5. Multi-temporal analysis techniques for mapping and monitoring changes in
coastal and coral reef environments
Effective management and monitoring of coastal environments requires an
integrative approach for selecting remotely sensed data to monitor changes, as
demonstrated in earlier sections of this chapter (and see Phinn et al., 2001a). The ability
of agencies to effect their monitoring requirements is dependent on the availability of
timely, accurate, and comprehensive information on the type, distribution, and rate of
change (Phinn et al., 2000a). Remotely sensed data are particularly suitable in change
detection applications, as this approach is relatively cost effective (Mumby et al., 1999)
and can provide repeated, non-intrusive sampling over large coastal areas (Green et al.,
1996). Remotely sensed data have been used in coastal and aquatic environments to
study reef geography and reef form, (Kuchler et al., 1986), assess water quality and
benthic and inter-tidal flora (Phinn et al., 2001a), and to map littoral and shallow
marine habitats, bathymetry, suspended sediment plumes, and coastal currents (Dekker
and Seyhan, 1988; Green et al., 1996; Dekker et al., 2001 a,b). However, the successful
monitoring of change and environmental processes requires significant additional
analysis of remotely sensed data products (Jensen, 1996b; Coppin, 2003; Treitz, 2003).
This section outlines the types of change that can be detected from remotely sensed
data, the image pre-processing requirements and change and trend detection techniques
required for operational change detection, and the presentation of change detection
results for managers, agencies, and stakeholders.
5.1 TYPES OF ENVIRONMENTAL CHANGE AND PROCESSES THAT CAN BE
DETECTED FROM REMOTELY SENSED DATA FOR COASTAL ECOSYSTEMS
The changes and processes that can be distinguished using remotely sensed data
can be loosely classified into coastal landcover, water quality and substrate/benthos
composition. The following sections outline selected previous work in these
ecosystems.
5.1.1 Coastal Landcover
Studies of coastal landcover change have primarily used the Landsat series of
sensors. Landsat data provides a synoptic view of landscape processes at a regional
scale, however more detailed mapping can be achieved with high spatial resolution
airborne and satellite sensors (Phinn et al., 2000a). Multi-temporal post-classification
studies using both the Landsat Multispectral Scanner (MSS) and Thematic Mapper
(TM) sensors to detect coastal landscape changes in the Majahual system, along the
Mexican Pacific were conducted by Ruiz Luna and Berlanga Robles (1999). They
classified change in six land-use classes (mangrove, lagoon, saltmarsh, dry forest,
secondary succession, and agriculture) initially using four (Ruiz Luna and Berlanga
Robles, 1999) and later six scenes (Berlanga Robles and Ruiz Luna, 2002) to evaluate
trends of changes between the classes. Other examples of mangrove mapping include
work by Hill et al. (1994), who used SPOT to classify mangrove change in the
Ba River Delta, Fiji. Jinnahtul Islam et al. (1997) used ancillary data to enhance their
change detection of mangrove forest of the Sunderbans region of Bangladesh over a 54
year period using interpreted aerial photography. More specialised work by Trepanier
et al. (2002) used SPOT images to determining the accumulation-erosion budget for a
14 km portion of coastline in Vietnam.
233
Integrated Information Acquisition and Management
coastal and coral reef environments
Effective management and monitoring of coastal environments requires an
integrative approach for selecting remotely sensed data to monitor changes, as
demonstrated in earlier sections of this chapter (and see Phinn et al., 2001a). The ability
of agencies to effect their monitoring requirements is dependent on the availability of
timely, accurate, and comprehensive information on the type, distribution, and rate of
change (Phinn et al., 2000a). Remotely sensed data are particularly suitable in change
detection applications, as this approach is relatively cost effective (Mumby et al., 1999)
and can provide repeated, non-intrusive sampling over large coastal areas (Green et al.,
1996). Remotely sensed data have been used in coastal and aquatic environments to
study reef geography and reef form, (Kuchler et al., 1986), assess water quality and
benthic and inter-tidal flora (Phinn et al., 2001a), and to map littoral and shallow
marine habitats, bathymetry, suspended sediment plumes, and coastal currents (Dekker
and Seyhan, 1988; Green et al., 1996; Dekker et al., 2001 a,b). However, the successful
monitoring of change and environmental processes requires significant additional
analysis of remotely sensed data products (Jensen, 1996b; Coppin, 2003; Treitz, 2003).
This section outlines the types of change that can be detected from remotely sensed
data, the image pre-processing requirements and change and trend detection techniques
required for operational change detection, and the presentation of change detection
results for managers, agencies, and stakeholders.
5.1 TYPES OF ENVIRONMENTAL CHANGE AND PROCESSES THAT CAN BE
DETECTED FROM REMOTELY SENSED DATA FOR COASTAL ECOSYSTEMS
The changes and processes that can be distinguished using remotely sensed data
can be loosely classified into coastal landcover, water quality and substrate/benthos
composition. The following sections outline selected previous work in these
ecosystems.
5.1.1 Coastal Landcover
Studies of coastal landcover change have primarily used the Landsat series of
sensors. Landsat data provides a synoptic view of landscape processes at a regional
scale, however more detailed mapping can be achieved with high spatial resolution
airborne and satellite sensors (Phinn et al., 2000a). Multi-temporal post-classification
studies using both the Landsat Multispectral Scanner (MSS) and Thematic Mapper
(TM) sensors to detect coastal landscape changes in the Majahual system, along the
Mexican Pacific were conducted by Ruiz Luna and Berlanga Robles (1999). They
classified change in six land-use classes (mangrove, lagoon, saltmarsh, dry forest,
secondary succession, and agriculture) initially using four (Ruiz Luna and Berlanga
Robles, 1999) and later six scenes (Berlanga Robles and Ruiz Luna, 2002) to evaluate
trends of changes between the classes. Other examples of mangrove mapping include
work by Hill et al. (1994), who used SPOT to classify mangrove change in the
Ba River Delta, Fiji. Jinnahtul Islam et al. (1997) used ancillary data to enhance their
change detection of mangrove forest of the Sunderbans region of Bangladesh over a 54
year period using interpreted aerial photography. More specialised work by Trepanier
et al. (2002) used SPOT images to determining the accumulation-erosion budget for a
14 km portion of coastline in Vietnam.
233
Integrated Information Acquisition and Management