were overlaid on the 2003 image. The polygon boundaries were then edited to
match the distribution of features in the 2003 image and corroborated using fieldsurvey data. Any changes in polygon boundaries were considered to be temporal
changes in the spatial extent and/or position of a community. Subsequently, the
edited 2003 polygons were used, along with field-survey data, to interpret the 2002
imagery and the 2002 polygons were used, along with field-survey data, to
interpret the 1997 imagery.
The 1989 image, acquired soon after tropical cyclone Firinga, was so different
from the 1997 image that the polygon overlay method could not be used. Because
of the lack of contemporaneous field-survey data, the 1987 map from Naim (1989)
was used as a reference to guide the visual interpretations prior to 1989. The lack
of historical field-survey data, a common limitation when using archived historical
imagery in studies, also prevented accuracy assessments from being performed for
these images.
GIS methods were used to identify community polygon changes between each
successive map pair. Community polygon differences were considered noise when
the area/perimeter ratio was near zero (±1 %). The resulting difference maps
allow users to track changes in reef community composition through time.
Scopélitis et al. (2007) showed that, despite the challenges of identifying features found consistently throughout the imagery and a lack of historical in situ
data, aerial photographs can be a valuable resource for a time series analysis of the
spatial extent of coral reef communities. By pushing back the baseline, a better
picture of a coral reef’s resilience (or lack of) can be observed and documented,
allowing managers the opportunity to better understand the long-term trends with
respect to current-day stressors.
2.4.3 Astronaut Photography as a Secondary Data Source
Coral pinnacles rising up from the lagoon floor and dotting the sea surface or nearsurface are an important feature, both biologically and navigationally, of many
atolls. Mapping of these elements using remotely sensed imagery can be a challenging task due to their spectral similarity with frequently present small clouds.
Comparing multi-temporal satellite datasets can help solve this problem; however,
the cost of obtaining multiple images can be prohibitive. In order to address this
problem, Andréfouët and Robinson (2003) undertook a study combining freely
available astronaut-acquired space photography and video from the NASA Space
Shuttle with satellite imagery to distinguish clouds from coral pinnacles.
For this study, both astronaut-acquired hard-copy photographs and deinterlaced
digital still images extracted from high-definition television (HDTV) video were
merged with previously acquired SPOT HRV and/or Landsat ETM+ imagery for
84 atolls throughout the South Pacific. The hard photographs were digitized at
2,400 ppi (10.6 lm/pixel). Geometric rectification of the Space Shuttle imagery
(both digitized hard-copy photographs and deinterlaced stills pulled from the
42
S. A. Cochran
match the distribution of features in the 2003 image and corroborated using fieldsurvey data. Any changes in polygon boundaries were considered to be temporal
changes in the spatial extent and/or position of a community. Subsequently, the
edited 2003 polygons were used, along with field-survey data, to interpret the 2002
imagery and the 2002 polygons were used, along with field-survey data, to
interpret the 1997 imagery.
The 1989 image, acquired soon after tropical cyclone Firinga, was so different
from the 1997 image that the polygon overlay method could not be used. Because
of the lack of contemporaneous field-survey data, the 1987 map from Naim (1989)
was used as a reference to guide the visual interpretations prior to 1989. The lack
of historical field-survey data, a common limitation when using archived historical
imagery in studies, also prevented accuracy assessments from being performed for
these images.
GIS methods were used to identify community polygon changes between each
successive map pair. Community polygon differences were considered noise when
the area/perimeter ratio was near zero (±1 %). The resulting difference maps
allow users to track changes in reef community composition through time.
Scopélitis et al. (2007) showed that, despite the challenges of identifying features found consistently throughout the imagery and a lack of historical in situ
data, aerial photographs can be a valuable resource for a time series analysis of the
spatial extent of coral reef communities. By pushing back the baseline, a better
picture of a coral reef’s resilience (or lack of) can be observed and documented,
allowing managers the opportunity to better understand the long-term trends with
respect to current-day stressors.
2.4.3 Astronaut Photography as a Secondary Data Source
Coral pinnacles rising up from the lagoon floor and dotting the sea surface or nearsurface are an important feature, both biologically and navigationally, of many
atolls. Mapping of these elements using remotely sensed imagery can be a challenging task due to their spectral similarity with frequently present small clouds.
Comparing multi-temporal satellite datasets can help solve this problem; however,
the cost of obtaining multiple images can be prohibitive. In order to address this
problem, Andréfouët and Robinson (2003) undertook a study combining freely
available astronaut-acquired space photography and video from the NASA Space
Shuttle with satellite imagery to distinguish clouds from coral pinnacles.
For this study, both astronaut-acquired hard-copy photographs and deinterlaced
digital still images extracted from high-definition television (HDTV) video were
merged with previously acquired SPOT HRV and/or Landsat ETM+ imagery for
84 atolls throughout the South Pacific. The hard photographs were digitized at
2,400 ppi (10.6 lm/pixel). Geometric rectification of the Space Shuttle imagery
(both digitized hard-copy photographs and deinterlaced stills pulled from the
42
S. A. Cochran