satellite imagery is more appropriate for studying reef
geomorphology than reef biology. The combined spatial
and spectral resolutions of satellite sensors were not
capable of reliably distinguishing habitats with relatively
high inter-habitat similarity (Bray±Curtis similarity, 60
to 80%). This was borne out by the high variability in
accuracy associated with individual habitat classes. The
poor separability of spectra rendered the underlying
discriminant function (supervised classi®cation) unable
to assign pixels to appropriate habitat classes, and re-
sulted in large and variable allocation errors.
CASI consistently provided the most accurate results.
Even ®ne habitat-discrimination was possible with an
accuracy of 81% (almost double that achieved with any
satellite). CASI has the advantage of oering tremen-
dous ¯exibility to the user. In this case, four narrow
spectral bands were set to penetrate water which in-
creased the likelihood of distinguishing habitat spectra
(band settings 402.5 to 421.8 nm, 453.4 to 469.2 nm,
531.1 to 543.5 nm, 571.9 to 584.3 nm). (Note: if we had
not also been interested in mangrove habitats, up to eight
spectral bands could have been selected for this purpose.)
Whilst CASI was found to provide signi®cantly greater
accuracies than aerial photography, the comparison was
not entirely balanced. Although the habitat categories
were comparable, Sheppard et al. (1995) mapped a much
larger area than that tested forCASI (14 600 ha vs 100 ha).
It is perhaps safest to conclude that for comparable areas,
CASI would be at least as good as aerial photography (for
further discussion see Mumby et al. 1997b).
The cost-eectiveness of remote-sensing methods for
detailed habitat-mapping is detailed in Table 4.
Potential limitations of this study
Satellite imagery is usually acquired from an image ar-
chive and it is common to ®nd considerable disparity
between the date of imagery acquisition and the date of
®eld work (Luckzovich et al. 1993). If the landscape (or
seascape) has undergone change in the intervening pe-
riod, ®eld data may be inappropriate and may result in
lower map accuracies than would have been expected if
sampling dates were similar. It is dicult to predict what
eect such temporal disparity might have had on the
present study. However, since Landsat TM was the
oldest data source and was still found to be the most
accurate, it seems unlikely to have aected our conclu-
sions markedly. It is possible that greater accuracies may
have been obtained for Landsat TM if more recent
satellite data had been available. We attempted to
ameliorate such eects wherever possible by avoiding
habitat boundaries when sampling.
Representativeness of results
The results presented here should be representative of
most Caribbean reef systems with banks and fringing
satellite imagery is more appropriate for studying reef
geomorphology than reef biology. The combined spatial
and spectral resolutions of satellite sensors were not
capable of reliably distinguishing habitats with relatively
high inter-habitat similarity (Bray±Curtis similarity, 60
to 80%). This was borne out by the high variability in
accuracy associated with individual habitat classes. The
poor separability of spectra rendered the underlying
discriminant function (supervised classi®cation) unable
to assign pixels to appropriate habitat classes, and re-
sulted in large and variable allocation errors.
CASI consistently provided the most accurate results.
Even ®ne habitat-discrimination was possible with an
accuracy of 81% (almost double that achieved with any
satellite). CASI has the advantage of oering tremen-
dous ¯exibility to the user. In this case, four narrow
spectral bands were set to penetrate water which in-
creased the likelihood of distinguishing habitat spectra
(band settings 402.5 to 421.8 nm, 453.4 to 469.2 nm,
531.1 to 543.5 nm, 571.9 to 584.3 nm). (Note: if we had
not also been interested in mangrove habitats, up to eight
spectral bands could have been selected for this purpose.)
Whilst CASI was found to provide signi®cantly greater
accuracies than aerial photography, the comparison was
not entirely balanced. Although the habitat categories
were comparable, Sheppard et al. (1995) mapped a much
larger area than that tested forCASI (14 600 ha vs 100 ha).
It is perhaps safest to conclude that for comparable areas,
CASI would be at least as good as aerial photography (for
further discussion see Mumby et al. 1997b).
The cost-eectiveness of remote-sensing methods for
detailed habitat-mapping is detailed in Table 4.
Potential limitations of this study
Satellite imagery is usually acquired from an image ar-
chive and it is common to ®nd considerable disparity
between the date of imagery acquisition and the date of
®eld work (Luckzovich et al. 1993). If the landscape (or
seascape) has undergone change in the intervening pe-
riod, ®eld data may be inappropriate and may result in
lower map accuracies than would have been expected if
sampling dates were similar. It is dicult to predict what
eect such temporal disparity might have had on the
present study. However, since Landsat TM was the
oldest data source and was still found to be the most
accurate, it seems unlikely to have aected our conclu-
sions markedly. It is possible that greater accuracies may
have been obtained for Landsat TM if more recent
satellite data had been available. We attempted to
ameliorate such eects wherever possible by avoiding
habitat boundaries when sampling.
Representativeness of results
The results presented here should be representative of
most Caribbean reef systems with banks and fringing
การแปล กรุณารอสักครู่..
