See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/5485760

Remote Sensing and GIS for Wetland Inventory,

Mapping and Change Analysis

Article in Journal of Environmental Management · April 2008

DOI: 10.1016/j.jenvman.2007.06.027 · Source: PubMed

CITATIONS READS

127 1,220

3 authors, including:

Lisa-Maria Rebelo Nidhi Nagabhatla

Consultative Group on International Agricult… United Nations University (UNU)
38 PUBLICATIONS 545 CITATIONS 96 PUBLICATIONS 392 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Water Security View project

All content following this page was uploaded by Nidhi Nagabhatla on 08 May 2015.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
 Journal of Environmental Management 90 (2009) 2144e2153
www.elsevier.com/locate/jenvman

Remote sensing and GIS for wetland inventory,

mapping and change analysis
L.-M. Rebelo*, C.M. Finlayson, N. Nagabhatla
International Water Management Institute, PO Box 2075, Colombo, Sri Lanka
Received 12 March 2007; received in revised form 8 May 2007; accepted 25 June 2007
Available online 25 March 2008

Abstract

A multiple purpose wetland inventory is being developed and promoted through partnerships and specific analyses at different scales in
response to past uncertainties and gaps in inventory coverage. A partnership approach is being promoted through the Ramsar Convention on
Wetlands to enable a global inventory database to be compiled from individual projects and analyses using remote sensing and GIS. Individual
projects that are currently part of this global effort are described. They include an analysis of the Ramsar sites’ database to map the distribution
of Ramsar sites across global ecoregions and to identify regions and wetland types that are under-represented in the database. Given the extent of
wetland degradation globally, largely due to agricultural activities, specific attention is directed towards the usefulness of Earth Observation in
providing information that can be used to more effectively manage wetlands. As an example, a further project using satellite data and GIS to
quantify the condition of wetlands along the western coastline of Sri Lanka is described and trends in land use due to changes in agriculture,
sedimentation and settlement patterns are outlined. At a regional scale, a project to map and assess, using remote sensing, individual wetlands
used for agriculture in eight countries in southern Africa is also described. Land cover and the extent of inundation at each site is being
determined from a multi-temporal data set of images as a base for further assessment of land use change. Integrated fully within these analyses
is the development of local capacity to plan and undertake such analyses and in particular to relate the outcomes to wetland management and to
compile data on the distribution, extent and condition of wetlands globally.
Ó 2008 Elsevier Ltd. All rights reserved.

Keywords: Wetland; Inventory; Mapping; Remote sensing; GIS

1. Introduction
The need for information in support of wetland manage-
ment is multi-scalar e from global, regional and national
assessments to guide policy-making and prioritization of
responses to more local information on specific wetlands to
guide management planning processes, including assessment
and monitoring. However, in many instances baseline wetland
inventory is incomplete, and what has been collected is often
inconsistent and/or is not readily accessible to those who need
to use it (Finlayson and Spiers, 1999). The Ramsar Conven-
tion on Wetlands has long recognised the need to develop
* Corresponding author.
E-mail address: l.rebelo@cgiar.org (L.-M. Rebelo).
techniques that can fill gaps in baseline inventory and has
supported the development and application of inventory
techniques including the application of remote sensing and
GIS (Finlayson et al., 1999; Lowry, 2006; Rosenqvist et al.,
2007; Davidson and Finlayson, 2007).
At the same time advancements in Earth Observation
coupled with ground analyses have provided opportunities
for identifying, describing and mapping the distribution of
wetlands at a range of scales from local to global (Sahagian
and Melack, 1996; Darras et al., 1999; Lehner and Döll,
2004; Fernandez-Prieto et al., 2006). However, the methods
by which wetlands have been identified or classified using
global data sets have varied and the results have often been in-
compatible or inconsistent (Finlayson et al., 1999; Lehner and
Döll, 2004). As a consequence there are some major gaps in
wetland inventory (e.g. in southern America and for peatlands)

0301-4797/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jenvman.2007.06.027
 L.-M. Rebelo et al. / Journal of Environmental Management 90 (2009) 2144e2153 2145

