International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004
unsupervised land use classification, cartography of settlements,
isolated buildings, plants and road infrastructure is feasible, but
significant editing work and ground verification, wherever
possible, is required to separate confused classes. This tool is
particularly helpful if there are neither quality, nor up-to-date
maps of the area available.
SPOT-5 collected spatial data on the elements at risk need to be
combined with demographic and socio-economic factors and
indicators on the economic activities for cost purposes. The
resulting cost study is then confronted to expected hazard
magnitudes and frequencies for vulnerability definition.
However, this method is far beyond being operational for
landslide, mud and debris flow events, since little work has
been done to assess the potential damages for a given terrain
instability (Wunderle, 2002).
4. CONCLUSIONS
SPOT-5 remote imagery is a useful tool for the detection of
large landslides and to a lesser extent mud flows and debris
flows (Table 1). A pseudo-color 2.5 m SPOT-S Pan+Hi image
draped on a DEM for 3D simulation allows to delimit hazard
areas at scales up to 1:25'000 (master plan) and to create a
comprehensive inventory map of occurred hazard phenomena,
consistent with field observations. The rate of vegetation
disturbance yields to some extent a qualitative monitoring of
the activity. However, the recognition of hazard characteristics
depends to a great extent on the ability and experience of the
interpreter. This remote sensing tool, although it does not reach
the ground resolution of aerial photography, usefully
complements this traditional surveying method. In Matagalpa,
semi-automatic approaches, analyzing image radiometry are not
suitable for retrieving landslide data. However, vegetation
indexes (NDVI), best filtered with the drainage network are
helpful for sorting out areas with debris flow deposits.
For hazard susceptibility maps (Table 2), integration of SPOT-5
products is less straightforward, since the derived thematic
information has to be filtered with slope classifications. At
Matagalpa, among the prime risk factors that indicates a
predisposition to landslides, only the information about
geological lineaments can be visually recovered from the
pseudo-color SPOT-5 image. Additional data, such as barren
soils and deforested areas can be extrapolated from land use and
change detection maps.
The contribution of SPOT-5 products for vulnerability purposes
(Table 3) is relevant where poor map coverage is available or
when up-to-date information is required. Land cover maps, after
demanding editing work and field validation are helpful for
mapping threatened elements such as settlements, buildings,
plants and road infrastructure. The later information must
nevertheless be combined with socio-economic data of the
elements at risk to assess their actual vulnerability.
In conclusion, SPOT-5 products are a complementary tool in
the process of risk analysis (hazard inventory, second-order risk
factors and vulnerable elements), which is tailored for mapping
large (kilometer-wide) hazardous phenomena at the watershed
scale. To fully take advantage of its potential, it needs to be
integrated as part of a GIS, together with other relevant
information, such as DEMs and geological and hydrographic
maps. However, a field survey remains essential for validating
SPOT-5 derived landslide inventories and land cover
classifications. Most of limitations observed with SPOT-5 could
probably be overcome with infra-metric resolution capabilities
in optical bands and metric resolution radar C-band.
REFERENCES
Cannon, S.H., Haller, K.M., Ekstrom, I., Schweig, E.S.HI,
Devoli, G., Moore, D.W., Rafferty, S.A., and Tarr, A.C., 2001.
Landslide initiation locations in Nicaragua. USGS Open-File
Report 01-412-A, U. S. Geological Survey, Denver, CO, USA,
17 p., 7 maps, http://pubs.usgs.gov/of/2001/0fr-01-0412-a
(accessed 26 April 2004).
Carrefio, R, and Barreto, H., 2000. Evaluación indicativa de
peligros derivados de fenómenos de inestabilidad y torrentiales.
COSUDE, Managua, Nicaragua, 28 p.
Dikau, R., 1999, The recognition of landslides. In: Casale, R.
and C. Margottini (eds.), Floods and Landslides. Integrated
Risk Assessment. Springer, Berlin, pp. 39-44.
Havli¢ek, P., Hradecky, P., Kycl, P., Mléoch, B., Mrázová, S.,
Novák, Z., Opletal, M., Rapprich, V., Sebesta, J., Sevéík, J.,
Vorel, T., and Devoli, G., 2002. Estudio geológico de riesgos
naturales, área de Matagalpa (Hoja de mapa 1:50'000 3054/IV)
Czech Geological Survey in collaboration with INETER.
Prague-Managua, 91 p., 4 maps.
Lateltin, O., 1997. Recommandations 1997, prise en compte
des dangers dus aux mouvements de terrain dans le cadre des
activités de l'aménagement du territoire. Série Dangers naturels,
OFEFP/OFEE/OFAD, OCFIM no 310.023f, Bern, Switzerland,
http://www.bwg.admin.ch/themen/natur/f/pdf/
empfmbg.pdf (accessed 26 April 2004).
Liu, J.K., Wong C.C., Huang J.H., and Yang, M.J., 2002.
Landslide-enhancement images for the study of torrential-
rainfall landslides. 23" Asian Conference on Remote Sensing,
Kathmandu, Nepal, http://www.gisdevelopment.net/aars/acrs/
2002/env/193.pdf (accessed 26 April 2004).
Schneider, T., 2001. Caractérisation multicritére des formations
géologiques du canton de Vaud (Suisse) et de leurs
prédispositions face aux dangers naturels. Mémoire de diplôme
d’études postgrades en “Géologie de l’Ingénieur et de
l’Environnement”, GEOLEP, EPFL, Lausanne, Switzerland.
Wunderle, S., Van Westen, C., Pasquali, P., 2002. Integration
of remote sensing techniques with statistical methods for
landslide monitoring and risk assessment, DUP-SLAM2 final
report, http://dup.esrin.esa.int/files/project/192-106-252-100
200415131610.pdf (accessed 26 April 2004).
ACKNOWLEDGMENTS
This contribution to the SPOT-5 validation program has been
undertaken in the frame of the project “Strengthening local
capacities for the management of local resources of Matagalpa
and Rio Grande watershed, Nicaragua” and benefited from the
assistance of the CIGMAT resource center in Matagalpa. They
should receive our best thanks for providing GIS products and
for carrying out the GPS field survey and land cover validation.
We would like to gratefully acknowledge U. Wegmiiller and A.
Wiesmann from Gamma RS who processed the interferometric
SAR data. Thanks are also due to S. Cannon for sharing the
USGS landslide data on the Hurricane Mitch program.
Technical support was generously provided by E. Bjorgo (data
processing), J. Delgado (figures) and I. McClellan (English).
Images © CNES 2003 courtesy of Spot Image.
“ Spot
b
|
KEY WO!
ABSTRA(
and commi
Thanks to
twin-instru
extensively
CNES anc
performing
CNES and
cadastral 1
Africa, ge
Sweden, a;
Finally, C
products, i
The SPOT
42P rocke
sensor wa:
since the |
required.
With SPO
user com
capability
Very Hig
large geog
resolution
metre and
60 km by
SPOT 5
applicatio:
In additio
quality an
Spot Imag
applicatio
Program (
This prog
selected |
evaluate
Principal
worldwid:
of variou
expertise
The PI se
site of the
an evalu:
applicatio