Identification of Continental Wetlands Using Different Orbital Remote Sensors

Abstract

This paper evaluates the potential of false-color composite images, from 3 different remote sensing satellites, for the identification of continental wetlands. Landsat 8, Sentinel-2, and CBERS-4 scenes from three different Ramsar sites (i.e., sites designated to be of international importance) two sites located within the Mato-Grossense Pantanal, and one within the Sul-mato-grossense were used for analyses. For each site, images from both the dry and rainy seasons were analyzed using Near-Infrared (NIR), Shortwave Infrared (SWIR), and visible (VIS) bands. The results show that false-color composite images from both the Landsat 8 and the Sentinel-2 satellites, with both SWIR 2-NIR-BLUE and NIR-SWIR-RED spectral band combinations, allow the identification of wetlands.

Full text

1
Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
DOI: 10.5212/TerraPlural.v.15.2115518.001
Identication of Continental Wetlands Using Different Orbital
Remote Sensors
Identicação de zonas úmidas continentais usando diferentes
sensores orbitais remotos
Identicación de humedales continentales utilizando diferentes
sensores orbitales remotos
Isadora Taborda Silva
https://orcid.org/0000-0001-9053-4329
isah.taborda@gmail.com
Universidade Federal de Mato Grosso do Sul,
UFMS, Campo Grande, MS
Jéssica Rabito Chaves
https://orcid.org/0000-0003-2763-3361
je.rabitochaves@gmail.com
Universidade Federal de Mato Grosso do Sul,
UFMS, Campo Grande, MS
Helen Rezende de Figueiredo
https://orcid.org/0000-0002-6580-8305
helenrezende.bio@gmail.com
Universidade Federal de Mato Grosso do Sul,
UFMS, Campo Grande, MS
Bruno Silva Ferreira
https://orcid.org/0000-0002-0697-2834
bsferreira@outlook.com
Universidade Federal de Mato Grosso do Sul,
UFMS, Campo Grande, MS
César Claudio Cáceres Encina
https://orcid.org/0000-0001-8061-9804
ccaceres.encina@gmail.com
Universidade Federal de Mato Grosso do Sul,
UFMS, Campo Grande, MS
Dhonatan Diego Pessi
https://orcid.org/0000-0003-0781-785X
dhonatan.pessi@gmail.com
Universidade Federal de Rondonópolis, UFR,
Rondonópolis, MT
Normandes Matos da Silva
https://orcid.org/0000-0002-4631-9725
normandes32@gmail.com
Universidade Federal de Rondonópolis, UFR,
Rondonópolis, MT
Eliane Guaraldo
https://orcid.org/0000-0003-2526-1293
eliane.guaraldo@gmail.com
Universidade Federal de Mato Grosso do Sul,
UFMS, Campo Grande, MS
Antonio Conceição Paranhos Filho
https://orcid.org/0000-0002-9838-5337
toni.paranhos@gmail.com
Universidade Federal de Mato Grosso do Sul,
UFMS, Campo Grande, MS

2Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
Isadora Taborda sIlva; eT al.
Abstract: This paper evaluates the potential of false-color composite images, from
3 different remote sensing satellites, for the identication of continental wetlands.
Landsat 8, Sentinel-2, and CBERS-4 scenes from three different Ramsar sites (i.e.,
sites designated to be of international importance) two sites located within the Mato-
Grossense Pantanal, and one within the Sul-mato-grossense were used for analyses. For
each site, images from both the dry and rainy seasons were analyzed using Near-
Infrared (NIR), Shortwave Infrared (SWIR), and visible (VIS) bands. The results show
that false-color composite images from both the Landsat 8 and the Sentinel-2 satellites,
with both SWIR 2-NIR-BLUE and NIR-SWIR-RED spectral band combinations, allow
the identication of wetlands.
Keywords: False-color composite imagery, multispectral sensor, Pantanal, photo
interpretation, Ramsar sites.
Resumo: Este artigo avalia o potencial de imagens compostas de cores falsas, de 3
satélites diferentes de sensoriamento remoto, para a identicação de zonas úmidas
continentais. Foram utilizadas para análise as cenas Landsat 8, Sentinel-2 e CBERS-4
de três locais diferentes de Ramsar (isto é, locais designados como de importância
internacional) dois locais localizados no Pantanal Mato-Grossense e um no Pantanal
Sul-mato-grossense para análise. Para cada local, as imagens das estações seca e chuvosa
foram analisadas usando bandas de infravermelho próximo (NIR), infravermelho de
ondas curtas (SWIR) e visível (VIS). Os resultados mostram que imagens compostas de
cores falsas dos satélites Landsat 8 e Sentinel-2, com combinações de bandas espectrais
SWIR 2-NIR-BLUE e NIR-SWIR-RED, permitem a identicação de áreas úmidas.
Palavras-chave: Imagens compostas em cores falsas, sensor multiespectral, Pantanal,
interpretação de fotos, sites Ramsar.
Resumen: Este artículo evalúa el potencial de imágenes compuestas de colores
falsos, de 3 satélites de teledetección diferentes, para la identicación de humedales
continentales. Para el análisis se utilizaron escenas Landsat 8, Sentinel-2 y CBERS-4
de tres ubicaciones diferentes en Ramsar (es decir, ubicaciones designadas como de
importancia internacional), dos ubicaciones ubicadas en el Pantanal Mato-Grossense y
una en el Pantanal Sul-Mato-grossense para análisis. Para cada ubicación, las imágenes
de las estaciones seca y lluviosa se analizaron utilizando infrarrojo cercano (NIR),
infrarrojo de onda corta (SWIR) y bandas visibles (VIS). Los resultados muestran que
las imágenes compuestas de colores falsos de los satélites Landsat 8 y Sentinel-2, con
combinaciones de bandas espectrales SWIR 2-NIR-BLUE y NIR-SWIR-RED, permiten
la identicación de áreas húmedas.
Palabras-clave: Imágenes compuestas en colores falsos, sensor multiespectral,
Pantanal, interpretación fotográca, sitios Ramsar
INTRODUCTION
Loss of wetlands is extensive with > 70% estimated to have been destroyed or
compromised by human action since the early 20
th
century (Ozesmi & Bauer, 2002; Nicholls,
2004; Gardner et al., 2015). Located in the Upper Paraguay Basin, Brazil, the Pantanal is the
largest tropical wetland in the world, covering approximately 140,000 km² (Adámoli, 1986).
Its drought and ood cycles make the Pantanal a dynamic ecosystem. Due to the
low incidence of precipitation, the waters found there originate from the surrounding river
basins, with varying intensities, as well as the specic compositions of landscape units.

