3.4.1 Integration with Other Sensor Modalities
In addition to using image-based techniques to enhance classification accuracy
(e.g., texture information and contextual editing; see Sect. 3.2), integrating multispectral with other mapping technologies is also a promising option. For
example, acoustic and optical instruments provide distinct but potentially complementary data regarding the nature of benthic communities (Riegl and Purkis
2005). Acoustic remote sensing can provide seabed roughness (rugosity), seabed
hardness, and water depth (Chaps. 8–9). An evaluation in Glovers Atoll, Belize
indicated that the accuracy of maps based on the depth invariant index using
IKONOS was enhanced when three acoustic measures were added into the analysis (Bejarano et al. 2010). LiDAR remote sensing can also provide measures of
seabed rugosity and water depth (Chaps. 5–7), suggesting another strong avenue
for data integration. Further, acoustic and LiDAR remote sensing can extend
observations of the seabed into more turbid and deeper waters, beyond where
observation becomes limited with just optical remote sensing. For example,
IKONOS imagery and acoustic remote sensing data were integrated to map sedimentary structure in a high-latitude reef-like setting in Cabo Pulmo, Mexico
(Riegl et al. 2007). Similarly, in New Caledonia, reef-associated geomorphology
was mapped using a combination of Landsat ETM+ multispectral remote sensing
and acoustic multibeam observations (Andréfouët et al. 2009a).
Another benefit of data integration is the ability to utilize depth information to
perform a water column correction of the optical bands (Purkis 2005; Bejarano
et al. 2010). Bathymetry data can also be utilized as an additional data layer in
spatial modeling. Garza-Perez et al. (2004) predicted reef bottom features using
spatial modeling based on a combination of environmental data, IKONOS imagery
and a digital topographic model. The maps generated by this procedure showed
higher classification accuracy than maps generated using only traditional unsupervised classification of the IKONOS image. In all cases of data integration, the
proper spatial alignment of different data sources is important to further enhance
classification accuracy (Andréfouët and Clareboudt 2000; Andréfouët 2008).
These studies indicate the effectiveness of integrating data from multiple data
sources. Additionally, the emerging importance of coral reefs at marginal (e.g.,
high-latitude, turbid, and moderately deep) settings as refugia for corals under
stress from climate change will further enhance the need for the use of such
integrated mapping techniques.
3.4.2 Integration with Field Monitoring
While in situ monitoring can cover small areas in superb detail, the measurements
and observations can be unrepresentative when extended over larger areas.
Linking remote sensing, which provide spatially extensive surveys, with in situ
3 Multispectral Applications
69
In addition to using image-based techniques to enhance classification accuracy
(e.g., texture information and contextual editing; see Sect. 3.2), integrating multispectral with other mapping technologies is also a promising option. For
example, acoustic and optical instruments provide distinct but potentially complementary data regarding the nature of benthic communities (Riegl and Purkis
2005). Acoustic remote sensing can provide seabed roughness (rugosity), seabed
hardness, and water depth (Chaps. 8–9). An evaluation in Glovers Atoll, Belize
indicated that the accuracy of maps based on the depth invariant index using
IKONOS was enhanced when three acoustic measures were added into the analysis (Bejarano et al. 2010). LiDAR remote sensing can also provide measures of
seabed rugosity and water depth (Chaps. 5–7), suggesting another strong avenue
for data integration. Further, acoustic and LiDAR remote sensing can extend
observations of the seabed into more turbid and deeper waters, beyond where
observation becomes limited with just optical remote sensing. For example,
IKONOS imagery and acoustic remote sensing data were integrated to map sedimentary structure in a high-latitude reef-like setting in Cabo Pulmo, Mexico
(Riegl et al. 2007). Similarly, in New Caledonia, reef-associated geomorphology
was mapped using a combination of Landsat ETM+ multispectral remote sensing
and acoustic multibeam observations (Andréfouët et al. 2009a).
Another benefit of data integration is the ability to utilize depth information to
perform a water column correction of the optical bands (Purkis 2005; Bejarano
et al. 2010). Bathymetry data can also be utilized as an additional data layer in
spatial modeling. Garza-Perez et al. (2004) predicted reef bottom features using
spatial modeling based on a combination of environmental data, IKONOS imagery
and a digital topographic model. The maps generated by this procedure showed
higher classification accuracy than maps generated using only traditional unsupervised classification of the IKONOS image. In all cases of data integration, the
proper spatial alignment of different data sources is important to further enhance
classification accuracy (Andréfouët and Clareboudt 2000; Andréfouët 2008).
These studies indicate the effectiveness of integrating data from multiple data
sources. Additionally, the emerging importance of coral reefs at marginal (e.g.,
high-latitude, turbid, and moderately deep) settings as refugia for corals under
stress from climate change will further enhance the need for the use of such
integrated mapping techniques.
3.4.2 Integration with Field Monitoring
While in situ monitoring can cover small areas in superb detail, the measurements
and observations can be unrepresentative when extended over larger areas.
Linking remote sensing, which provide spatially extensive surveys, with in situ
3 Multispectral Applications
69