SciELO - Scientific Electronic Library Online

vol.68 número2  suppl.TIIProcFRONTS AND FRONTAL WAVES OVER THE SOUTHERN CONEANALYSIS OF SPACE-TIME VARIABILITY OF THE PLATA RIVER PLUME índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Servicios Personalizados




Links relacionados


Gayana (Concepción)

versión impresa ISSN 0717-652Xversión On-line ISSN 0717-6538

Gayana (Concepc.) v.68 n.2 supl.TIIProc Concepción  2004 

  Gayana 68(2): 476-481 , 2004



Joel Pellerin, Lucia Pinto Camargo & Clarice Maria Neves Panitz

Depto. De Geociências/Centro de Filosofia e Ciências Humanas, Depto. De Ecologia e Zoologia/Centro de Ciências Biológica. Universidade Federal de Santa Catarina. Campus Universitário, Florianópolis, Santa Catarina, Brazil. 88049-900,,


With the availability of multisensor, multitemporal, multiresolution and multifrequency image data from operational Earth observation satellites the fusion of digital image data has become a valuable tool in remote sensing image evaluation. In this paper a new multispectral image wavelength representation is introduced of the integrated Landsat Thematic Mapper TM and Spot HRV Pan data to establish river Tavares' Mangrove vegetation classification in Santa Catarina Island - Brazil. The combining of data from these different sensors is necessary for preserving spatial integrity of the higher-resolution data (PAN) set spectral information and of the lower resolution components in (TM). The fusion of these different data contributed to the understanding of the objects observed. Image fusion has many aspects to be looked at. First of all, the wavelength approach has been compared to hue-lightness-saturation (HIS) transformed image fusion technique and this has showed to possess the advantage of minimal distortion of the data spectral visible characteristic and the enhancement of spatial quality. It has exhibited the potential application of wavelength higher accuracy transformation for fusing spectral. In this study, mangrove's areas mapping were performed with an overall accuracy (Kappa) of 0.93 and this was possible because a good field knowledge.


Mangrove are transitional ecosystems between ocean and earth being associated with estuaries and coastal lagoons. They are present in tropical and sub-tropical regions and their vegetation is constituted by trees and herbs that are adapted to low or none soil oxygen content and to a great salinity variation. They have diverse roles such as nursery, coastal erosion control, wind control, diversity, coastal productivity, animal refuge, birds watching and population riverine subsistence.

Brazilian coast is privileged by its geographic position that favors mangroves development and along its 7,400 km it has approximately 25.000 km2 of these ecosystems. But we can see the great impact upon them which is mainly determined by human tensors, especially urbanization. Cities like Sao Luiz do Maranhão, Aracaju, Recife, Vitoria are constructed in mangroves'areas (FAO, 1998).

Urban concentration in the coastal zone is as historical factor and this expansion is very dynamic what makes mangroves'ecosystems vulnerable to environmental degradation.

Considering the crescent date volume until now produced about mangroves' impacts and the quickness that this are happening it is necessary to represent them in a fast, efficient and consistent way and this can be done by remote sensing techniques. Image of remote sensing became to represent one of the more viable way of environment monitoring due to its fastness, periodicity and synoptic vision that characterize them (Costa, 1992). These techniques permit to follow in a efficient way the anthropogenic occupation over mangrove.

The evaluation of the classification exactness employed in this study was done by Kappa Index Agreement (KIA) and this process can compare map date to some other supposed date considered 100% corrects. Although this seems to be a job without much value, many applied methodologies in orbital images classification consider this index vital because they serve to evaluate derived maps exactness.

Tavares'mangrove river is located in Santa Catarina island, south Brazil and it occurs along estuaries. Its vegetation is composed by three characteristic species: Avicennia schaueriana Stapf & Leechm, Laguncularia racemosa (L.) Gaerth and Rhizophora mangle L. In the herb layer is the cordgrass Spartina alterniflor Loiseleur (Panitz, 1997). But it is impacted like the other mangroves by diverse human activities especially urbanization that are altering its diversity, structure, development, productivity. The present papers intended to elaborate a mangrove's thematic map showing its vegetation and all tensor present.


