, Forest Ecology, Department of Life Sciences, University of the West Indies (Mona Campus); Imagery grant received from the DigitalGlobe Foundation for WorldView-2 and GeoEye-1 satellite imagery, and tide data donated by the United Kingdom Hydrographic Office (UKHO) Tidal Prediction Service (TPS).
• id: 7356935
• date added to LUP: 2015-06-23 14:35:51
• date last changed: 2015-08-06 11:32:34