3
Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
IdentIfIcatIon of contInental Wetlands UsIng dIfferent orbItal remote sensors
Thus, its wetlands or oodplains are highly diverse and occupy transition zones between
higher and well-drained environments and environments that remain constantly ooded
(Owen & Chiras, 1998).
Despite its importance, the region has been suffering from anthropic actions, causing
signicant loss of native vegetation and the consequent increase in pasture areas (Paranhos
Filho, Moreira, Oliveira, Pagotto & Mioto, 2014; Peres, Mioto, Marcato Junior & Paranhos
Filho, 2016). Understanding the nature of this wetland ecosystem is critical to maintaining
its balance and mitigating potential anthropogenic threats. To improve wetland protection,
understanding the dynamics of oods and identifying effective methods for describing
and monitoring them is critical.
However, wetland areas are difcult to monitor due to their relative inaccessibility
and seasonally dynamic nature (Hewes, 1951; Lee & Lunetta, 1996; Mitch & Gosselink,
2007). Remote sensing is an efcient and practical method that can be used to identify
landscape distribution of wetlands over a large area with advantages including multispectral
and multitemporal data collection (Ozesmi & Bauer, 2002; Rundquist, Gitelson, Derry,
Ramirez, Stark, & Keydan, 2001). Furthermore, makes it possible to reliably analyze and
monitor the Pantanal, which is vital for the preservation of the area (Santos, Pereira,
Shimabukuro & Rudorff, 2009; Paranhos Filho, Moreira, Oliveira, Pagotto & Mioto, 2014;
Guo, Li, Sheng, Xu & Wu, 2017). Multispectral sensors are particularly useful because their
multiple spectral bands allow many different band combinations for false-color composite
images, which allow visualization of wavelengths beyond the visible spectrum. Different
false-color composite images can be used to highlight different features and thus help to
identify different targets.
Different sensors have already been used to identify the spatial distribution of different
land cover and vegetation characteristics of the many habitats present in the Pantanal
complex system. Evans and Costa (2013) use in their study multi-temporal L-band ALOS/
PALSAR, C-band RADARSAT-2, and ENVISAT/ASAR data to map ecosystems and create
a lake distribution map of the Lower Nhecolândia subregion in the Brazilian Pantanal,
achieving a satisfactory result showing the spatial distribution of aquatic, terrestrial and
transitional habitats.
The most suitable band combination and resulting false-color composite image varies
for different targets, making it necessary to analyze which false-color composite provides
the best target distinction of the studied object. Besides, since understanding the different
spectral resolutions is the main factor inuencing the recognition of different types of
ground cover (Paranhos Filho, Moreira, Oliveira, Pagotto & Mioto, 2014), it is important
to identify the spectral behavior of each band to be used, as well as to understand the
capabilities and limitations of each sensor.
This study aims to analyze the potential for individualization of the continental
wetlands and compares different selected orbital sensors. It’s intended to establish possible
spectral response patterns.

4Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
Isadora Taborda sIlva; eT al.
MATERIALS AND METHODS
Study area
Located in central South America and mostly covering Brazilian territory, the
Pantanal is characterized by a tropical climate, with a monomodal ood pulse that varies
in intensity from year to year, and well-dened dry and rainy seasons (Adámoli, 1986).
It consists of an extensive sediment accumulation plain that, due to its low slope relief,
is annually ooded by the Paraguay River and its tributaries during the rainy season
(Adámoli, 1986; Penatti, 2014).
While the PantanaI complex can be divided into multiple sub-regions with very
different characteristics, it can generally be divided into three groups based on water
availability: approximately 10% to 20% is permanently covered by water or ooded for
long periods, 35% of the complex is made up of fully dry areas, and approximately 50%
is made up of aquatic-terrestrial transition areas, comprising permanently aquatic and
permanently terrestrial habitats (lightly ooded or ooded for about 3 to 6 months) (Junk
& Wantzen, 1989).
It is composed of a mosaic of habitats, including the Cerrado, the Chaco, some
components of the Caatinga, and ecosystems of the periamazonic region with a spatial
habitat heterogeneity characteristic of this type of formation (Ab’saber, 1988; Guimarães,
Trevelin & Manoel, 2014). The Pantanal can be considered a large environmental transition
zone and, like other ecotones, is characterized by high biodiversity and is among the most
productive areas in the world (Calheiros & Ferreira, 1997).
Given its environmental and cultural relevance and unique characteristics, it is
recognized as a National Heritage Site by the Federal Constitution and considered a
Biosphere Reserve and a World Heritage Site by UNESCO (United Nations Educational,
Scientic and Cultural Organization). Also, the region is home to three sites recognized
by the Convention on Wetlands of International Importance (Ramsar), namely the Private
Reserve of the Fazenda Rio Negro Natural Heritage (RPPN/FRN, for its acronyms in
Portuguese), the Pantanal Mato-Grossense National Park (PARNA, for its acronyms in
Portuguese) and the Private Reserve of the SESC Pantanal Natural Heritage (RPPN/SESC,
for its acronyms in Portuguese) (Fig. 1).

