solution for mapping in general and cadastral surveying in
particular (Harvey and Hill, 2001).
The current and upcoming high spatial resolution satellite
imagery is expected to have a significant impact on mapping
applications of primary data acquisition, especially on attending
surveyor's interests and accuracies. — Spatial resolution of
multispectral data was reduced from 20m to 5m for SPOT
products in less than two decades. The use of panchromatic
products improved with the launch of high resolution
commercial satellite like IKONOS and Quickbird, and the
production of accurate terrain model is a big success mainly for
developing countries (Bernard and Munier, 2003).
Few cadastral projects have used satellite imagery to delimitate
parcel boundaries. SPOT images (SPOT 4) have been used in
Argentina and Nicaragua fiscal and physical cadastral pilot
projects, yet because of its spatial resolution it was conclude
that it was not possible to produce accurate boundary maps for
cadastral purpose (Axes, 2004). In general, because of the
specification of the images, high resolution satellite imagery can
be used for pre-cadastral projects as an analysis instrument for
the delimitation process, also as a support in the procedure and
methodology decisions process (GDTA, 1997; Axes, 2004).
Descriptions of area of study, creation of cadastral maps, less
expensive updates, multifunctional cadastral data as land value
maps via remote sensing and GIS techniques (Nisanci and
Yomrahoglu, 2002) are some of the potential use of the satellite
imagery in cadastral projects. The combination of pre-cadastral
data from satellite images and traditional surveying techniques
allows an execution of cadastral projects in only 20-50 months
(Lebeau, 1999; Axes, 2004). The significant reduction of costs
is one of the most important attractiveness of the use of satellite
images for cadastral projects (Garcia, 2001); in Guatemala the
cost of 1/30 000 aerial photographies for a nearby region was
US$ 30/Km° (Kadaster, 2001) compared to US$ 4/Km’ for the
orthoimage used in this study .
SPOT 5 was designed to improve the geometric performance of
its previous models. SPOT 5 data is constituted of
panchromatic band, visible infrared and shortwave infrared
band with a swath width of 60 Km. Stereopairs of 5m spatial
resolution are acquired systematically from across track
solutions, allowing the production of 2.5m orthoimages by a
resampling algorithm. As Bernard and Munier (2003)
reminded, this production will prove interesting and efficient
maps for rural cadastre where the accuracy needed in not less
than the orthoimage spatial resolution.
3. DESCRIPTION OF THE STUDY AREA AND DATA
3.1 Study area
The study area is located in the department of Izabal, Guatemala
(figure 1). Located in the northeast region of the country, this
tropical area includes a number of plains and low rugged
mountain ranges. Generally the study area was chosen for its
diverse set of terrain features. The flood plains, characterized
by rich deposits, are occupied by rice, banana plantations and
cattle breeding Bare soil, scrub, survival plantations and
forests cover the rugged areas.
3.2 Database
Two types of data were integrated to provide the database used
in this study; namely SPOT 5 orthoimage and field data. The
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004
image processing systems used for this project were ERDAS
Imagine 8.3, AutoCAD 2000 and ArcView 3.2.
Ground control points were provided to SPOTIMAGE to
relocate the image to 2.5 m accuracy. Field data included the
measurement of 669ha. with total station and GPS equipments
for rural and urban properties. ^ Sub-areas were divided
depending on the extension and topography. by this, large
extensions in flat terrain, medium and small estates in rugged
topography as well as peri-urban and urban properties were
surveyed and identified on the orthoimage.
2 | 4 5
| x
| a x
i m FE” Ten
i ^ i
| T i i
Í e. i
i tu i
i * |
| j i
| HL |
i i
i i
| i
i i
i i
Ï i
i s
i i
i
i
i .
i Á
i A
{ : 23
Ï x i
i € 2 £
i Y n
| : 3 |
- NEN
ab ar
“uh i
Figure l. Map of Guatemala, the image extension is indicated
with a rectangular box.
4. METHODOLOGY AND RESULTS
A number of display scales were evaluated for the orthoimage
identification process. Depending on the region and nature of
the area — topography, size of the parcel and image
identification utility — the appropriate scale was chosen,
providing the clearest visual distinction of boundaries and
vertices.
As the quality of the image was considered poor, the
identification was held on the image itself and two derived
enhanced images — 3x3 and 5x5 windows.
The evaluation was made by comparing the resulted digitised
data from the total station or GPS measurements — reference
data — and the one from the orthoimage identification —
extracted data — for each parcel encountered in the study arca.
Land surveying analysis are mainly composed of comparisons
of areas and distance between vertices among the distinct data
as shown in table ! and figure 4 and 5.
4.1 Rural properties
For large extension in flat terrain, boundaries were easily
identified in the enhanced orthoimage as observed in figure 2
and 3. In this case, field borders are trees, live enclosure or
fences with presence of vegetation, roads or foot paths and
water drainage with the presence of low altitude vegetation.
Variation of surface between the reference data and the
extracted one is very low (0.14%) and in average the
localization of the boundary in the identification process is of
4.3 meters away from the reference data.
Intern
MÀ
