distance to means of 72.3% in Avicennia schaueriana Stapf &
Leechman coverage and 27.7% of canopy gaps was considered
as most accurate results.
CLASSIFICATION | Avicennia Canopy Not
TECHNIQUE schaueriana gap classified
Stapf & coverage
Leechman (96) (96)
coverage
(96)
parallelepiped 37.2 29,5 33.3
minimum distance 57.6 42.4 0.0
maximum 50.2 49.8 0.0
likelihood
Table 1 comparative results for different supervised
classifications techniques used in sample A window.
IAPRS & SIS, Vol.34, Part 7, “Resource and Environmental Monitoring”, Hyderabad, India,2002
CLASSIFICATION Avicennia Canopy Not
TECHNIQUE schaueriana gap classified
Stapf & coverag
Leechman e (%)
coverage (%)
(%)
parallelepiped 56.6 29.2 14.2
minimum distance 723 27.7 0.0
maximum 64.0 36.0 0.0
likelihood
Table 2. Comparative results for different supervised
classifications techniques used in sample B window.
4. DISCUSSION
Canopy gaps for Avicennia schaueriana Stapf & Leechman
trees in Itacorubi mangrove were quantified by using four
techniques: conventional stereo photointerpretation,
parallelepiped, minimum distance to means and maximum
likelihood supervised classifications.
Considering results for sample windows A and B, data fusion
technique applied for spectral improvement of the automatic
interpretation base and later determination of canopy gaps in
Itacorubi mangrove was a successful task. Obtained results in
sample A window for minimum distance to means were
consistent with observed and measured field-truth results.
Parallelepiped classification was less accurate after maximum
likelihood results.
Sample B window results for the same technique were assumed
as more accurate. Considering these results as true,
parallelepiped classification was the less accurate with an
expressive number of pixels not classified, after maximum
likelihood results.
5. CONCLUSSIONS
As described, this technique could be used as objective
environmental monitoring tool for canopy gaps determination
in Itacorubi mangrove. Used as field-truth, conventional stereo
photointerpretation results were obtained from panchromatic
colour aerial photographs.
These photographs usually have a high spatial resolution, fact
that is a vantage when this technique is applied. But design of
layers is usually slow and sometimes subjective, carrying an
alteration of features metric properties. Most part of
classification software had no capabilities for 3D automatic
features reconnaissance, so human perception of elevation is
main source of features differentiation.
Changing spectral resolution is a way for classification base
improvement, but not always same spatial resolution as aerial
photogrammetric systems is available. Data fusion is an
alternative for merging metric properties of aerial photographs
with thematic richness of infrared digital surveyings.
When adequate training sites are defined, optimal results can be
obtained using minimum distance to means supervised
classification techniques based on data fusion conventional
panchromatic colour aerial photographs and digital infrared
surveyings.
6. REFERENCES
Albertz, J.; 1991. Grundlagen der Interpretation von Luft- und
Satellitenbildern. Wissenschaftliche Buchgesellschaft,
Darmstadt, Deutschland.
Cavalcanti Walcacer, F. Meio-ambiente urbano no Brasil. In:
Benjamin, Antonio H. Dano ambiental, prevençäo, reparaçäo e
repressáo. Sáo Paulo : Editora Revista dos Tribunais, 1993.
470p.
Cintron, G.; Schaeffer-Novelli, Y.; 1981. Los manglares de la
costa brasileña: revisión preliminar de la literatura. II. Los
manglares de Santa Catarina. Informe técnico preparado para
la Oficina Regional de Ciencia y Tecnología para América
Latina y el Caribe. Universidade Federal de Santa Catarina e
UNESCO. 68p.
Duke, N.; 1992. Mangrove floristics and biogeography. In:
Robertson, A.; Alongi, D. Tropical mangrove ecosystems.
Coastal and estuarine series 41. Australian Institute of Marine
Science, Queensland, Australia. 329p.
Florianópolis; 1998. Prefeitura Municipal. IPUF - Instituto de
Planejamento Urbano de Florianópolis. Plano Diretor do
Distrito Sede do Município de Florianópolis. Florianópolis.
IPUF. 238p.
IBGE; 1990. Geografia do Brasil - Regiäo Sul. Fundaçäo
Instituto Brasileiro de Geografia e Estatística, Diretoria de
Geociéncias - Rio de Janeiro : IBGE. 420p.
Kraus, K.; 1990. Fernerkundung, Band - Auswertung
photographischer und digitaler Bilder. Dümmler, Bonn.
New Mexico Highlands University, 2000a. Mangroves in
http://bos.nmhu.edu/mangrove.html, 20/09/2001. The
Man
last i
New
man
20/0
Unis
New
in
Man
last
Pani
man
Flor
Unis
Ciér
Reci
Rob
ecos
Insti
Sänc
sens
Intei
Bue:
Smit
disti
tidal
Smit
Troj
Aus
329
Smi
at.
20/0
Wol
hydi
man
Aus
329;
Wo
In: ]
Coa
Scie
The
Eng
Ferr
Uni