5
Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
IdentIfIcatIon of contInental Wetlands UsIng dIfferent orbItal remote sensors
Figure 1: Pantanal location map highlighting the region’s three Ramsar sites: the Pantanal Mato-Grossense
National Park (135.000 ha), the RPPN Fazenda Rio Negro (38.000 ha), and the RPPN SESC Pantanal
(106.000 ha). Source: Mioto, Albrez & Paranhos Filho (2012 (Pantanal Limits), ICMBio 2018a (Limit of
Conservation Units), and MMA 2019 (Limit of Latin America and the Federative States).
Located in the municipality of Barão de Melgaço (MT), RPPN/SESC is a privately
owned nature reserve managed by the Social Service of Commerce (SESC, for its acronyms
in Portuguese) who is responsible for delivering environmental education and low impact
ecotourism activities, under the supervision of the Brazilian Institute of Environment and
Renewable Natural Resources (IBAMA, for its acronyms in Portuguese) (Alho et al., 2011).
Positioned in the so-called Poconé Pantanal, much of its territory is predominantly ooded
during the rainy season (ICMBio, 2018b).
The PARNA is located in the Upper Pantanal, in the southwest of Mato Grosso,
at the conuence of the Paraguay and Cuiabá rivers (ICMBio, 2003). As a protection
mechanism, in addition to this Ramsar site, there are four conservation units considered
as natural world heritage sites by UNESCO (2000), including the National Park under the
management of the Chico Mendes Institute for Biodiversity Conservation (ICMBio, for its
acronyms in Portuguese) and three RPPNs: Dorochê (northern border of the park, plain
region, and predominance of Cerrado), Acurizal and Penha (to the south, located in the
state of Mato Grosso do Sul, Serra do Amolar region).

6Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
Isadora Taborda sIlva; eT al.
The RPPN/FRN is located in an area delimited by the Negro River from east to west,
in the Abobral and Nhecolândia sub-regions. Its relief is slightly undulating and contributes
to the formation of a diverse range of landscapes (ICMBio, 2018c). Another characteristic
feature is the shallow lake basins that range from slightly acidic to highly alkaline, known
as bays and salines, respectively (Machado, Silva, Pinto, Camargo & Ribeiro, 2009).
Because these areas are Ramsar sites, it is easier to obtain support and access to
international funds for project nancing, in exchange for maintaining the ecological
processes and conserving the local biodiversity. Due to their high potential to present a
native vegetation cover, they were selected for the spectral analysis.
Data Processing
In this study, scenes obtained in 2017 and 2018 from the medium resolution satellites
Landsat 8 ‘Earth Explorer 2018a,b,c,d,e,f’, Sentinel-2 ‘Earth Explorer 2018g,h,e,i,k,l’ and
CBERS-4 ‘INPE 2018a,b,c,d,e,f’, were used for analyses.
Six Landsat 8 Operational Land Imager (OLI) images with a 30 m spatial resolution
referring to the orbits/points 227/72, 226/72, and 226/74, twelve CBERS-4 images, six
with the MUX sensor (multispectral camera), and six with the IRS sensor (multispectral
and thermal imaging), with spatial resolutions of 20 m and 40 m, respectively, referring
to orbits/points 165/122, 165/119 and 167/120; and six Sentinel-2 images, with an MSI
(Multispectral Instrument) sensor, with spatial resolutions of 10 m and 20 m, referring to
orbits/points 21KWU, 21KVA, and 21KWB were used.
According to the geographical position of each analyzed site, the dry and rainy
seasons were identied, and one scene from each season was selected per satellite from
those with the highest image quality and the lowest cloud coverage (Table 1). Initially,
scenes from the same year (2017 or 2018) were selected; however, it was not possible to
obtain CBERS-4 and Sentinel 2 scenes with the above characteristics in the same year and,
therefore, images from both years were selected here.
Table 1: Data from the scenes used in this study being one scene from each season of each analyzed site
per satellite.
SATELITE ORBIT / OINT SENSOR RAINY DRY
LANDSAT 8
227/72 OLI 29/04/2017 16/09/2017
226/72 OLI 01/02/2017 13/09/2017
226/74 OLI 21/03/2017 28/08/2017
CBERS-4
165/122 MUX 15/04/2017 23/08/2017
165/122 IRS 15/04/2017 23/08/2017
166/119 MUX 08/05/2017 15/09/2017
166/119 IRS 08/05/2017 15/09/2017
167/120 MUX 08/04/2018 12/09/2017
167/120 IRS 08/04/2018 12/09/2017
SENTINEL-2
21KVA MSI 27/04/2018 26/07/2017
21KWU MSI 22/04/2018 10/08/2018
21KWB MSI 07/01/2017 04/09/2017

