Practically, this means that a series of field measurements from known locations
are used to model a distribution function of the measured parameter over time and
space. This function can then be used to estimate a value at the time of the remote
sensing data acquisition (e.g., water depth as a function of tidal variations), which
then can be compared with the measured value derived from the remote sensing
data (Gregg and Casey 2004).
14.3 Validation Literature Review
The methods used to collect reference data and validate image products described
above were mostly derived from terrestrial remote sensing, but they are also used
in coral reef remote sensing studies (Ma and Redmond 1995; Stehman and
Czaplewski 1998; Foody 2002). To provide context as to how these methods are
routinely applied in coral reef environments, the following section presents a
review of 80 papers (Table 14.2; Roelfsema 2009). The 80 remote sensing papers
focused on coral reef and seagrass mapping in coastal environments, including 66
published in the period 2002–2009, 9 in 1992–2002, and 5 in 1976–1992. Of the
80 publications, 41 were published in remote sensing journals and 39 in other
journals related to reef or coastal environments (e.g., Coral Reefs, Journal of
Coastal Research, Estuaries). The information from the review covers: (1) mapping approaches used (Sect. 14.3.1; Table 14.3); (2) the associated sampling
designs used (Sect. 14.3.2; Table 14.4); and (3) accuracy measures used for specific mapping purposes (Sect. 14.3.3; Table 14.5).
14.3.1 Mapping Approaches
To characterize the types of mapping approaches applied in coral reef environments, we divided 80 peer-reviewed studies on coastal and marine remote sensing
into two main habitats, coral reefs (54 papers) and seagrass (26 papers). We then
subdivided papers on coral reefs based on the reef zones studied: barrier, platform
and fringing reefs (7 papers); barrier and platform reefs (14 papers); fringing reefs
(26 papers); and lagoons (8 papers). The areal extent covered in each of the papers
varied from: 0–10 km
2 (13 papers); 10–100 km
2 (37 papers); 100–360 km
2 (15
papers) and [360 km
2 (10 papers). Studies included in the 0–10 km
2 group were
mainly proof-of-concept studies for developing specific new applications for coral
reef habitat mapping (e.g., Rowlands et al. 2008).
The various mapping approaches were characterized according to: remote
sensing data type, calibration and validation data type(s), and mapping purpose
(Table 14.3). Studies were also grouped by type of image data, method for collecting calibration and validation data, and mapping technique used to transform
image data to a map product. It should be noted that some of the 80 papers used
382
C. M. Roelfsema and S. R. Phinn
are used to model a distribution function of the measured parameter over time and
space. This function can then be used to estimate a value at the time of the remote
sensing data acquisition (e.g., water depth as a function of tidal variations), which
then can be compared with the measured value derived from the remote sensing
data (Gregg and Casey 2004).
14.3 Validation Literature Review
The methods used to collect reference data and validate image products described
above were mostly derived from terrestrial remote sensing, but they are also used
in coral reef remote sensing studies (Ma and Redmond 1995; Stehman and
Czaplewski 1998; Foody 2002). To provide context as to how these methods are
routinely applied in coral reef environments, the following section presents a
review of 80 papers (Table 14.2; Roelfsema 2009). The 80 remote sensing papers
focused on coral reef and seagrass mapping in coastal environments, including 66
published in the period 2002–2009, 9 in 1992–2002, and 5 in 1976–1992. Of the
80 publications, 41 were published in remote sensing journals and 39 in other
journals related to reef or coastal environments (e.g., Coral Reefs, Journal of
Coastal Research, Estuaries). The information from the review covers: (1) mapping approaches used (Sect. 14.3.1; Table 14.3); (2) the associated sampling
designs used (Sect. 14.3.2; Table 14.4); and (3) accuracy measures used for specific mapping purposes (Sect. 14.3.3; Table 14.5).
14.3.1 Mapping Approaches
To characterize the types of mapping approaches applied in coral reef environments, we divided 80 peer-reviewed studies on coastal and marine remote sensing
into two main habitats, coral reefs (54 papers) and seagrass (26 papers). We then
subdivided papers on coral reefs based on the reef zones studied: barrier, platform
and fringing reefs (7 papers); barrier and platform reefs (14 papers); fringing reefs
(26 papers); and lagoons (8 papers). The areal extent covered in each of the papers
varied from: 0–10 km
2 (13 papers); 10–100 km
2 (37 papers); 100–360 km
2 (15
papers) and [360 km
2 (10 papers). Studies included in the 0–10 km
2 group were
mainly proof-of-concept studies for developing specific new applications for coral
reef habitat mapping (e.g., Rowlands et al. 2008).
The various mapping approaches were characterized according to: remote
sensing data type, calibration and validation data type(s), and mapping purpose
(Table 14.3). Studies were also grouped by type of image data, method for collecting calibration and validation data, and mapping technique used to transform
image data to a map product. It should be noted that some of the 80 papers used
382
C. M. Roelfsema and S. R. Phinn