Caribbean Island Maps On- and Off-shore Habitats

Balancing Growth with Preservation

Figure 1

St. Lucia’s island-wide GIS now allows for integrated coastal resources management and smart planning to avoid urban sprawl and protect the island’s pristine coastline.

Figure 2

Image taken by the Fugro and Envision field verification crew in St. Lucia. Picture courtesy of Envision.

Figure 3

Segmentation of the imagery using Definiens software resulting in polygons.

Figure 4

The polygons are labeled using Classification and Regression Tree (CART) analysis.

Figure 5

The final labeled polygons.

Figure 6

Orthophotos off the coast of St. Lucia at 25-cm pixel resolution.

Kevin P. Corbley
Geospatial Business
ConsultantCorbley Communications
Denver, Colo.

On the island of Saint Lucia, where tourism drives the economy, managing new development and protecting natural resources require a delicate balancing act. As it seeks economic growth, the tiny Caribbean nation is determined to avoid the mistakes of other island destinations by planning construction of future hotels and resorts to accommodate additional visitors without damaging its number one tourist attractions – the pristine coastline and lush mountain habitats.

“We realized that development along the coast impacts more than just the beaches,” said Suzanna Aurelien, Senior Cartographer with the Saint Lucia Ministry of Physical Development and Environment. “So we needed a national GIS to help us in planning, development and environmental management.”

Editor’s Note: An article about CRS in coral reef mapping also appears here.

In 2007, the Ministry envisioned a GIS capable of addressing issues unique to island nations. Rather than stopping at the water’s edge, coverage of the system would extend from the highest interior mountain peak, across the white sandy beaches, and out into the ocean to a depth of 20 meters. Just three years later, Saint Lucia is on its way to developing a GIS that will enable it simultaneously to plan resort construction and agricultural programs while monitoring the health of its beaches, seagrasses and coral reefs.

Mapping the Coastal Zone

Saint Lucia is located among the Lesser Antilles Islands of the eastern Caribbean. Beach activities such as sunbathing, sport fishing and snorkeling are the primary attractions on this island, explained Peter Felix, Chief Surveyor for the Ministry of Physical Development. But the thousands of annual tourists are also drawn to the natural beauty of the island’s interior rain forests, hot springs and ‘drive-through’ volcano.

A major challenge at the start of GIS development was the lack of existing geospatial data relating to these popular areas, or to any other parts of the island that 200,000 people call home. The only maps were in paper form and years old. The first order of business was to map the island from scratch. Because different funding sources were involved, the decision was made to perform the GIS base mapping as separate coastal and land-based projects.

With Saint Lucia’s banana-based agriculture industry attempting to diversify into other farming and business activities, its Banana Industry Trust (BIT) spearheaded the coastal mapping project. BIT selected Fugro GEOID of France to assemble the project team, which ultimately included Fugro EarthData in the U.S., and WS Atkins International and Envision, both of the U.K. Fugro GEOID handled ground control and geodatabase development, while the U.K. firms provided technical assistance in several capacities.

Fugro EarthData was assigned the role of acquiring and processing high-resolution digital imagery to create the coastal and benthic habitat basemap. BIT requested aerial image collection at 75-centimeter resolution and classification of multiple habitat types covering both the on- and off-shore environments, including mangroves, marshes, sand beach, rocky shoreline, seagrasses and coral reefs (see Figure 1). Following aerial acquisition, Envision and Fugro personnel filmed the benthic habitat with a GPS-enabled underwater video to identify the key classes (see Figure 2).

“We had mapped benthic habitats before, but never to a depth of 20 meters,” said Debbie Simerlink, Fugro EarthData Project Manager. “In fact, the aerial acquisition proved just as challenging as the processing of the image data itself.”

Fugro EarthData deployed a twin-engine aircraft equipped with a Leica ADS40-SH52 digital camera to the island in fall 2008. Due to the irregular shape of the coastline, standard linear flight lines were drawn to cover almost the entire 620 square kilometers. During processing, the 170-square-mile coastal zone would be separated out from the other data. Although only three days of total flight time were needed, the actual deployment took much longer.

“In the tropical locale, the flight crew had to find openings in the cloud cover that coincided with relatively calm ocean conditions with as little turbidity as possible,” said Simerlink. “On any given day, they could only capture small segments of flight lines, if they flew at all.”

The four-band image sets, comprised of natural-color and near-infrared data, were periodically delivered to Fugro EarthData’s headquarters in Frederick, Maryland, for quality checking. If the subsurface images were not clear enough due to sediment in the water or rough surf, the lines were re-flown. Once the entire coastal area had been flown, the imagery was orthorectified in preparation for the coastal and benthic habitat mapping.

Applying Object-Oriented Classification

Mapping subsurface habitats with airborne multispectral imagery is a relatively new technique that Fugro EarthData played a major role in developing, in partnership with the National Oceanic and Atmospheric Administ-ration (NOAA). The semi-automated methodology, referred to as an object-oriented image classification process, was originally developed by Fugro EarthData in 2008 through a series of projects spearheaded by NOAA’s Coastal Services Center to map seagrass beds along the Gulf Coast of Texas. Although the company had applied the process in other commercial projects since then, Saint Lucia again presented a special challenge due to the depth of mapping requested by BIT.

“The collection of field data with the underwater video proved critical in the analysis,” said Chad Lopez, Senior Digital Imaging Analyst at Fugro EarthData. “The video allowed us to see that the appearance of identical habitats changed from shallow water to deep, and that meant their spectral signatures in the imagery would change as well.”

In the first phase of analysis, Lopez masked the orthorectified coastal image at the water line into on- and off-shore data-sets. He then used Definiens software to segment the images into polygons (see Figure 3). Other than allowing Lopez to set parameters relating to the size and shape of the output polygons, the software ran in an automated fashion without inputs of classification training data.

It segmented the image data into discrete polygons comprised of multiple pixels, which were later classified, rather than classifying individual pixels by spect-ral value. The three visible bands and one infrared band were processed simultan-eously. “The output was a vector dataset in which the polygons outlined the various habitats,” said Lopez. “Training and classification came in the second phase.”

Next, the analysts used commercially available statistical analysis software to classify the polygons in a technique referred to as Classification and Regression Tree (CART) analysis. CART utilized ground truth, or habitat training information, collected by the field teams onsite and extracted from the digital underwater video to create a set of rules for polygon classification. The analysis routine took into account a multitude of variables in establishing these rules, which ultimately generated an orthoimage layer with classes labeled according to the desired habitat schema.

“An advantage of the CART analysis is that you can quickly run several iterations to improve the overall classification accuracy,” said Lopez. “From there, the CART-generated classes are joined to the image segments using ESRI’s ArcGIS to produce classified polygons.” (See Figures 4-5.)

Fugro GEOID has developed a multi-layer geodatabase structure containing the coastal habitat classes for delivery to BIT. The database is designed for compatibility with the national GIS under development and will accommodate updated habitat information as the system evolves.

The decision to create the GIS came at an important time for Saint Lucia. Early examination of the imagery indicates there has already been damage inflicted upon the reefs and significant beach erosion near existing resorts.

Capturing the Land Base

The Ministry of Physical Development and Environment awarded a second contract directly to Fugro EarthData to create a high-resolution basemap of the entire island. This would serve as the primary digital land base for the nascent GIS. At the time of the contract award, the Maryland firm had already deployed its ADS40-equipped aircraft to Saint Lucia for the coastal mapping, and the same crew was able to alternate acquisitions for the two projects in late 2008 into early 2009.

Mapping subsurface habitats with airborne multispectral imagery is a relatively new technique that Fugro EarthData played a major role in developing, in partnership with the National Oceanic and Atmospheric Administration (NOAA).

“We had to fly the two projects separately because the Ministry requested much higher resolution for the land-based mapping, which required a lower altitude,” said Simerlink.

An interesting aspect of the land-based acquisition was the additional request from the government to collect the terrestrial imagery at two different spatial resolutions. The Ministry wanted the majority of the island mapped at 25-centimeter resolution (see Figure 6 ), while eleven populous urban areas would be collected at 12.5 centimeters. By taking advantage of the ADS40’s extra detector array, which can be used to increase resolution without flying separate lines at different altitudes, Fugro was able to meet this requirement without significantly increasing the flying time and overall cost.

Through this technique, known as HiRES mode, the firm was able to produce orthoimages at the two resolution levels, generate digital elevation models and extract framework data layers (buildings, roads, etc.) at the required scale and accuracy.

The Saint Lucia land-based map was delivered to the Ministry in the form of an ArcGIS geodatabase file. It contained the orthoimagery, DEMs, contours and planimetrics that will serve as the foundation for the nationwide GIS.

Managing the Island Ecosystem

Fugro EarthData’s Lopez believes the ability of digital airborne cameras to capture clear subsurface imagery and the development of processing techniques capable of generating habitat maps from those images have dramatically enhanced the value of GIS-based coastal zone management because they add details relating to an important part of the ecosystem.

“Benthic habitat mapping is a fundamental component of coastal management,” said Lopez.

Without these new technologies, mapping offshore would have been more costly because the data would have to be collected manually, a process that could take years even on an island like Saint Lucia. Once its GIS is fully developed, the island’s Ministry will have a complete picture of how activities on the land impact the sea and vice versa. The government hopes the GIS will assist it in devising strategies to prevent further habitat damage without stifling growth.

Submit a comment

Comments [ 1 ]

  1. July 9, 2011 4:46am MST
    by Mccayde
    Yup, that'll do it. You have my apprceiation.
Sensors & Systems | Monitoring, Analyzing and Adapting to Global Change | Stay in tune with the transformation. Subscribe to the free weekly newsletter.