7
Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
IdentIfIcatIon of contInental Wetlands UsIng dIfferent orbItal remote sensors
The QGIS 2.18 (QGIS Development Team, 2017) software, licensed under the General
Public License (GNU), was used for all operations.
Once the images were gathered, the imaged bands of the three satellites were
compared in order to dene the corresponding bands to be used for each sensor to create
the false-color composite images (Fig. 2).
The IRS sensor bands (CBERS-4) were resampled from 40 m to 20 m to be stacked
with the MUX bands (CBERS-4). The MSI (Sentinel-2) sensor bands were converted to
GeoTIFF format for uniformity, and the 20 m bands resampled to 10 m and then stacked.
The Landsat 8 images did not require resampling or GeoTIFF conversion, only stacking.
Subsequently, all stacks were redesigned for the Universal Transverse Mercator
(UTM), Zone 21 South, and Datum SIRGAS 2000 projection. The images were then cut
using the clip tool, remove all but the analyzed regions. To this end, we used the shapeles
available on the ICMBio website (2018).
Figure 2: Comparative analysis of Landsat 8, CBERS-4, and Sentinel-2 imaged bands. Bands 10 and 11 of
IRS (CBERS-4), and 11, and 12 of MSI (Sentinel-2) were resampled (shaded boxes). Source: USGS (2018),
INPE (2018), and ESA (2018).
Three different combinations of three spectral bands for false-color composite
image generation and their corresponding bands for each sensor were selected for photo-
interpretive analysis of the Pantanal sub-regions as follows:
• Near-Infrared (NIR), Shortwave Infrared (SWIR), and red corresponding to the
5 - 6 - 4 bands in Landsat 8 (OLI sensor); 8 - 10 - 7 in CBERS-4M (MUX and IRS
sensors), and 8 - 11 - 4 in Sentinel-2M (MSI sensor), respectively;

8Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
Isadora Taborda sIlva; eT al.
• Red, NIR, and green corresponding to the 4 - 5 - 3 bands in Landsat 8 (OLI), 7 - 8 -
6 in CBERS-4M (MUX) and 4 - 8 - 3 in Sentinel-2M (MSI), respectively;
• Shortwave Infrared 2 (SWIR 2), NIR, and blue corresponding to the 7 - 5 - 2 bands
in Landsat 8 (OLI), 11 - 8 - 5 in CBERS-4M (MUX and IRS sensors) and 12 - 8 - 2 in
Sentinel-2M (MSI), respectively.
The images were subsequently enhanced to achieve better visualization with the naked
eye, aiding to identication of ground cover patterns. To improve our understanding of
the nature of the dynamic Pantanal ecosystem, permanently dry and permanently ooded
areas were analyzed. The classes used in the ground cover identication were: dry area
with undergrowth; dry area with tree-shrub vegetation; humid undergrowth area; ood
area with tree-shrub vegetation; wetland; free water surface; sediment-water; water. The
Morro Caracará (MT), in the Pantanal Mato-Grossense National Park, was considered the
most signicant among the sites. After analyzing all images from both the dry and rainy
seasons from all three study sites, it was decided that only the Pantanal Mato-Grossense
National Park will be discussed here.
The three studied sites show a high degree of correlation among the types of vegetation
present; however, the Morro do Caracará site has a greater potential of having a native
vegetation cover. As stated by the National System of Conservation Units (SNUC), the
objective of this National Park is the preservation of natural ecosystems of great ecological
relevance and scenic beauty.
According to the management plan, the Morro do Caracará is characterized by
pediplains of varying sizes surrounded by ooded forests. In the rainy season, the humid
soil has a thick layer of peat. Seasonal deciduous forest covers much of the Morro do
Caracará, but some areas are forest-savannah transition or semi-deciduous seasonal forest
biomes. It presents ‘differences in the herbaceous species cover between seasons; this being
smaller during the drought and with more presence of peat, resulting from the leaf fall of
the trees. (ICMBio, 2003, p. 70). On gradients with slopes of 30% and 45%, the herbaceous
vegetation is predominant, identied as rupestrian eld vegetation. Interspersed with this
herbaceous vegetation, unsteady shrubs and cacti occur.
RESULTS
Correlation between satellites
The rst false-color composite image used the NIR-SWIR-RED bands (5-6-4 in Landsat
8, 8-10-7 in CBERS-4M, and 8-11-4 in Sentinel-2M) (Figs. 3, 4, and 5). With high reectance
in the NIR band, the photosynthetically active tree-shrub vegetation corresponds to a
reddish color. In the CBERS-4
M
scenes, the contrast between these vegetation areas and
other areas is greater, and thus presents a more vibrant coloration. Wetlands or ooded
areas, with or without the presence of sediment or macrophytes, are indicated by a navy-
blue to black color in all sensors. This response occurs because the visible band (red),

9
Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
IdentIfIcatIon of contInental Wetlands UsIng dIfferent orbItal remote sensors
which presents high water reectance, is assigned to the blue channel. Dry areas with
undergrowth correspond to water-green tones.
In the composite image of the PARNA (Fig. 3), areas with humid tree-shrub vegetation
(orange) were identied. In the eastern portion of the park this vegetation is intertwined
with large lagoons (blue), and in the western portion with wetlands (shades of dark blue).
In the RPPN/FRN, the presence of lagoons (blue coloration) in the western portion
of the site is highlighted, which during the drought present areas of undergrowth in its
surroundings (Fig. 4).
The RPPN/SESC scenes (Fig. 5) feature more areas with undergrowth without a
free water surface (water-green). During the dry season, the tree-shrub vegetation in the
eastern portion of the region also has a reddish tone, indicating that it is photosynthetically
active. During the rainy period, these regions moisten and turn orange. Few wetlands are
found, appearing mainly during the oods in the northeast region.
Figure 3: False-color composite images of the Pantanal Mato-Grossense National Park for the scenes of the
three satellites using a NIR-SWIR-RED band combination.

10 Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
Isadora Taborda sIlva; eT al.
Figure 4: False-color composite images of the SESC Pantanal Natural Heritage for the scenes of the three
satellites using a NIR-SWIR-RED band combination.

11
Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
IdentIfIcatIon of contInental Wetlands UsIng dIfferent orbItal remote sensors
Figure 5: False-color composite images of the Fazenda Rio Negro Natural Heritage for the scenes of the
three satellites using a NIR-SWIR-RED band combination.

12 Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
Isadora Taborda sIlva; eT al.
The second spectral band combination, RED-NIR-GREEN (4-5-3 in Landsat 8, 7-8-
6 in CBERS-4
M
, and 4-8-3 in Sentinel-2
M
), produced false-color composite images with
predominantly pink and green coloration (Figs. 6, 7, and 8). Associated with ooded areas,
shades of pink to lilac are visual responses of the largest water reections in the visible
region (red and green), assigned to the red and blue channels of the multispectral image.
The darker shades of the same colors indicate areas with cleaner waters.
Figure 6: False-color composite images of the Pantanal Mato-Grossense National Park for the scenes of the
three satellites using a RED-NIR-GREEN band combination.

13
Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
IdentIfIcatIon of contInental Wetlands UsIng dIfferent orbItal remote sensors
Figure 7: False-color composite images of the Fazenda Rio Negro Natural Heritage for the scenes of the
three satellites using a RED-NIR-GREEN band combination.
In this type of false-color composite image, the wetlands are confused with the
ooded, clean water areas in the Landsat 8 and CBERS-4M scenes, while in the Sentinel-
2M scenes cleaner water surfaces correspond to black, facilitating their identication from
other areas.
In the RPPN SESC composition, wetlands are confused with areas of undergrowth.

14 Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
Isadora Taborda sIlva; eT al.
Figure 8: False-color composite images of the SESC Pantanal Natural Heritage for the scenes of the three
satellites using a RED-NIR-GREEN band combination.

15
Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
IdentIfIcatIon of contInental Wetlands UsIng dIfferent orbItal remote sensors
The last false-color composite image, using a combination of SWIR 2-NIR-BLUE
spectral bands (7-5-2 in Landsat 8, 11-8-5 in CBERS-4
M
, and 12-8-2 in Sentinel-2
M
), presents
more differentiated coloration between the three satellites (Figures 9, 10, and 11). The
ooded areas in the Landsat 8 scenes are brown, while they correspond to a navy-blue
tone in the CBERS-4M scenes, and purple in the Sentinel-2M scenes. This response occurs
due to differences in the imaged range of each sensor.
Since the spectral response of water is more intense in the visible range, while that
of the exposed soils is more intense in the SWIR 2 range, the false-color composite images
with these ranges resulted in equally higher reectance values in the red and blue channels,
respectively. The differences in the spectral ranges imaged by the sensors, especially for
the SWIR 2 band, cause these slightly distinct visual responses from purple to brown. Due
to the reectance of the target in the NIR region, and since they correspond to the green
channel of the composite image, the areas with vegetation show shades of green.
Figure 9: False-color composite images of the Pantanal Mato-Grossense National Park for the scenes of the
three satellites using a SWIR 2-NIR-BLUE band combination.

16 Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
Isadora Taborda sIlva; eT al.
Figure 10: False-color composite images of the Fazenda Rio Negro Natural Heritage for the scenes of the
three satellites using a SWIR 2-NIR-BLUE band combination.

17
Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
IdentIfIcatIon of contInental Wetlands UsIng dIfferent orbItal remote sensors
Figure 11: False-color composite images of the SESC Pantanal Natural Heritage for the scenes of the three
satellites using a SWIR 2-NIR-BLUE band combination.

18 Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
Isadora Taborda sIlva; eT al.
Correlation between the different false-color composite images of the same satellite
To facilitate the interpretation of the different results, the composite images were
separated by the sensor, and the specic color gradient legends were compiled for each
image, allowing detailed evaluation of the ground cover identication and the specic
behavior of each sensor.
At the top of Morro Caracará of PARNA, the vegetation is undergrowth (without
free water surfaces), on the slope tree-shrubs. In the Landsat 8 NIR-SWIR-RED false-color
composite image, the water-green tone corresponds to the areas of dry undergrowth, while
the yellow to orange areas correspond to tree-shrub vegetation with higher photosynthetic
activity. The black and navy-blue areas correspond to clean water (Fig. 12).
Figure 12: False-color composite images of the Landsat 8 scenes of the Pantanal Mato-Grossense National
Park in the Morro Caracará area.
In the RED-NIR-GREEN false-color composite image, the difference in the color of
the areas of undergrowth and areas with free water surfaces is very subtle, corresponding
to light pink and pink tones, respectively. In the NIR-SWIR-RED and the SWIR 2-NIR-BLUE
false-color composite images, the contrast between the two areas is clear, corresponding
to the color scheme displayed in the gradation legend.

19
Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
IdentIfIcatIon of contInental Wetlands UsIng dIfferent orbItal remote sensors
In the SWIR 2-NIR-BLUE false-color composite image, the paths of the water bodies
are evident and, as in the NIR-SWIR-RED composite image each component has a different
color. However, the wetlands and the ood areas with tree-shrub vegetation are easy to
recognize around the lagoon in the SWIR 2-NIR-BLUE image (opaque moss green) but
only subtly different in the NIR-SWIR-RED (brown) and RED-NIR-GREEN (dark green)
images, creating in the latter two an illusion of a ooded area with undergrowth in the
northeastern portion of the analyzed area.
The CBERS-4
M
images show prevalent cloud coverage throughout the analyzed
period, resulting in false-color composite images that are difcult to interpret (Fig. 13).
Figure 13: False-color composite images of the CBERS-4M scenes from the Pantanal Mato-Grossense
National Park in the Monte Caracará area.
The combination of the MUX sensor (with the blue, green, red, and NIR bands)
and the IRS sensors (SWIR and SWIR 2) was not realizable without geometric correction,
resulting in spatially distorted images, that made comprehension difcult, and a gradation
legend with very little contrast. The RED-NIR-GREEN false-color composite image, a
product of bands from the MUX sensor only, gave a slightly clearer image of the terrain.
In the RED-NIR-GREEN false-color composite image, the dry undergrowth at the
top of Morro Caracará corresponds to a light pink color, similar to that in the Landsat 8