and inconsistencies in methods and efforts to assess the extent wetlands worldwide have been lost was seen as speculative
and condition of wetlands, and the estimates of the rate of and not supported by available data (Finlayson and D’Cruz,
wetland loss and degradation globally are incomplete or based 2005). A separate analysis of wetland extent using multiple
on unsubstantiated assertions (Finlayson et al., 2005). data sets (Lehner and Döll, 2004) produced a global estimate
Despite, or perhaps because of the absence of reliable data of 917 million hectares with substantial differences to the
the continuing loss and degradation of wetlands is of ongoing previous estimates by Finlayson et al. (1999) particularly in
concern to the Ramsar Convention that has through its in- relation to the area of wetlands in Europe and the Neotropics
volvement in recent remote sensing initiatives (Fernandez- (incorporating the Caribbean and Central and South America)
Prieto et al., 2006; Davidson and Finlayson, 2007; (Table 1). Some differences are perhaps not unexpected given
Rosenqvist et al., 2007) and the Millennium Ecosystem the differences in data and approaches being used for inven-
Assessment (Finlayson et al., 2005) promoted the need to tory analyses, but the disparity between estimates also sug-
make available a comprehensive information base on the sta- gests that further standardisation may be useful.
tus and trends of wetlands, their values, and the major drivers In response to the ongoing uncertainties over wetland area
of change. Through its triennial Conference of Parties the the compilation of a global wetland inventory is being pro-
Convention has also promoted further analyses of its existing moted through the Ramsar Convention in partnership with
data on wetlands listed as internationally important (Wetlands international, national and local organisations (Davidson and
International, 2005), and an assessment of the interactions Finlayson, 2007) and the incorporation of information from in-
between wetlands and agriculture. In response to the well- dividual projects based on existing and new data sources (e.g.
recognised data limitations and need for further information Rosenqvist et al., 2007). In particular, methods are being eval-
and analyses we report on several initiatives using Earth uated for their applicability and congruence with criteria rele-
Observation to provide such information. These initiatives vant to building a global assessment and monitoring capability
cover: (i) steps to compile a global overview of wetlands; based upon Earth Observation applications. This brings into
(ii) an analysis of the geographic representivity of wetlands play a series of methods e high amongst these are the devel-
of international importance; and (iii) analyses of change in opment of improved wetland specific methods for using newly
specific wetlands. available and advanced remote sensing data (Fernandez-Prieto
et al., 2006; Davidson and Finlayson, 2007; Rosenqvist et al.,
2. Global initiatives 2007) as well as further consideration of the usefulness of
multiple-scale approaches such as those suggested by the Con-
2.1. A global wetland inventory vention (Davidson and Finlayson, 2007).
The global inventory includes an analysis of the geographic
The achievements and limitations of global wetland inven- distribution of wetlands, based initially on existing data such
tory have been assessed and reported as part of the Millennium as that contained in the continental directories largely com-
Ecosystem Assessment (Finlayson and D’Cruz, 2005). The piled more than a decade ago (Finlayson and van der Valk,
Assessment confirmed that the extent of wetland mapping 1995) and supported by global analyses of remotely sensed
and inventory was inadequate and that the most recent esti- data (Sahagian and Melack, 1996; Darras et al., 1999; Lehner
mates of wetland extent were underestimates with significant and Döll, 2004). Further methodological developments that
gaps regionally and for various types of wetlands. Further, are under consideration include the methods and databases
the global estimate by Finlayson et al. (1999) of 1280 million being assembled for mapping irrigated areas globally (see
hectare was an underestimate (Table 1); this was derived from Thenkabail et al., 2006) and the application of new satellite
national sources and was much higher than previous estimates data resources to the challenges of a global remote sensing-
produced through Earth Observation (Finlayson et al., 1999). based approach to wetland mapping, including improved and
Additionally, assertions (e.g. Dugan, 1990) that 50% of cost effective methods to deal with seasonal and episodic inun-
dation of forested and non-forested ecosystems (Rosenqvist
et al., 2007). Additionally, geospatial analysis of existing in-
Table 1
formation is being used to assess eco-regional gaps in wetland
Estimates of global wetland area (million hectare with percentage area in
parentheses) for each of the six geopolitical regions used by the Ramsar conservation and protection within a land cover and land use
Convention on Wetlands (www.ramsar.org) context, such as that outlined in southern Africa (see below)
Region Global review of Global lakes and and at specific sites, such as the recently completed mapping
wetland resources wetlands database of wetlands in Sri Lanka (IUCN Sri Lanka and the Central
(Finlayson et al., 1999) (Lehner and Döll, 2004) Environmental Authority, 2006).
Africa 125 (10) 131 (14) As a further step, an inventory with scale-related core data
Asia 204 (16) 286 (32) elements, based on the Ramsar-recognised Asian Wetland
Europe 258 (20) 26 (3) Inventory (Finlayson et al., 2002), is being built through
Neotropics 415 (32) 159 (17) a combination of individual initiatives. This will provide an
North America 242 (19) 287 (31)
Oceania 36 (3) 28 (3)
up-to-date analysis of the extent of wetland mapping and
inventory data and techniques and make these available to
Total 1280 917
managers. Importantly, it will also develop regional
2146 L.-M. Rebelo et al. / Journal of Environmental Management 90 (2009) 2144e2153
knowledge exchange and capacity to undertake wetland inven-
tory through integrated analyses and specific capacity building
exercises and encourage the operationalisation of Earth Obser-
vation research. Fernandez-Prieto et al. (2006) point out that
simply having more data through Earth Observation research
does not mean that we have better analyses.
2.2. A partnership for inventory
In order to accomplish meaningful, accurate and up-to-date
mapping, inventory and assessment of global wetlands, it is
essential to develop a platform, which will allow the many