Study area

Tavares mangrove river is located in the southeast portion of Santa Catarina island, extending from 2738'40 "e 2740'06" south latitude to 4830'17 "e 4833'39"west longitude in 13,3 km2. The mean annual temperature is 19C and the mean annual rainfall is of 1500 mm, (IBGE, 1999). The main river - Tavares has 8,22 km2 that together with Fazenda stream make the second greatest hydrographic basin in Santa Catarina island.

Tavares mangrove's river is the biggest one and had its area reduced by the Aeronautic Basin and Hercílio Luz International Airport implantation and by landfill and artificial drainage. The road that cross the mangrove to the north-south direction forms a dick that functions as a dam for the tides. The Carianos village is one of the areas that has been embanked and in its east and south portion mangrove has lost terrain by forest cutting and drainage that have given place to prairies (Fig.1). In 1982 part of this mangrove was included in a Natural Extractivist Marine Reserve in order to conservate one of the most population extractivist resource the mollusk Anomalocardia brasiliana.

Figure 1 . Study area ­ Tavares mangrove's river, Santa Catarina, south Brazil.

The methodology for Tavares' mangrove vegetation unities identification was based upon tonality and texture pattern scales and utilized by Ponzoni et al. (1995), Panitz (1997) and Froidefond et al. (1997).

Aerial photographs were employed for the study. All digital procedure was made with Spot PAN, 714-405 orbit images, 07/11/1995, 10 spatial resolution, 1o:35h, incidence angle 16. In the moment of image acquire the tide level was a + 0,80 m above the mean sea level of 0,63 m. The Landsat Thematic Mapper TM images, channels 1,2,3,4,5 and 7, digital form, orbit 220-078, 15/07/93 and 30 m spatial resolution, mean tide of + 0,20 m (DHN, 1995). The images were imported to a GIS/IDRISI.2.0 (Clark University) and manipulated by diverse statistical modules and spatial available analyze. In field work date about species distribution were acquired. New thematic maps were elaborated and a model constructed, classifying and actualizing soil occupation.

The electronic optical systems are capable of detect and register under image form or not, energy flux reflected or emitted by distant objects. The electromagnetic spectral band important for this research was in the visible region (380 to 750 nm) and infrared very important of remote sensing.

In this work we utilized two satellites - the TM (Thematic Mapper) a systems of multispectral VARREDURA that has a better spatial resolution, a better object discrimination in the earth surface, more geometric fidelity in a colored composition in the bands 3,5,4 RGB. The other satellite the SPOT in the panchromatic mode with a spatial resolution of 10 m.

First was analyzed the ALVO form (spatial information) and the sensors response ­ seven in the Landsat and four in the SPOT (spectral information). Pixels' position in the TM image were corrected to make them coincide with the true soil position by referring to a image in the SPOT-Pan utilizing control points of high precision. In this way, the resolution of 30 m and 10 m were w attained through the two satellite date combining with photo-interpretation date. The acquisition and date confirmation in the field were of great importance to the multispectrals image interpretation and they permit to make clear the information that were not good in the previous view of Landsat/TM and Spot/PAN images.

The digital images procedure techniques followed the classic ones for remote sensing. We employed the transformation technique HIS (I-""Intensity", H-'hue, S"-saturation"). These attributes were analyzed and manipulated individually, contrary to the RGS systems where they are interconnected intrinsically.

The IHS systems utilized in the process through three band containing spectral information, calculating the hue intensity and saturation through mathematical algorithms that relate colors spaces. Later we applied a linear contrast enhancement (ALC). The last phase involved an image reversion I, H and S to the RGB coordinates, because video monitors work only in this system. The use of this technique is necessary to make the fusion or "fuzzy"of the SPOT/PAN and TM images.

This procedure involved the component I, H and S calculation from the three selected band of TM 3,5 and 4 applying contrast enhancement of component I by the Spot image and applied to inverse transformation HIS-RGB. In this way it was possible to obtain a colored composition with a spectral resolution corresponding to the three TM band and the spatial resolution of 10 m of the SPOT/Pan.

The phase of the integration process of the applied image are presented in Figure 2.

Figure 2: Phases of images fusion SPOT/P and TM.


Among the combination series between the effected TM band we choosed as the best the composition 3,5 and 4 RGB (Fig. 3).

Figure 3: Colored composition 3,5 and 4 (RGB) TM image - Tavares river mangrove.

