Y GAPS 
al surveyings. 
nd sometimes 
tial resolution 
ips using data 
input, aerial 
] on a 900nm 
al cartography 
is and Albertz 
ons were used 
canopy gaps 
lata fusion of 
VOODROFFE 
& Leechman 
id Rhizophora 
1 (SANCHEZ 
lopy gaps are 
ina Stapf & 
ist border of 
y grows, the 
jally in terms 
| construction. 
'e areas in the 
IITHSONIAN 
monitoring is 
legislation has 
t Preservation 
ays this legal 
WALCACER, 
TANOPOLIS, 
constructions 
s available for 
ay, Kraus and 
ransformation 
an alternative 
tographs and 
-ular goal was 
1 schaueriana 
ipplication of 
parallelepiped, minimum distance to means and maximum 
likelihood supervised classification techniques using as input 
data fused images. Kraus and Albertz techniques were used and 
optimal results were obtained using minimum distance to means 
supervised classification techniques based on data fusion 
conventional panchromatic colour aerial photographs and 
digital infrared surveyings. 
2. METHOD 
Method consisted of three core tasks: first task was data 
acquisition and processing of input data. Second task was the 
creation of classification raster base by application of data 
fusion transformations. Third task was output processing by 
determination of field-truth and use of supervised 
classifications for canopy gaps quantification. 
2.1 Data acquisition 
2.1.1. High spatial resolution raster base: On a first stage, 
two positive colour aerial photographs of Itacorubi mangrove 
were scanned using a Carl Zeiss photogrammetric device and 
associated software. A 28mm scanning resolution was used. 
Final image was a 120Mb true-colour tag information format 
file (tiff), with specific hardware parameters header fields. 
Aerial photographs were taken on November 2000. Images 
were resampled and georreferenced using digital version of an 
original Florianópolis Municipality 1:5000 scale topographic 
sheet. Georreferenced photographs were added as raster layers 
into a geographic information system environment. 
2.1.2. Aerial surveying and 900nm infrared raster base: On 
a second stage, on September 2001 a thematic aerial infrared | 
surveying was carried out. Then, a digital sensor, associated 
hardware and software were installed on board a Cessna 
Skyhawk plane. Sensor's main axis was located in a vertical 
position near the plane's main axis using a topographic bubble- 
level. A Universal Serial Bus (USB) 2m length cable made the 
connection to a standard notebook. 
Flying height over Itacorubi mangrove was about 452m. Two 
flying lines, in North-South direction, covered the area and 28 
true-colour non-compressed images were created. Each of them 
covers 225m by 170m. Caption time interval was fixed to 5 
seconds. During the flight, surveying optical quality by real- 
time monitoring, hardware communication and system 
variables were controlled. 
As main input source for the thematic surveying, a digital 
sensor with USB data connection to PC was used. The sensor 
was controlled by software using Visual Basic programming 
language and owner's specific drives. Basically, the software 
has a frame-to-frame or continuous record capability, with user- 
defined caption time intervals. Output images can be recorded 
using both true-colour compressed or non-compressed file 
structures. 
The charge-coupled device (CCD) was a 6mm width CIF type, 
352 elements (horizontal) by 288 elements (vertical), with 
integrated analogical/digital electronic interface. CCD spectral 
sensitivity ranges from 300nm to 1050nm. The optical system 
has a 3.9mm focal length objective, and a 52? field-of-view 
angle. ; 
IAPRS & SIS, Vol.34, Part 7, “Resource and Environmental Monitoring”, Hyderabad, India,2002 
Two infrared filters were coupled to digital sensor for the aerial 
thematic surveying. For these filters, spectral transmittance 
begins at 720nm and 900nm. Best results were obtained with a 
900nm based reconnaissance system. 
During the aerial surveying, hardware connections, caption 
time interval between images, image size, light, colour, bright, 
gain, real time monitoring and hard disk storage were 
controlled using a 650MHz notebook. 
Images were resampled and georreferenced using digital 
version of the Florianópolis Municipality 1:5000 scale 
topographic sheet. Georreferenced files were added as raster 
layers into the geographic information system environment. 
2.2 Data fusion transformation 
Performing into the GIS a spatial overlapping of the high 
spatial resolution and 900nm infrared bases, two spatially 
referenced sample windows (A and B) were created. Sample A 
window served as reference for field truth and used 
classification methods. Sample B window was used for a case 
study where canopy gap area was automatically calculated. 
Sample windows were true-colour 703 pixels width by 957 
pixels height (133m by 181m) and 328 pixels width by 691 
pixels height (62m by 130m). The studied windows have 
representative environmental conditions of the habitat situation, 
with minimum man interference. Since November 2000 to 
September 2001, no significant spatial or phenologic variations 
were detected. 
Data fusion was performed applying Kraus and Albertz hue- 
saturation-value trans-formation technique. Thus, hue and 
saturation were taken from high spatial resolution samples 
window, and value was taken from 900nm infrared image. 
Image processing tasks were performed with the aid of Erdas 
version 8.4 specific software. Spatial relations, images cutting 
and georreferencing tasks were carried out using ArcView 
version 3.2 geographic information system software. 
2.3 Output processing 
2.3.1  Field-truth: For a quantification of mangrove 
coverage and canopy gaps, in sample A window a conventional 
stereo photointerpretation over polyester sheet was made. 
Ground tasks consisted of trees, gaps and soil visual 
reconnaissance. By scanning the polyester sheets, Avicennia 
schaueriana Stapf & Leechman trees and canopy gaps features 
were digitised. The raster base was converted to a vector layer 
using a tracing program. 
During the interpretation activities, there were no technical 
difficulties for features separation. Trees elevation over the 
ground is expressively different than herbaceous vegetation and 
naked soil. In some cases, trees shadows were a restrictive 
element for a right interpretation. Also, it was considered ° 
previous field studies such as CINTRON & SCHAEFFER- 
NOVELLI, 1981; PANITZ, 1986 and SANCHEZ DALOTTO, 
2002. These previous Itacorubi mangrove field studies have 
calculated Avicennia schaueriana Stapf & Leechman density 
and coverage values, which were used as support for 
determination of field-truth.