agencies and groups working in wetlands to work together
towards this goal. This requires not only organizational coher-
ence, but also a solid technical and scientific foundation. To
this end, a voluntary multi-stakeholder initiative for global
wetlands inventory and mapping has been proposed by the
Food and Agricultural Organisation (FAO) of the United
Nations, through the Coastal Panel of the Global Terrestrial
Observing System (C-GTOS), and the International Water
Management Institute (IWMI) and accepted by the Ramsar
Convention’s Scientific and Technical Review Panel (STRP)
(June 2006). The initiative has been developed in line with
the requirements of the United Nations Commission for Sus-
tainable Development for partnerships that promote sustain-
able development (http://www.un.org/esa/sustdev/partnerships/
csd11_partnerships_decision.htm).
The aim of the partnership is to support and introduce
coherence to global wetland and mapping efforts that will
eventually provide both the foundation and the results that
can be assembled into an accurate global inventory. The focus
of the partnership is on collaboration and capacity building.
Goals include the development of regional knowledge ex-
change and capacity to undertake wetland inventory through
integrated analyses of wetland condition and trends, and to sup-
port national-based efforts and capacity building for national
monitoring programs. Development of a standardized wetland
classification system suitable for land cover/use is an important
component of the proposal. As an initial step the partnership is
contributing to the inventory tasks being addressed by the
Ramsar Convention’s STRP covering wetland classification
and delineation, and assessing the extent and consequences
of adverse change.
The partnership is an open process and depends on collab-
oration between multiple agencies and organisation; these
features have been embraced through the relationships already
established with international and regional organisations as
represented through input to the Ramsar Convention’s Scien-
tific and Technical Review Panel (STRP). An established
example involving the STRP is the Kyoto & Carbon (K&C)
Initiative which is an international collaborative effort led by
the Japanese space agency, JAXA, to support information
needs posed by carbon cycle science and environmental con-
ventions by using data acquired by the L-band SAR instrument
(PALSAR) on the Advanced Land Observing Satellite
(ALOS), which was launched in January 2006. This contains
a dedicated wetlands theme to provide repetitive and
consistent data for wetlands characterization and disturbance
monitoring through the establishment of a fine resolution
(10e20 m) multi-temporal L-band SAR image archive
(Rosenqvist et al., 2007).
2.3. The global data set of wetlands of international
importance
The global network of wetland sites known popularly as
Ramsar sites now contains over 1600 wetlands covering
more than 135 million hectares. The database containing in-
formation on the sites is regularly updated (see www.ramsar.
org/index_list.htm) and a formal overview provided to the
Convention every 3 years (Wetlands International, 2005).
While this database constitutes the most extensive source of
inventory information on individual wetlands available to
the public it contains many gaps with some wetland types
under-represented, although exactly what is meant by under-
represented has proven difficult to define as the database is
not based on spatial data and the selection of sites is not fully
supported by a consistent or complete information base. In the
last triennial overview four categories of under-representation
were identified (Wetlands International, 2005). These were (i)
global coverage e the global distribution of designated
wetlands in relation to the global wetland resource, (ii) re-
gional/national under-representation according to appropriate
biogeographic classifications, (iii) numbers of sites designated
for certain wetland types, and (iv) numbers of sites designated
for their wetland-dependent biodiversity.
As part of an ongoing initiative to support the listing of
wetlands of international importance an analysis is underway
through the STRP to determine the geographic representivity
of the distribution of listed sites globally. The aim of this
work is to highlight geographic areas and wetland types of pri-
ority for further inventory, and areas where the use of Earth
Observation for wetland identification and delineation may
be targeted. Under-represented areas identified through this
analysis will determine future research directions as part of
the global wetland mapping and inventory efforts.
This analysis has initially been undertaken using the 825
terrestrial ecoregions identified by WWF (Olson et al.,
2001). In this instance the ecoregions are defined as units of
land containing a distinct assemblage of natural communities
and species and have been identified through a classification
framework that builds on existing biogeographic knowledge
as well as a detailed level of biogeographic resolution. In ad-
dition to sharing environmental conditions and a large majority
of their species and ecological dynamics, the components of
these ecoregions interact ecologically in ways that are critical
for their long-term persistence. As biomes are the first tier in
this global classification the distribution of the global Ramsar
sites has been investigated across these. While it has been
suggested (Olson et al., 2001) that this classification is likely
to reflect the distribution of species and communities more ac-
curately than landscape units based on models derived from
gross biophysical features such as temperature and rainfall
(Holdridge, 1967), vegetation structure or remote sensing
 L.-M. Rebelo et al. / Journal of Environmental Management 90 (2009) 2144e2153
derived spectral information (Loveland and Belward, 1997),
the distribution of wetland sites across these or similar systems
will be examined subsequently in a more detailed analysis.
The distribution of global Ramsar sites across the WWF
biomes is displayed in Figs. 1 and 2. The greatest concentra-
tion of Ramsar sites within a single biome is found in Europe
in the temperate broadleaf and mixed forests biome (30%) and
in the Mediterranean forests, woodlands and shrubs biome
(10%). This is likely a reflection of the effectiveness of the
listing processes in the countries concerned rather than an
indication of the areal extent of wetlands within these two
biomes. The area of wetlands in Europe has been estimated
as between 28 and 36 million hectares or 3e20% of the global
total, while the greatest areal wetland extent has been identified
in the Neotropics or Asia and North America (Table 1). In con-