20 Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
Isadora Taborda sIlva; eT al.
images (Figure 12), resulting in a very similar color scheme for the wetlands and the ood
areas. Areas with dry tree-shrub vegetation (green), however, became evident, presenting
a different color from the other areas.
In general, the Sentinel-2
false-color composite images exhibit similar trends to those
of the Landsat 8 images, with high contrast for dry areas in the NIR-SWIR-RED and SWIR
2-NIR-BLUE composite images compared to wetlands and ood areas (Fig. 14).
Figure 14: False-color composite images of Sentinel-2 scenes from the Pantanal Mato-Grossense National
Park in the Monte Caracará area.
The difference between dry and ooded areas in the RED-NIR-GREEN false-color
composite image is small, with similar shades of color (pink, light pink, and lilac) for
regions of different vegetation types. The water body path visibility is subtle in the NIR-
SWIR-RED false-color composite image when compared to the RED-NIR-GREEN and
SWIR 2-NIR-BLUE composite images. The spectral response to water is shown in black
in all images, but the macrophyte and sediment visualization are lighter in the latter two,
corresponding to shades of pink to black in the RED-NIR-GREEN and blue to black in the
SWIR 2-NIR-BLUE false-color composite images.
For a better understanding of which sensors and band combinations obtained the
better results in the identication of wetlands and ood areas (Tab. 2).

21
Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
IdentIfIcatIon of contInental Wetlands UsIng dIfferent orbItal remote sensors
Table 2: Visual identication of wetlands through orbital images. Legend: A = Appropriate; N = Not
appropriate. NSR = NIR-SWIR-RED; RNG = RED-NIR-GREEN; S2NB=SWIR 2-NIR-BLUE.
Satellite/Sensor Landsat 8 Sentinel-2 CBERS-4
Composition NSR RNG S2NB NSR RNG S2NB NSR S2NB RNG
Seasonality AAAAAANNN
Details of the ooded landscape
(Wetland and Flood Areas) NNAANANNN
DISCUSSION
The visualization of ground cover in false-color composite images evaluated for three
sites in this study revealed that the OLI (Landsat) and MSI (Sentinel-2) sensor composite
images have similar photo interpretive ground cover results. Due to their higher spatial
resolution, the Sentinel-2
M
images were more suitable for small area analysis or large-
scale mapping. The Landsat 8 scenes, on the other hand, are more suited for the analysis
of wider areas.
The resampling proved to be adequate for the MSI (Sentinel-2) sensor, possibly due
to the different bands and spatial resolutions being imaged by the same sensor. All the
composite images obtained for scenes from this sensor allow clear identication of the areas
analyzed. On the other hand, the composite images generated from CBERS-4
M
multi-sensor
scenes proved difcult to interpret, especially regarding small areas. It is recommended
to further study the multi-sensor false-color composite images of this satellite, especially
with geometric and atmospheric corrections.
The gradation legends (from dry to rainy) allowed comparison among the different
false-color composite images of the same satellite. Among the Landsat 8 composite images,
the 7-5-2 (SWIR 2-NIR-BLUE) image shows the best results about the color distinction,
ease of wetlands visualization, and differentiation of water-generated paths. For the
Sentinel-2M scenes, the same spectral band combination exhibits similar behavior, while
the image generated by the 12-8-2 (SWIR 2-NIR-BLUE) band combination shows higher
color contrast than the NIR-SWIR-RED and RED-NIR-GREEN composite images. Seasonal
variability is evidenced in images from both the Landsat 8 and the Sentinel-2M satellites,
both in the SWIR 2-NIR-BLUE and NIR-SWIR-RED composite images, demonstrating the
high potential use of both for ground cover identication in wetland studies.
It is important to highlight that the perusal of the management plan provided valuable
insights for the comparative analysis of the images within the local context, serving as a
reference for the photo interpretation phase.
FINAL REMARKS
This study was motivated by the popularity of wetland classication using remote
sensing data and should support the advancement of wetland classication methods. The

22 Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
Isadora Taborda sIlva; eT al.
scientic community needs to assess historical methodological trends before deciding which
approaches to take in the future. This article facilitates that analysis. However, the impact
of classication on conservation, restoration, and other important social enterprises can
also be extended by considering the application requirements and logistical limitations
of operational mapping programs when developing classication protocols. For example,
some orbital images may not be widely available, and some protocols may be impractical
to implement on a regional scale, but only on a national or continental scale. Besides the
methods may work well in one place of study, but not in another. This broader perspective
can also help guide the development of classication protocols.
Concerning the current rates of loss of wetlands and the importance of monitoring
these areas, this study contributed by revealing the potential of the analyzed sensors and the
high spatial and temporal resolution available free of charge for detailed characterizations
of wetlands at various levels. Given the limitations of this study, to improve the accuracy of
wetland characterization and to better characterize wetland dynamics, future studies may
involve more sophisticated methods for analyzing Sentinel-1 time series and additional
indices or measurements of Sentinel-2 bands and Landsat-8. Also, on-site measurements
for data collection data will largely contribute to the accuracy assessments of this study.
ACKNOWLEDGMENTS
This study was nanced in part by the Coordenação de Aperfeiçoamento de Pessoal
de Nível Superior - Brasil (CAPES) - Finance Code 001. We thank CAPES for the Master’s
Scholarships of Chaves, Ferreira and Silva, the PNPD CAPES scholarship of Guaraldo and
the doctoral scholarship of Encina. We also thank the National Council for Scientic and
Technological Development (CNPq), for the Research Productivity Scholarship (Process
304122/2015-7) of Paranhos Filho.
To CNPq for the award of a Productivity Subsidy in technological development
and innovative extension granted to Normandes Matos da Silva (Process 315170/2018-
2), and the nancial support for the project of a remote piloted aircraft as a strategy to
monitor res and monitor the restoration ecological protection in protected areas. (Subject
441975/2018-6);
REFERENCES
Ab’saber A. N. (1988). O Pantanal Mato-Grossense e a teoria dos refúgios. Rev Bras Geog, 50(2), 9-57.
Adámoli, J. (1986). Fitogeograa do Pantanal. Anais do Simpósio sobre recursos naturais e sócio-econômicos do
Pantanal. Corumbá, MS, Brasil,1. pp.90-106.
Alho, C.J.R., Schneider, M., Conceição, P.N., Vasconcellos, L.A.S., Oliveira, D.M. M. , Strussmann, C., &
Morais, R.M.G. (2011). Plano de manejo da estância ecológica SESC Pantanal. Conhecendo o Pantanal, 1, 5-7.
Calheiros, D.F., & Ferreira, C.J. (1997). Alterações limnlógicas no rio Paraguai “dequada” e o fenômeno natural de
mortalidade de peixes no Pantanal matogrossense. Corumbá: Embrapa-CPAP.

