INSAR MONITORING OF THE NYIRAGONGO – NYAMURAGIRA
VOLCANOES (DR OF CONGO). STUDY OF THE NYIRAGONGO JANUARY 2002- AND
NYAMULAGIRA NOVEMBER 2006 ERUPTIONS.
N. d’Oreye
(1), F. Kervyn
(2), C. Wauthier
(3), V. Cayol
(3), and the GVO team
(4)
(1)
Geophys./astrophys. Dept., Nat. Museum of Natural History, 19 rue Josy Welter, 7256 Walferdange, Luxembourg,
Dept. of Geology, Royal Museum for Central Africa. Leuvensesteenweg, 13, 3080 Belgium.
Université Louis Pasteur, Lab. Magmas et volcans, Clermont Ferrand, France. [email protected] &
[email protected] , (4)
Goma Volcanological Observatry, Goma, DR Congo
ABSTRACT
This paper presents the ongoing efforts developed in
North Kivu (DRC) for the implementation of the
understanding, the monitoring, and the management of
the risks associated to the activity of the Nyiragongo
and Nyamulagira volcanoes.
Most of the ongoing activities in the region have been
motivated by the eruption in January 2002 of the
Nyiragongo that had drastic consequences for the city of
Goma situated at the foot of the volcano and its
inhabitants that had to evacuate in difficult conditions.
In the framework of the SAMAAV (Study And
Monitoring of Actives African Volcanoes) and
GORISK projects, InSAR monitoring coupled to ground
based measurements has started. Ground deformations
associated to that January 2002 eruption and the more
recent Nyamulagira November 2006 eruption are
described. A preliminary model for part of the
deformation associated to the Nyiragongo eruption is
presented.
1. INTRODUCTION
This work is performed in the framework of multiple
initiatives:
1.1. The SAMAAV project
SAMAAV was developed based on an ESA CAT-1
(CT3224) project dedicated to the study and monitoring
of active African volcanoes using ERS and ENVISAT
SAR data. The four sites under concern are the Fogo
(Cape Verde), the Ol Doinyo Lengai (Tanzania), the
Nyiragongo-Nyamulagira (DR Congo) and the Mt
Cameroon. SAMAAV that is coordinated by the Royal
Museum for Central Africa (RMCA, B) and the
National Museum of Natural History (NMNH, L)
involves other partners in Europe and Africa: University
of Liège (B), the University of Ghent (B), the Instituto
Superior Tecnico of Lisbon (P), the University of
Clermont-Ferrand (F), the Dar es Salaam University
(Tanzania), the Geological Survey of Tanzania,
(Tanzania), the Instituto Nacional de Meteorologia e
Geofísica de Cabo Verde, the Goma Volcanological
Observatory (Dem. Rep, Congo), the University of Buea
(Cameroon).
1.2. The GORISK project
Based on the positive results obtained in SAMAAV, the
GORISK project was set up which focuses on the
Nyiragongo – Nyamulagira volcanoes. It aims at
implementing the local ground deformation monitoring
capacity using both spaceborne and ground-based
techniques. Ground measurements of geochemical
parameters are also involved; they include the
monitoring of water quality as well as gas emanation
from the sub-surface.
GORISK collaborates with two external initiatives, the
US-VISOR and EU-NOVAC projects that are providing
respectively spaceborne and ground based
measurements of the volcanic plume. The type and
amount of gases are monitored in addition to the plume
dispersion path that is provided to GORISK for further
GIS integration.
GORISK is a service-oriented project dedicated to three
local users: The GVO, the United Nation unit for the
volcanic risks assessment and management, and the
CEMUBAC, a Belgian NGO active in the health
domain and wich performs an epidemiological study
based on the volcanic plume dispersion maps provided
by GORISK.
GORISK is supported by the STEREO-II program of
the Belgian Ministry of Science Policy and the
Luxemburg National Research Fund. The project is
coordinated by the RMCA and the partners are the
NMNH, Univ. of Luxemb., Univ of Naples, Univ. of
Clermont (F).
