International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004
15°N
N
À
em
s
0 50 100
2e EEE
kilometres
14°N —
fem
Study area
Matagalpa
2
patagaip® me
s
^. ae Gr. ere
rl
12°N
CARIBBEAN SEA
87°W] 86°W|
Figure 1. Simplified map of Nicaragua with the location of the
SPOT-5 Hi+Pan images. The white frame represents
the area of investigation (equal to the Matagalpa
INETER 3054-1V topographic sheet).
85°W|
mentioned maps were used for validating the input of SPOT-5
products for hazard identification and risk assessment.
2. VALIDATION PARAMETERS AND
METHODOLOGY
2.1 GIS and RS database
CNES and Spot Image provided UNOSAT with two 60 x 60-
km-wide satellite images centered on Matagalpa (Fig. 1),
acquired during the dry season (19 April, 2003), with the
following specification:
— a level 1A, 2.5 m pixel size SPOT-5 Panchromatic
image, and
— alevel lA 4-band, 10 m pixel size SPOT-5 Hi image.
Both images were orthorectified to level 3 using a DEM based
on the 3-arc-second SRTM data and 13 Ground Control Points
(GCPs) from a Global Positioning System (GPS) field survey.
The combination of both images gives a high-resolution 2.5 m
4-band color image.
In addition to SPOT-5 imagery, the project database, built by
UNOSAT in cooperation with CIGMAT encompasses the
following Remote Sensing (RS) and GIS products:
— a l4-class land cover map, built from an unsupervised
classification of the previous SPOT-5 Hi image, field-
validated by CIGMAT,
— 1:50°000 INETER topographic sheets (mosaic)
covering the Rio Grande of Matagalpa watershed,
— aregional, 90 m pixel size DEM based on the 3-arc-
second SRTM mission, covering the upper part of the
Rio Grande of Matagalpa watershed,
— 8 20 m pixel size DEM rasterized from topographic
curves of the 1:50'000 INETER Matagalpa sheet,
— a 100 m pixel size slope map, created from the 20 m
DEM product, and a corresponding classified polygon
slope map (7 classes in degrees: «5, 5-10, 10-15, ]5-
25, 25-40, 40-60, 760),
— SAR land deformation maps of the Matagalpa urban
area, processed by UNOSAT partner Gamma RS from
a 1997-1999 set of six ERS images,
— a8 GIS inventory of the water bodies (rivers, lakes,
springs), extracted from the INETER maps and the
study of Havlíéek et al. (2002),
— several 1:50'000 geological vector layers (lateritic
soils, landslide deposits, scarps, etc), based on the
work of Havliéek et al. (2002), and
— vector layers of the initiation points and run out areas
of 1998 active terrain movements (Cannon et al,
2001), covering a 10-km-wide square north of the city
of Matagalpa.
All these raster and vector products are in a standard map
projection (NUTM 16, WGS 84).
2.2 Hazard definition
In this contribution, we will focus the suitability analysis of
SPOT-5 imagery to the most common geological hazards that
have occurred or might occur in the Matagalpa region and city.
These are landslides, mud flows and debris flows (Carrefio and
Barreto, 2000; Cannon et al, 2001; Havlièek et al, 2002).
According to the nomenclature of Lateltin (1997) and Schneider
(2001), landslides are defined as displacements by shearing of
compact masses of loose or rock grounds along a failure
surface. Mud flows are movements of material mass of
polyphasic nature (solid fragments and water) similar to a fluid
of variable viscosity. Lastly, debris flows are a mix of solid
materials (blocks, gravels, etc.) transported by a viscous fluid
(composed of fine sediments, clays and water) under the gravity
and which occurs in the drainage network. Whereas mud and
debris flows are instantaneous phenomena mainly driven by
huge rainfalls (e.g., Hurricane Mitch), landslides are slow
movements (from a few mm/y if substabilized to a few cm/y if
active) that may last for years, with abrupt accelerating phases
that might create along valleys stream blockage and subsequent
breaking up (high flood risk, particularly at Matagalpa). Key
risk factors, indicating a strong predisposition to terrain
movements are, in addition to geology, the presence of water
(proximity to river network, soil moisture, and discharge zones)
and moderate to steep slopes. Second-order factors that
reinforce this susceptibility include faulted zones, barren soils,
deforested areas and informal settlements.
2.3 Methodology
There are two ways for assessing geological hazards: (a) by
describing occurred and ongoing phenomena (inventory map
that reports and represents signs and indicators of terrain
movements) and (b) by classifying soil areas according to their
predisposition to documented natural hazards (susceptibility
map that orders according to three degrees which sectors are
potentially risk areas for each natural hazard type).
For the first approach, we tested the visual recognition on
SPOT-5 data of morphologies or soils typical of terrain
movements (disturbed vegetation, scarps, disturbed drainage,
etc.; Dikau, 1999) and their spatial delimitation. The medium
for this analysis is a combined Hi+Pan 2.5 m image, processed
in pseudo-natural colors and draped over a DEM for 3D
visualization. We also evaluated the potential of semi-automatic
methods, based on image radiometry, and particularly the use of
vegetation indices (NDVI) for generating low- or non-vegetated
thematic vector layers (Liu et al., 2002), that are further filtered
for extracting possible landslide or debris flow deposits.
For the susceptibility approach, we evaluate the integration of
SPOT-5 data to the multi-factor risk methodology developed by
Inte:
EPF
(CA
Sch
info
pho
up
hyd
ove
pre
doc
calc
rem
pot
gro
con
theı
infc
acc
mu
The
con
wit
be.
or |
wh
pro
be |
3.1
On
sin
act
toe
Fi
in
H: