Scopélitis et al. (2009) incorporated aerial photographs covering a span of
30 years (1973, 1978, 1989, 1997, and 2003) with two QuickBird satellite images
(2002, 2006) to scale-up quantitative in situ data to document changes on SaintLeu, a high-energy, fringing coral reef off La Réunion in the South West Indian
Ocean. During the 35 year time span covered by the imagery, Saint Leu was
regularly impacted by seasonal storms; however, two large tropical cyclones
caused significant additional disturbances. Sedimentation from storm run-off, in
addition to the mechanical wave damage caused by category 5 tropical cyclone
Firinga in 1989, caused severe mortality to the reefs (Naim et al. 1997). A significant bleaching event occurred 2 months after category 4 tropical cyclone Dina
(2002), but physical damage to the coral reef during that storm was not as severe as
the damage from Firinga. Table 2.2 summarizes the overlap of these storms’ time
periods with the data used in this study.
A pan-sharpened 2006 QuickBird image was used as a reference to geometrically correct the historical imagery. A minimum of 10 control points were selected
on each image, choosing easily identifiable objects such as buildings, roads, and
sand patches when possible, and a first-order polynomial algorithm was applied
(RMSE \ 0.5). Because the images were individually interpreted in GIS using
visual analysis, no attempt at radiometric normalization was made between
imagery.
Scopélitis et al. (2007) defined different coral habitats using a combination of
coral growth form, life status, taxonomy, and substratum. Because of the challenges in identifying some of these parameters consistently throughout the historical images, a broader classification scheme was created on a community basis
for this study, defined only by coral growth form and substratum. This reduced
level of detail resulted in 15 thematic classes that could be tracked throughout
most of the historical imagery.
Beginning with the 2006 QuickBird image and working backward through
time, coral community maps were created from all the images. Prior to visual
interpretation, a mask was applied to the images to exclude any land, exposed
beaches and breaking waves. An identical, comprehensive mask was used for all
the images in order to eliminate variability between these features. Field-survey
data from 2007 were considered to be representative of coral communities seen in
the 2006 QuickBird image since no major disturbance occurred between the date
of image acquisition and the in situ data observations; these data were used to
guide the 2006 map creation. To create the 2003 thematic map, the 2006 polygons
Table 2.2 Summary of available data and major events on Saint-Leu reef (modified from
Scopélitis et al. 2009)
1973 1978 1987 1989
1993 1997 2000 2002 2003 2006 2007
Aerial photos
X
X
X
X
X
Satellite images
X
X
In situ data
X
X
X
X
X
X
Tropical cyclone
Firinga
Dina
2 Photography Application
41
30 years (1973, 1978, 1989, 1997, and 2003) with two QuickBird satellite images
(2002, 2006) to scale-up quantitative in situ data to document changes on SaintLeu, a high-energy, fringing coral reef off La Réunion in the South West Indian
Ocean. During the 35 year time span covered by the imagery, Saint Leu was
regularly impacted by seasonal storms; however, two large tropical cyclones
caused significant additional disturbances. Sedimentation from storm run-off, in
addition to the mechanical wave damage caused by category 5 tropical cyclone
Firinga in 1989, caused severe mortality to the reefs (Naim et al. 1997). A significant bleaching event occurred 2 months after category 4 tropical cyclone Dina
(2002), but physical damage to the coral reef during that storm was not as severe as
the damage from Firinga. Table 2.2 summarizes the overlap of these storms’ time
periods with the data used in this study.
A pan-sharpened 2006 QuickBird image was used as a reference to geometrically correct the historical imagery. A minimum of 10 control points were selected
on each image, choosing easily identifiable objects such as buildings, roads, and
sand patches when possible, and a first-order polynomial algorithm was applied
(RMSE \ 0.5). Because the images were individually interpreted in GIS using
visual analysis, no attempt at radiometric normalization was made between
imagery.
Scopélitis et al. (2007) defined different coral habitats using a combination of
coral growth form, life status, taxonomy, and substratum. Because of the challenges in identifying some of these parameters consistently throughout the historical images, a broader classification scheme was created on a community basis
for this study, defined only by coral growth form and substratum. This reduced
level of detail resulted in 15 thematic classes that could be tracked throughout
most of the historical imagery.
Beginning with the 2006 QuickBird image and working backward through
time, coral community maps were created from all the images. Prior to visual
interpretation, a mask was applied to the images to exclude any land, exposed
beaches and breaking waves. An identical, comprehensive mask was used for all
the images in order to eliminate variability between these features. Field-survey
data from 2007 were considered to be representative of coral communities seen in
the 2006 QuickBird image since no major disturbance occurred between the date
of image acquisition and the in situ data observations; these data were used to
guide the 2006 map creation. To create the 2003 thematic map, the 2006 polygons
Table 2.2 Summary of available data and major events on Saint-Leu reef (modified from
Scopélitis et al. 2009)
1973 1978 1987 1989
1993 1997 2000 2002 2003 2006 2007
Aerial photos
X
X
X
X
X
Satellite images
X
X
In situ data
X
X
X
X
X
X
Tropical cyclone
Firinga
Dina
2 Photography Application
41