the most recent general applications in these areas needed to evaluate the potential for
data synthesis using different sensors.
3.1 LOW SPATIAL RESOLUTION (≥500 METER) GLOBAL LAND (COASTAL)
PRODUCTS
Multiple global or nearly global land cover products are now available. These
global products make it possible to use GIS and modeling tools to develop regional and
global derived analyses include land cover adjacent to the coastal zone without
conducting primary classifications. With all the varied uses of land cover maps, it is
important to know specific information about each map that describes the author’s
purpose, methods, errors uncertainties or biases in the final product, and validation
results or accuracy assessments. Land cover products and measurement of land cover
change are important components in many analyses of coastal habitats. Change
detection can often require that multiple sources of data be combined (e.g. Petit and
Lambin 2001). Newly released global digital elevation models (DEMs) built from
Shuttle Radar Topography Mission (SRTM) data can be combined with land cover data
to allow improved estimation of vulnerability to erosion or flooding. For example, by
using land cover data in the context of the physical and environmental factors of the
landscape (slope, soil type, precipitation regime), it is possible to estimate relative
erosion rates (Bryant et al., 1998) which can affect coast waters.
3.1.1 IGBP DISCover global landcover
The International Geosphere Biosphere Program (IGBP) began in 1992 to produce
the first global, validated land cover data set called DISCover. The completion of this
enormous project was preceded by five years of technical planning and oversight by the
IGBP-DIS Land Cover Working Group who defined the technical specifications
involved in producing a new global land cover database (Loveland et al., 1999). The
first global classification was completed in 1997 and the validation of this data set was
concluded in 1999. The data set consists of 17 land cover classes derived from 1-km
Advanced Very High Resolution Radiometer (AVHRR) imagery (Table 2). The
classification process was divided into four major steps: 1) data set preparation and
assessment of AVHRR image quality; 2) computation of monthly Normalized
Difference Vegetation Indices; 3) unsupervised clustering derived from the K-Means
algorithm to construct preliminary greenness classes (Belward et al., 1999);
4) development of a method to interpret land cover regions based on seasonal changes
in greenness; and 5) final land cover product generation (Loveland et al., 1999). Major
caveats to the use of the dataset are the image quality of the AVHRR data that was used
in the analysis and in difficulties with Landsat TM imagery that was used for validation
(Husak et al., 1999). Classification accuracy of the different land cover types varied
from 60 to 90% (Scepan, 1999). The IGBP DISCover data, or “Global Land Cover
Characteristics Database Version 2.0 Based on 1 km AVHRR data (April 1992-March
1993)” is available from EROS Data Center (EDC) Distributed Active Archive Center
(DAAC).
Global landcover datasets are most suited to studies that need to include the extent
of human modified areas (such as agriculture) compared to natural habitats (such as
forest or wetland). Unfortunately for use in coastal management studies, the IGBP
classification scheme, does not specifically distinguish coastal wetland types such as
mangroves (Table 2).
,
288
Robinson, Andréfouët and Burke
data synthesis using different sensors.
3.1 LOW SPATIAL RESOLUTION (≥500 METER) GLOBAL LAND (COASTAL)
PRODUCTS
Multiple global or nearly global land cover products are now available. These
global products make it possible to use GIS and modeling tools to develop regional and
global derived analyses include land cover adjacent to the coastal zone without
conducting primary classifications. With all the varied uses of land cover maps, it is
important to know specific information about each map that describes the author’s
purpose, methods, errors uncertainties or biases in the final product, and validation
results or accuracy assessments. Land cover products and measurement of land cover
change are important components in many analyses of coastal habitats. Change
detection can often require that multiple sources of data be combined (e.g. Petit and
Lambin 2001). Newly released global digital elevation models (DEMs) built from
Shuttle Radar Topography Mission (SRTM) data can be combined with land cover data
to allow improved estimation of vulnerability to erosion or flooding. For example, by
using land cover data in the context of the physical and environmental factors of the
landscape (slope, soil type, precipitation regime), it is possible to estimate relative
erosion rates (Bryant et al., 1998) which can affect coast waters.
3.1.1 IGBP DISCover global landcover
The International Geosphere Biosphere Program (IGBP) began in 1992 to produce
the first global, validated land cover data set called DISCover. The completion of this
enormous project was preceded by five years of technical planning and oversight by the
IGBP-DIS Land Cover Working Group who defined the technical specifications
involved in producing a new global land cover database (Loveland et al., 1999). The
first global classification was completed in 1997 and the validation of this data set was
concluded in 1999. The data set consists of 17 land cover classes derived from 1-km
Advanced Very High Resolution Radiometer (AVHRR) imagery (Table 2). The
classification process was divided into four major steps: 1) data set preparation and
assessment of AVHRR image quality; 2) computation of monthly Normalized
Difference Vegetation Indices; 3) unsupervised clustering derived from the K-Means
algorithm to construct preliminary greenness classes (Belward et al., 1999);
4) development of a method to interpret land cover regions based on seasonal changes
in greenness; and 5) final land cover product generation (Loveland et al., 1999). Major
caveats to the use of the dataset are the image quality of the AVHRR data that was used
in the analysis and in difficulties with Landsat TM imagery that was used for validation
(Husak et al., 1999). Classification accuracy of the different land cover types varied
from 60 to 90% (Scepan, 1999). The IGBP DISCover data, or “Global Land Cover
Characteristics Database Version 2.0 Based on 1 km AVHRR data (April 1992-March
1993)” is available from EROS Data Center (EDC) Distributed Active Archive Center
(DAAC).
Global landcover datasets are most suited to studies that need to include the extent
of human modified areas (such as agriculture) compared to natural habitats (such as
forest or wetland). Unfortunately for use in coastal management studies, the IGBP
classification scheme, does not specifically distinguish coastal wetland types such as
mangroves (Table 2).
,
288
Robinson, Andréfouët and Burke