23
Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
IdentIfIcatIon of contInental Wetlands UsIng dIfferent orbItal remote sensors
Earth Explorer. (2018a). United States Geological Survey. Landsat 8 Images, sensor OLI. Orbit 226, point 072.
Acquisition Date: February 1, 2017. Accessed on June 5, 2018.
Earth Explorer. (2018b). United States Geological Survey. Landsat 8 Images, sensor OLI. Orbit 226, point 072.
Acquisition Date: September 13, 2017. Accessed on June 5, 2018.
Earth Explorer. (2018c). United States Geological Survey. Landsat 8 Images, sensor OLI. Orbit 226, point 074.
Acquisition Date: March 21, 2017. Accessed on June 5, 2018.
Earth Explorer. (2018d). United States Geological Survey. Landsat 8 Images, sensor OLI. Orbit 226, point 074.
Acquisition Date: August 28, 2017. Accessed on June 5, 2018.
Earth Explorer. (2018e). United States Geological Survey. Landsat 8 Images, sensor OLI. Orbit 227, point 072.
Acquisition Date: September 16, 2017. Accessed on June 5, 2018.
Earth Explorer. (2018f). United States Geological Survey. Landsat 8 Images, sensor OLI. Orbit 227, point 072.
Acquisition Date: April 29, 2017. Accessed on June 5, 2018.
Earth Explorer. (2018g). United States Geological Survey. Sentinel 2A Images, sensor MSI. Orbit/point 21KVA.
Acquisition Date: August 26, 2017. Accessed on June 5, 2018.
Earth Explorer. (2018h). United States Geological Survey. Sentinel 2A Images, sensor MSI. Orbit/point 21KVA.
Acquisition Date: April 27, 2017. Accessed on June 5, 2018.
Earth Explorer. (2018i). United States Geological Survey. Sentinel 2A Images, sensor MSI. Orbit/point 21KWB.
Acquisition Date: September 4, 2017. Accessed on June 5, 2018.
Earth Explorer. (2018j). United States Geological Survey. Sentinel 2A Images, sensor MSI. Orbit/point 21KWB.
Acquisition Date: January 7, 2017. Accessed on June 5, 2018.
Earth Explorer. (2018k). United States Geological Survey. Sentinel 2A Images, sensor MSI. Orbit/point 21KWU.
Acquisition Date: April 22, 2018. Accessed on June 5, 2018.
Earth Explorer. (2018l). United States Geological Survey. Sentinel 2A Images, sensor MSI. Orbit/point 21KWU.
Acquisition Date: August 26, 2017. Accessed on June 5, 2018.
Evans, T.L., & Costa, M. (2013). Landcover classication of the Lower Nhecolândia subregion of the
Brazilian Pantanal Wetlands using ALOS/PALSAR, RADARSAT-2, and ENVISAT/ASAR imagery.
Remote Sensing of Environment, 128, 118-137. 10.1016/j.rse.2012.09.022
Gardner, R.C., Barchiesi, S., Beltrame, C., Finlayson, C.M., Galewski, T., Harrison, I., … & Walpole, M.
(2015). State of the World’s Wetlands and their Services to People: A compilation of recent analyses. (Ramsar
Brieng,.7). Gland: Switzerland: Ramsar Convention Secretariat.
GIS Development Team. (2017). (Version 2.18.23 Las Palmas). [Software]. QGIS Geographic Information
System.
Guimarães, E., Trevelin, C.C., & Manoel, P.S. (2014). Pantanal: paisagens, ora e fauna. São Paulo: Cultura
Acadêmica/ Ed. UNESP.
Guo, M., Li, J., Sheng, C., Xu, J., & Wu, L.A. (2017). review of wetland remote sensing. Sensors, 17(4), 777.
10.3390/s17040777
Hewes, L. (1951). The northern wet prairie of the United States: Nature, sources of information, and
extent. Annals of the Association of American Geographers, 41(4), 307-323. 10.2307/2561071
ICMBIO. (2003). Plano de Manejo do Parque Nacional do Pantanal Matogrossense. Brasília: IBAMA. Recovered
from https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ve
d=2ahUKEwili7PWxrvrAhUlHrkGHQW3DUsQFjABegQIAxAB&url=https%3A%2F%2Fwww.icmbio.
gov.br%2Fportal%2Fimages%2Fstories%2Fimgs-unidades-coservacao%2Fparna_matogrossensee.
pdf&usg=AOvVaw0Rh9mWiQ5cC-ie8UdabmFb
ICMBIO. (2018a). Mapa Temático e Dados Geoestatísticos das Unidades de Conservação Federais. Shapele
Unidades de Conservação Federais.