According to Pohl et al. (1998) the image fusion integrates different date that give more information from each exclusive sensor giving better date interpretation and more precise and reliable results.

The obtained image in the I, H and S components through Colspace module from IDRISI can be viewed in figures 4 , 5 and 6 respectively.

Figure 4. Image of component I (intensity).

Hue band derived from COLSPACE
Figure 5: Image of component H (hue)

Comparing figure 3 and 7 we can see that mangrove's vegetation is better discriminated, observed by more dark and clear tone of green because the two sensors combination of the crossed image maintains the diverse tonality in the TM images classes, with a spatial resolution of 10m against a 30 m resolution. In figure 3 mangrove's vegetation appears in homogeneous red ton because vegetation intensely reflects electromagnetic energy in the wavelength relative to the infrared, appearing in red colors.

Figure 6. Image of component S (saturation).

Sensors'fusion proportioned the spectral resolution necessary to typical mangrove's vegetation discrimination associated with other diverse soil uses. A supervisioned classification was done through the MAXVER algoritm of MAXLIKE/IDRISI.

The colored composition resulting from the HIS components is presented in figure 7
Figure 7: Colored composition 3, 5, and 4 (RGB) of Landsat TM/SPOT image fusion.

The black mangrove Avicennia schaueriana, the predominant specie was classified as high mangrove 1 and 2. Besides the specie ecological characteristic as salt excretion, probably this differentiation is due to different arboreal layer height and this shows how sensors'fusion is important. This specie due to its dominance gives to the mangrove an homogeneous physiognomic aspect; that was also viewed by Cintron (1981), Panitz (1992) and Soriano-Sierra (1993).

After this classification we got the confusion matrix ("root mean square"- RMS) to evaluate through a probabilistic view, the relation of the classified targets with their physical distribution compability. The model calculated the classes'global values through the Kappa Index (Kappa Index Agreement ­KIA), (Tab. 1). This index has been recommended as an exactness appropriate measure because it really represents the confusion matrix.

Table 1: Kappa (Kappa Index Agreement-KIA).

Through Kappa's formule, the reached results through the supervisioned classification (MAXLIKE) obtained a good achievement in the classification getting 93% of efficiency. The hybrids products of image fusion were superiors than the original images.

We can conclude that this integration is an adequate tool to classify areas of small portion with details richness.

The utilized technique in this study has proportionated a good distinction interes's classes and a better significance in the Tavares' mangrove river's texture patterns discrimination.

Satellites images utilization represent a rapid and efficient way to follow anthropogenic mangroves' occupation and tends to direct future research filling the existing gaps besides, they promote knowledge enhancement in adequate time and of the mangrove's research available resources along the Brazilian coast.



Cintrong,G. & SCHAEFFER-NOVELLI,Y. Introduccion a la Ecologia del Manglar. Instituto Oceanográfico, Universidade de São Paulo/SP. 147p. 1980. [1]

FAO. Forestry and Food Security.Paper 90. Roma, 129 p. [2]

GONG, P.O. Application of remote sensing to Determine the Status osf Mangrove Forest along Kenyan Coast. 1997. [3]

IBGE. Projeto Gerenciamento Costeiro. Diagnóstico Ambiental do Litoral de Santa Catarina. Relatório Final (Integração dos Domínios Natureza e Sociedade). 1999. [4]

Panitz. C. M. N. Produção e decomposição de serapilheira no manguezal do rio Itacorubi, ilha de Santa Catarina, Florianópolis, Brasil. Tese de Doutorado, Universidade Federal de São Carlos, Brasil. 1986. [5]

Panitz, C.M.N.1997.Ecological description of Itacorubi mangrove, Santa Catarina island, Santa Catarina, Brazil. P.204-225. in Kjervfe, B.; Lacerda, L.D. & Diopp S. El Haddji eds. Mangrove Ecosystem Studies in Latin America and Africa. ISME. UNESCO, Paris. [6]

Ponzoni,F.J. Comportamento espectral da vegetação. Instituto de Pesquisas Espaciais - INPE-5619-PUD/065. São José dos Campos, SP.,37p. 1995. [7]


Creative Commons License Todo el contenido de esta revista, excepto dónde está identificado, está bajo una Licencia Creative Commons