trast the biomes with the greatest area of Ramsar wetland sites
listed are the tropical and sub-tropical grasslands, savannas and
shrublands, and flooded grasslands (Figs. 1 and 2b). Whilst
a metric for recording the representivity of Ramsar sites, or
wetland types within biomes, has not been developed, it is
evident that the number of Ramsar sites currently listed is
not representative of the estimated areal extent of wetlands in
the Ramsar regions, even though these estimates themselves
differ. The current analysis does, however, provide a basis for
considering the proportionate importance of different biomes
within the Convention, and can also be used to assess the
uses of wetlands and the key causes of wetland loss or degra-
dation. This capacity will be enhanced with the placement on
the Internet of the information on wetland distribution and
areal extent contained in the continental wetland directories
undertaken in the late 1980s and early 1990s. The wetland
directories for the Neotropics (Ducks Unlimited Inc., 2006,
Fig. 1. Distribution of Ramsar sites across WWF Terrestrial Biomes.
2147
compiled by Scott and Carbonell, 1986) and that for Asia
(IWMI, 2006, compiled by Scott, 1989) are now available on
the Internet.
Wetland sites utilised for agriculture are of interest in the
current research given the importance of wetlands for agricul-
ture and the important role of agriculture in wetland degrada-
tion (Finlayson et al., 2005). Given the duality of interaction
between agriculture and wetlands, particular emphasis is be-
ing placed on the use of Earth Observation for the provision
of improved information for better management and under-
standing of agricultural interfaces with the water regimes
that characterise specific wetland types. Analysis of the infor-
mation in the Ramsar sites’ database (as of 11 June 2006 with
1603 listed sites) shows that more than 78% of listed wetlands
support agricultural activities. The distribution of these across
the global biomes is shown in Fig. 3; over 50% of the sites in
each biome are used for agriculture.
The importance of wetlands for agriculture globally is
shown by 80e90% of sites in the forest and savanna biomes
containing agriculture; these biomes support a large proportion
of the world’s population. Mangroves are also important for
agriculture. On the whole, however, the database does not
provide further detailed information on these activities, e.g.
specific products or seasonality or area. Earth Observation
could usefully be applied to provide some of this further infor-
mation and to record seasonal and longer-term change. At the
same time, with the exception of the montane biome (No. 10),
more than 50% of wetlands sites in these biomes are threat-
ened by agriculture (Fig. 4). Again, Earth Observation could
be applied to supply further information on, for example, the
extent and seasonality of these activities and the threats they
pose to the wetlands.
2148 L.-M. Rebelo et al. / Journal of Environmental Management 90 (2009) 2144e2153
Fig. 2. Number and area of Ramsar sites within the WWF Terrestrial Biomes
(1: Tropical, subtropical moist broadleaf forest; 2: Tropical, subtropical dry
broadleaf forest; 3: Tropical, subtropical coniferous woodland; 4: Temperate
mixed broadleaf forest; 5: Temperate coniferous woodland; 6: Boreal forest/
taiga; 7: Tropical, subtropical grassland savanna and shrubland; 8: Temperate
grassland savanna and shrubland; 9: Flooded grassland; 10: Montane grass/
shrubland; 11: Tundra; 12: Mediterranean forest wood/shrubland; 13: Desert
and xeric shrubland; 14: Mangrove).
Fig. 3. Percentage of Ramsar sites which support agricultural activities in each
of the WWF Terrestrial Biomes.
Fig. 4. Percentage of Ramsar sites within each WWF Terrestrial Biomes which
are under threat from agricultural activities.
3. Case studies
3.1. Wetland change detection, Sri Lanka
In line with the hierarchical approach to wetland inventory
promoted by the Ramsar Convention and outlined in the Asian
Wetland Inventory (Finlayson et al., 2002) satellite data and
GIS are being used to quantify the condition of a wetland
complex in Sri Lanka. Effective inventory, mapping and mon-
itoring of wetlands along with assessment of the ecosystem
services they provide is being used to improve local under-
standing of the inter-related ecological and socio-economic
factors that affect the wetlands. They are also helping to
identify the inter-linkages between the various components
that comprise the wetland and that support livelihoods.
The Muthurajawela Marsh and the adjoining Negombo
Lagoon located along the western coast of Sri Lanka cover
a combined area of approximately 12,000 hectares (IUCN
Sri Lanka and the Central Environmental Authority, 2006).
Muthurajawela is the largest saline coastal peat bog in Sri
Lanka, which together with the Negombo Lagoon forms an
integrated coastal wetland ecosystem. It was listed as one of
12 priority wetlands in Sri Lanka in 1996 and its northern
part was declared a wetland sanctuary due to endemicity and
uniqueness in both marine and terrestrial components. The
surrounding landscape has long been used for agriculture
and in more recent times for aquaculture. In addition as it is
one of the most urbanised and industrialised parts of the coun-
try there is increasing pressure to convert the wetlands for
other uses. The wetland complex is located alongside a rapidly
developing urban area and is being rapidly degraded by
inadequately planned development activities and growing an-
thropogenic pressure. An inventory and assessment has been
undertaken to quantify the extent of biophysical change in
the wetland complex over the past few decades and to investi-
gate some of the socio-economic drivers of change over this
period. This included the identification of ‘indicators of stress’
in the wetland and a geospatial analysis of the extent of
ecological change and future potential threats.
 L.-M. Rebelo et al. / Journal of Environmental Management 90 (2009) 2144e2153
Fig. 5. Methodological flow of the process involved in linking ecological and socio-economic variables.
In order to detect trends in land uses a geospatial platform
(Fig. 5) was established to identify and assess the extent of
specific areal and ecological change and to analyse gaps in
the existing management and conservation strategies for the
marsh and lagoon. Multi-spectral Landsat satellite imagery
from 1992 to 2002 was used to spatially capture changes in
the land cover and land use observed in the wetlands and
adjacent land. The multi-seasonal images were processed
and classified with a supervised classifier (Maximum Likeli-
hood Algorithm). Ground data were used to refine, verify