24 Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
Isadora Taborda sIlva; eT al.
ICMBIO. (2018b). RPPN Estância Ecológica SESC Pantanal. Shapele. Accessed on June 12, 2018.
ICMBIO. (2018c). RPPN Rio negro. Shapele.
INPE. (2018a). Divisão de Geração de Imagens. Catálogo de Imagens. CBERS-4 Images, sensor MUX e IRS. Orbit
165 point 122. Acquisition Date: April 15, 2017. Accessed on June 22, 2018.
INPE. (2018b). Divisão de Geração de Imagens. Catálogo de Imagens. CBERS-4 Images, sensor MUX e IRS. Orbit
165 point 122. Acquisition Date: August 23, 2017. Accessed on June 22, 2018.
INPE. (2018c). Divisão de Geração de Imagens. Catálogo de Imagens. CBERS-4 Images, sensor MUX e IRS. Orbit
166 point 129. Acquisition Date: May 8, 2017. Accessed on June 22, 2018.
INPE. (2018d). Divisão de Geração de Imagens. Catálogo de Imagens. CBERS-4 Images, sensor MUX e IRS. Orbit
166 point 129. Acquisition Date: Setember 15, 2017. Accessed on June 22, 2018.
INPE. (2018e). Divisão de Geração de Imagens. Catálogo de Imagens. CBERS-4 Images, sensor MUX e IRS. Orbit
167 point 120. Acquisition Date: September 12, 2017. Accessed on June 22, 2018.
INPE. (2018f). Divisão de Geração de Imagens. Catálogo de Imagens. CBERS-4 Images, sensor MUX e IRS. Orbit
167 point 120. Acquisition Date: April 8, 2018. Accessed on June 22, 2018.
Junk, W.J., & Wantzen, K. M. (1989). The Flood Pulse Concept in River-Floodplain-Systems. Canadian
Special Publications for Fisheries and Aquatic Sciences, 106(1), 110-127.
Lee, K.H., & Lunetta, R.S. (1996). Wetland Detection Methods. In: J.G. Lyon, & J. McCarthy (Orgs.),
Wetland and Environmental Applications of GIS. (pp. 249-284). New York:.Lewis Publishers.
Machado, R.B., Silva, S.M., Pinto, E.C.T., Camargo, G., & Ribeiro, A.P. (2009). Reserva Particular do
Patrimônio Natural Fazenda Rio Negro - Plano de Manejo. Recovered from https://www.google.com/url?s
a=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwiP14Tqy7vrAhXmLLkGHdxaBSgQFjAAegQI
AxAB&url=https%3A%2F%2Fwww.imasul.ms.gov.br%2Fwp-content%2Fuploads%2F2015%2F06%2FP
lano-de-Manejo-RPPN-FRN-.pdf&usg=AOvVaw2nqKstruQM27YptY8CSop _
Mioto, C.L., Albrez, E.A., & Paranhos Filho, A.C. (2012). Contribuição à caracterização das sub-regiões
do Pantanal. Rev Entre-Lugar, 3(6), 165-180.
Mitch, W.J., & Gosselink, J.G. (2007). Wetlands. New Jersey: Wiley.
Nicholls, R.J. (2004). Coastal ooding and wetland loss in the 21st century: Changes under the SRES
climate and socio-economic scenarios. Glob. Environ. Chang., 14, 69-86. 10.1016/j.gloenvcha.2003.10.007
Owen, O.S., & Chiras, D.D. (1998). Natural Resource Conservation - Management for a Sustainable Future.
New Jersey: Prentice-Hall.
Ozesmi, S.L., & Bauer, M.E. (2002). Satellite remote sensing of wetlands. Wetlands ecology and management,
10(5), 381-402. 10.1023/A:1020908432489
Paranhos Filho, A.C., Moreira, E.S., Oliveira, A.K.M., Pagotto, T.C.S., & Mioto, C L. (2014). Análise da
variação da cobertura do solo no Pantanal de 2003 a 2010 através de sensoriamento remoto. Engenharia
Sanitária e Ambiental, 1(1), 69-76. 10.1590/S1413-41522014019010000305
Penatti, N.C. (2014). Geobotânica e fenologia da superfície terrestre no estudo da diversidade do Pantanal:
uma abordagem multisensor. Tese de doutorado. Universidade de São Paulo, São Paulo, SP, Brasil.
10.11606/T.44.2014.tde-06022015-084733
Peres, P.N., Mioto, C.L., Marcato Junior, J. & Paranhos Filho, A.C. (2016). Variação da Cobertura do
Solo no Pantanal de 2000 a 2015 por Sensoriamento Remoto com Software e Dados Gratuitos. Anuário do
Instituto de Geociências, 39(2), 116-123. 10.11137/2016_2_116_123
Rundquist, D., Gitelson, A.A., Derry, D., Ramirez, J., Stark, R, & Keydan, G. (2001). Remote estimation
of vegetation fraction in corn canopies. In G. Grenier, & S. Blackmore. Proceedings of the Third European
Conference on Precision Agriculture. (v 1, pp. 301-306).

25
Terr@Plural, Ponta Grossa, v.15, p. 1-25, e2115518, 2021.
IdentIfIcatIon of contInental Wetlands UsIng dIfferent orbItal remote sensors
Santos, J.S., Pereira, G., Shimabukuro, Y.E., & Rudorff, B. F. T. (2009). Identicação de áreas alagadas no
bioma Pantanal–Brasil utilizando dados multitemporais Terra/MODIS. Geograa, 34, 745-755.
Data de submissão: 23/maio/2020
Data de aceite: 01/set./2020