and validate the changes in land cover and land use. Ancillary
data (e.g. topographical and land survey maps) were integrated
with the geospatial analysis to refine the land cover/use types.
Land cover change and future projections were undertaken
using a customised algorithm available in IDRISI-Andes
(Land Change Modeller) for ecological sustainability. The
utility of this algorithm for assessing accelerated land conver-
sion in tropical coastal wetlands was tested against the ancil-
lary data. Changes in wetland cover between 1992 and 2002
are shown in Fig. 6. Key changes included a loss of deep water
areas (33%), mangroves (49%) and marsh (40%) due to sedi-
mentation, infilling and clearing of vegetation and the expan-
sion of human settlements. The areal extent of agricultural
land was not significantly different. The main activities driving
change in the marsh and lagoon are expanding settlements
along with industrialisation. Industrial, domestic and agricul-
tural discharges over many years have disturbed the pattern
of water flow into the lagoon and increased sedimentation.
This increase in sedimentation is assumed to have adversely
affected fish productivity and hence the livelihood patterns
of the local people. Ground data are being collected to corrob-
orate these results in more detail and to identify more localised
2149
and specific change and how this may relate to national and
local policy and land use management.
The accuracy of the derived map was checked in the field
using ground points collected using a differential GPS with
23 ground control points collected during the pre-classification
process and 58 points collected post-classification. The accu-
racy calculated for the refined supervised classification
approach was 86% although there were ambiguities in classes
such as marshland with shrub patches and open moist area
with shrubs.
Further to the primary analysis of change between the
defined time periods, an attempt was made to depict the
contribution to overall changes in the marsh by other classes.
An analysis of change using different thresholds and defined
algorithms in IDRISI to capture the gradients of change was
used to determine the contribution to net changes in associated
wetland types. Contributions to the increase in settlements are
shown in Fig. 7. These illustrate that the largest conversions to
settlements have occurred in marshland, followed by areas of
littoral vegetation; infilling and the conversion of marshes to
settlements are the major cause of shrinkage in areal extent.
In summary, increased urbanisation and industrialisation
have resulted in the loss of marshland.
3.2. Inland wetlands in southern Africa
At a broader scale wetlands used for agriculture are being
mapped and assessed through Earth Observation in southern
Africa. This constitutes part of the global inventory project.
It is also part of a project to assess the manner in which local
communities use wetlands for multiple purposes, both now
and in the future, and to put in place or enhance mechanisms
2150 L.-M. Rebelo et al. / Journal of Environmental Management 90 (2009) 2144e2153
Fig. 6. Land cover change in the Muthurajawela Marsh and the adjoining Ne-
gombo Lagoon, Sri Lanka.
that minimise the degradation of wetlands in order to optimise
the livelihood benefits that are generated (Darradi et al., 2006).
In addition, generic guidelines, tools and methods for sustain-
able and improved land and water management in wetlands
are being generated. As wetlands in sub-Saharan Africa
typically play a vital role in supporting rural populations, the
sustainable management of these ecosystems is critical. In
order to prevent depletion of these resources a balance is
required between environmental and socio-economic needs
(Wood, 2006).
Priority wetland sites that are vulnerable to both climatic
variability and agricultural (subsistence and commercial)
Fig. 7. Contribution of changes in land cover to the increased area of settlements (km2).
activities have been selected within eight countries in southern
Africa (Fig. 8). Population pressure and increased exploitation
of resources within these wetlands and the surrounding
catchments are leading to serious degradation and loss of bio-
diversity and ecosystem services. Initial analyses have been
undertaken at one of the larger (>2000 km2) sites e Lake
Chilwa in Malawi. The inundation extent and spatial distribu-
tion of wetland types within the Lake Chilwa wetland complex
is being determined from a multi-temporal Earth Observation
data set acquired during the wet and dry seasons. In addition to
validation of the mapping results, the ecological characteris-
tics and the cultivation patterns within and across ecological
zones has been characterized through field investigations.
While remote sensing data sets are available (wet and dry sea-
son Landsat TM, and wet season ASTER), on-ground access
to the site is extremely difficult. Access roads into and around
the site are minimal, and the vegetation is near-impenetrable
even during the dry season. As ground validation to accom-
pany the remote sensing analysis is problematic, air photos
have been acquired in order to assist with image interpretation
and classification. Where possible ground data were also
collected with wetland types across each site classified accord-
ing to the FAO Land Cover Classification System (LCCS; Di
Gregorio and Jansen, 2000) for major land cover categories.
This will provide a comparison with the approach recommen-
ded by the Ramsar Convention to classify wetlands on the
basis of their hydro-geomorphology (Finlayson et al., 1999).
The spatial distribution of the dominant wetland types
across the site was classified from Landsat TM and ASTER
data using a decision tree classifier and a series of binary
decision rules. Decision Tree classifiers allow multistage clas-
sifications to be performed, recursively partitioning the input
data set into increasingly homogenous subsets. A particular
advantage of this approach is that data sets with different
spatial resolutions, as well as ancillary data sets can be used
together during the classification process. In the absence of
sufficient ground data, but acknowledging the availability of
suitable alternate data sources (an ASTER Digital Elevation
Model (DEM) and a data set of aerial photographs) which
may aid in image interpretation in addition to knowledge of
the area gained during field visits, the use of a decision tree
classifier is considered a more appropriate method than tradi-
tional statistical techniques. A set of decision rules have been
applied to a data set consisting of a dry season Landsat TM
 L.-M. Rebelo et al. / Journal of Environmental Management 90 (2009) 2144e2153 2151

Fig. 8. Location and size of wetland sites in southern Africa.

image acquired on 18th November 2005, a wet season ASTER
image (VNIR, SWIR and TIR subsystems) acquired on 21st
May 2006, and a 15 m spatial resolution ASTER DEM
(derived from the nadir and backward VNIR stereoscopic
pair and calibrated through ground-based GPS measurements).
The decision rules were derived through an analysis of the
remote sensing data sets in conjunction with a subset of the
field data (50% of the 90 sites were selected at random) and
the georeferenced air photos (Fig. 9). The Lake Chilwa wet-
land complex consists of approximately a third open water,
a third swamp and a third floodplain. Initial analysis for the
first node of the decision tree partitions the image into two
classes, terrestrial and wetland, following the definitions
used in the Wetland Classification System for South Africa
(Ewart-Smith et al., 2006). The wetland area is subsequently
partitioned into permanent open water and seasonal flood
plain, and the latter into two wet season classes, open water
and swamp. Based on field data and air photos, the area of
swamp is classified into four composite vegetation classes
based on the dominant land cover and land use. The spatial
distribution of the wetland classes is shown in Fig. 10. The
classification had an accuracy of 89%, based on the 45 field
sites that were not used in the initial analysis.
The geospatial analyses involving a combination of satellite
imagery, air photographs and ground information relating to
land cover and land use at multiple-scales (Figs. 9 and 10)
are being used to develop tools that can be used to assist man-
agers to make decisions about future land uses in wetlands that
are intensively used for agriculture and/or fisheries and are an
important resource for local communities. The land cover/use

Fig. 9. Lake Chilwa, Malawi (Landsat TM real colour composite, May 2006, ground and air photography sites).
2152 L.-M. Rebelo et al. / Journal of Environmental Management 90 (2009) 2144e2153
Fig. 10. Vegetation distribution, Lake Chilwa wetland, Malawi.
analyses will provide a base for further livelihood assessments
and management interventions across the region.
4. Capacity building
Integrated fully within the above described analyses is the
need to develop national and local community capacity to
undertake analyses and interpret the responses in a local oper-
ating environment. This is in line with international emphasis
on developing capacity within countries to undertake wetland
inventory and assessment and to ensure that local communities
are involved at all stages of the analyses and can make use of
the outcomes (Fernandez-Prieto et al., 2006). The capacity
building component of the present assignments involves
a number of stages e involvement of stakeholders, local part-
ners and/or planners in planning and implementing the project
and specific components, assisting with interpretation and the
development of models for predictive analyses, and impor-
tantly, in the transfer of information and outcomes to the wider
community. This is especially important where specific man-
agement interventions and/or further monitoring is being
recommended, as in the case of the Sri Lankan and southern
African cases outlined above. These activities also support
the need to develop dedicated capacity building activities in
order to better transfer Earth Observation technology to a wider
wetlands’ community (Fernandez-Prieto et al., 2006).
In all cases stakeholders have been involved in the planning
and development of the projects e this is seen as a critical not
just an initial step in capacity building. Where practicable the
users and data custodians are involved in the analyses, and
where field work is undertaken this is done in conjunction
with local partners. The former has enhanced knowledge ex-
change and personal development through mutual on-the-job
interaction. The latter supports better contact with local land
owners and enhances information exchange, and importantly,
also enables specific on-ground technical training. Specific
training and personal development have featured in the
analyses in Sri Lanka and southern Africa, e.g. covering image
interpretation, field survey and land cover assessment, as well
as familiarity with GPS, GIS and spatial modelling software.
The challenge with capacity building though is not to in-
clude suitable exercises in the projects; the challenge is to en-
sure they are effective. We can readily justify capacity
building, but understanding and supporting the wider capacity
needs of local partners may not be as readily achievable even
when working within an integrated or holistic project. Within
the context of the global wetland initiative research that sup-
ports local communities is not an insignificant issue e it is
seen as essential when dealing with sustainability of liveli-
hoods and landscapes. There are many examples though where
this has not been done successfully (Carbonell et al., 2001),
but it is acknowledged as part of the global wetland inventory
project that the establishment of effective relationships with
stakeholders along with the technical analyses is an important
component.
5. Summary and conclusions
In response to past uncertainties and gaps in wetland inven-
tories, a global wetland inventory is being developed through
the application of remote sensing and GIS technologies. The
inventory is being promoted through an international partner-
ship centred on the requirements of the Ramsar Convention on
Wetlands for further information on the extent and condition
of wetlands globally. The inventory is being compiled through
individual projects that consider different spatial and temporal
scales.
An analysis of the geographic distribution of wetlands is
being undertaken using continental directories largely com-
piled more than a decade ago with the explicit goal of making
these available through the Internet for further interrogation
and analysis. In conjunction with this effort the Ramsar sites
database has been analysed to determine the geographic repre-
sentivity of sites listed in the database. Whilst the site data is
not representative of the areal extent of wetlands globally it
does provide a base for considering the proportionate distribu-
tion of wetlands across ecoregions, and is also being used to
assess the regional uses of wetlands (e.g. for agriculture) and
the key causes of wetland degradation (e.g. by agriculture).
In terms of wetland management and the application of inven-
tory for management purposes a specific example of wetland
change detection is illustrated from Sri Lanka where changes
in population and land use are contributing to wetland loss and
degradation. Through the use of remote sensing and modelling
of change in land use the particular causes of wetland conver-
sion could be identified, in this instance the expansion of
settlements occurred through the abandonment of paddy fields
and the loss of littoral vegetation.
At a broader scale the extent of land cover changes within
and around selected wetlands in southern Africa has been
assessed using remote sensing. This is part of a larger project
considering the sustainability of agriculture in wetlands and
the maintenance of local livelihoods through targeted man-
agement interventions. In this respect it illustrates the
 L.-M. Rebelo et al. / Journal of Environmental Management 90 (2009) 2144e2153
connection between wetland condition and human livelihoods
and specifically the application of wetland inventory data for
planning and management purposes. Previous analyses of
wetland inventory have highlighted the difficulty of obtaining
wetland inventory data unless it is linked with practical pur-
poses such as the analysis of land cover or land use change.
Given the increasing extent of such analyses it is anticipated
that the global wetland inventory database can be extended
through combining individual analyses linked with immedi-
ately obvious practical outcomes.
Integrated fully within these analyses in the development of
local capacity to plan and undertake such analyses, and in par-
ticular to relate the outcomes to wetland management as well
as to compile data on the distribution, extent and condition of
wetlands globally.
References
Carbonell, M., Nathai-Gyan, N., Finlayson, C.M. (Eds.), 2001. Science and
Local Communities: Strengthening Partnerships for Effective Wetland
Management. Ducks Unlimited Inc., Memphis, USA, 93 pp.
Darradi, Y., Grelot, F., Morardet, S. 2006. Analysing stakeholders for sustain-
able wetland management in the Limpopo River basin: the case of
Ga-Mampa wetland, South Africa. In: Communication to the 7th WATER-
NET/WARFSA/GWP-SA Symposium ‘‘Mainstreaming IWRM in the
Development Process’’, Lilongwe, Malawi, 1e3 November 2006, 25 pp.
Darras, S., Michou, M., Sarrat, C., 1999. IGBP-DIS Wetland Data Initiative:
a First Step Towards Identifying a Global Delineation of Wetlands.
IGBP-DIS Office, Toulouse, France.
Davidson, N.C., Finlayson, C.M., 2007. Earth Observation for wetland inven-
tory, assessment and monitoring. Aquatic Conservation: Marine and Fresh-
water Ecosystems 17, 219e228.
Di Gregorio, A., Jansen, L.J.M., 2000. Land Cover Classification System (LCCS),
Classification Concepts and User Manual. FAO, ISBN 92-5-104216-0.
Ducks Unlimited Inc., 2006. A Directory of Neotropical Wetlands.
http://www.ducks.org/Conservation/International/2449/ADIRECTORYOF
NEOTROPICALWETLANDS.html.
Dugan, P.J. (Ed.), 1990. Wetland Conservation: a Review of Current Issues and
Required Action. IUCN, Gland, Switzerland.
Ewart-Smith, J., Ollis, D.J., Day, J.A., Malan, H.L., 2006. National Wetland
Inventory: Development of a Wetland Classification System for South
Africa. WRC. Report: KV 174/06.
Fernandez-Prieto, D., Arini, O., Borges, T., Davidson, N., Finlayson, M.,
Grassl, H., MacKay, H., Prigent, C., Pritchard, D., Zalidis, G., 2006. The
GlobWetland Symposium: Summary and Way Forward. Proceedings Glob-
Wetland: Looking at Wetlands from Space, October 2006, Frascati, Italy.
Finlayson, C.M., van der Valk, A.G. (Eds.), 1995. Classification and Inventory
of the World’s Wetlands. Advances in Vegetation Science, vol. 16. Kluwer
Academic Publishers, Dordrecht, 192 pp.
Finlayson, C.M., Spiers, A.G. (Eds.), 1999. Global Review of Wetland Re-
sources and Priorities for Wetland Inventory. Supervising Scientist Report
144. Supervising Scientist, Canberra.
View publication stats
2153
Finlayson, C.M., Davidson, N.C., Spiers, A.G., Stevenson, N.J., 1999. Global
wetland inventory e current status and future priorities. Marine and Fresh-
water Research 50 (8), 717e727.
Finlayson, C.M., Begg, G.W., Howes, J., Davies, J., Tagi, K., Lowry, J., 2002.
A Manual for an Inventory of Asian Wetlands (version 1.0). In: Wetlands
International Global Series, vol. 10. Wetlands International, Kuala
Lumpur, Malaysia.
Finlayson, C.M., D’Cruz, R., 2005. Chapter 20 inland water systems. In:
Hassan, R., Scholes, R., Ash, N. (Eds.), Ecosystems and Human Well-
being Volume 1: Current Status and Trends: Findings of the Condition
and Trends Working Group. Island Press, Washington DC, USA.
Finlayson, C.M., D’Cruz, R., Davidson, N.J., 2005. Ecosystem Services and
Human Well-being: Water and Wetlands Synthesis. World Resources
Institute, Washington DC, USA.
Holdridge, L.R., 1967. Life Zone Ecology. Tropical Science Centre, San Jose,
Costa Rica.
IUCN Sri Lanka and the Central Environmental Authority, 2006. National
Wetland Directory of Sri Lanka. The Central Environmental Authority
CEA, The World Conservation Union (IUCN) and the International Water
Management Institute (IWMI), Colombo, Sri Lanka.
IWMI, 2006. A Directory of Asian Wetlands. http://www.iwmi.cgiar.org/
wetlands/WetlandDir.asp.
Loveland, T.R., Belward, A.S., 1997. The IGBP-DIS global 1 km land cover
data set, DISCover first results. International Journal of Remote Sensing
18, 3289e3295.
Lehner, B., Döll, P., 2004. Development and validation of a global database of
lakes, reservoirs and wetlands. Journal of Hydrology 296 (1e4), 1e22.
Lowry, J.B.C., 2006. The Application of Low-cost GIS Software and Data for
Wetland Inventory, Assessment and Monitoring. Ramsar Convention Sec-
retariat, Gland, Switzerland. http://www.ramsar.org/lib/lib_rtr_index.htm
Ramsar Technical Report No. 2.
Olson, D.M., Dinerstein, E., Wikramanayake, E.D., Burgess, N.D., Powell, G.V.,
Underwood, E.C., D’Amico, J.A., Itoua, I., Strand, H.E., Morrison, J.C.,
Loucks, C.J., Allnutt, T.F., Ricketts, T.H., Kura, Y., Lamoreux, J.F.,
Wettengel, W.W., Hedao, P., Kassem, K.R., 2001. Terrestrial ecoregions of
the world: a new map of life on earth. Bioscience 51, 933e938.
Rosenqvist, A., Finlayson, C.M., Lowry, J., Taylor, D., 2007. The potential of
long wavelength satellite borne radar to support implementation of the
Ramsar Wetlands Convention. Aquatic Conservation: Marine and Fresh-
water Ecosystems 17, 229e244.
Sahagian, D., Melack, H. (Eds.), 1996. Global Wetland Distribution and
Functional Characterization: Trace Gases and the Hydrologic Cycle.
IGBP Secretariat, Stockholm IGBP Report No. 46.
Scott, D.A., 1989. A Directory of Asian Wetlands. IUCN, Gland, Switzerland/
Cambridge, United Kingdom.
Scott, D.A., Carbonell, M., 1986. A Directory of Neotropical Wetlands. IUCN
Conservation Monitoring Centre/International Waterfowl Research
Bureau, Cambridge, United Kingdom/Slimbridge, United Kingdom.
Thenkabail, P., Chandrashekhar, B., Turral, H., Noojipady, P., Yuanjie, L.,
Vithanage, J., Venkateswarty, D., Velpuri, M., Schull, M., 2006. An Irri-
gated Area map of the World (1999) Derived from Remote Sensing,
IWMI Research Report 105, Colombo, Sri Lanka.
Wetlands International, 2005. Ramsar Sites: Directory and Overview, eighth
ed. Wetlands International.
Wood, A., 2006. Environmental role of wetlands in headwaters. NATO
Science Series IV 63, 211e220.