2. GEOLOGICAL SETTINGS
The Nyiragongo and Nyamulagira volcanoes are located
in the western branch of the east African rift system
(Figure 1). Compared to the eastern rift branch
characterized by a moderate seismicity and intense
volcanic activity, seismicity in the western branch is
_____________________________________________________
Proc. ‘Envisat Symposium 2007’, Montreux, Switzerland 23–27 April 2007 (ESA SP-636, July 2007)
very high whereas volcanism concentrates in four
provinces: the Toro Ankole in western Uganda, the
Rungwe in south-west Tanzania, the Kivu, and the
Virunga in East DR Congo. Nyiragongo and
Nyamulagira are the only active volcanoes of the highly
alkaline Virunga volcanic chain which started to
develop ~11 Ma ago [1].
Figure 1: Location of the study area in the east African
rift (inset). In red are the lava flows produced by the
2002 eruption. Cyan: fractures extracted from the
geological map. The ground based network is composed
of seismometers (triangles), tiltmeters (stars), and GPS
(dots). In background: LANDSAT draped on SRTM-3.
The Nyiragongo (3470 m asl), a steep stratovolcano is
characterized by silica-undersatured and ultrapotassic
lavas as well as a permanent crater lava lake. Located at
less than 15 km from the city of Goma, the last eruption
in January 2002 caused significant casualties and
damages and had an important socio-economic impact
on the region.
The Nyamulagira is a shield volcano located at ~10 km
NNW of Nyiragongo. It is the most active volcano of
Africa as it regularly erupts every two years [2]. Most of
the time, the eruptions are producing lava flowing
northwards in uninhabited areas.
3. THE NYIRAGONGO JANUARY 2002
ERUPTION
The eruption of the Nyiragongo on 17th
January 2002
occurred along an opening fractures network on the
southern flank in the direction of the city and the
airport. Two major lava flows reached the city
destroying about 15% of the town. In total, the volume
of lava erupted has been estimated to ~25 million m³[3],
[4]. Most of the 400.000 inhabitants evacuated the city
towards neighbouring Rwanda although many of them
came back two days later immediately after the
eruption; about 170 people died during the eruption.
The eruption occurred following sequences of tectonic
earthquakes that occurred in October and later in
beginning of January. These seismic events with
unusual magnitude for the region were accompanied by
significant changes in the phenomenology (e.g. dark
plume and rumbling on top of Nyiragongo, lake Kivu
level change…). The fractures that started to open from
North towards the South has produced three main flows
partly following the last flows of 1977; two of them
have reached the city and one of them reached the Lake.
3.1. Deformation observed from InSAR
Interferograms computed with ERS data acquired before
and after the eruption display complex regional
deformation with large fringes wrapping around the
Nyamulagira volcano and narrow linear fringes
suggesting a subsidence of Goma following an axis
parallel to the newly open fissure (Figure 2). This
subsidence was also detected in the field with asymetric
lake level changes observed along the shoreline in the
weeks following the eruption [4].
Figure 2: Deformation ERS interferogram (Sep. 2000 –
July 2002). The subsidence of the city of Goma (dashed
white line) is suggested by the narrow and linear
fringes. To the West, the fringes are wrapping around
the Nyamulagira volcano as a result of a deflation.
A signal possibly associated to a dyke injection bellow
the fissure is also observed (see modelling section).
Although the time baseline is unfavourably large (2000-
2002) due to some ERS technical reasons, another work
realized by [5] using RADARSAT pair from December
2001 and February 2002 shows similar pattern
indicating that the deformation took place in a short
period around the eruption.
3.2. Ground based monitoring system
The development of monitoring systems started in 2002
with the deployment by INGV of a seismic network that
encompasses the two volcanoes [6]. Networks of
telemetred tiltmeters and permanent GPS have been
recently implemented in the frame of the GORISK
project. The network configuration was driven by
InSAR observations and field safety considerations that
impose a concentration most of most equipment on the
southern flanks area.
3.3. Modelling
The complex pattern of deformation shown in the
InSAR ERS interferogram suggest that the deformation
is the sum of different contributions from several
phenomena such as a dyke associated with the eruptive
fissure, and probably a normal fault(s) close to the city
of Goma (Figure 3). In the Nyamuragira area, the
fringes could also contain atmospheric artifacts.
As these sources are close to each other, their
contributions add and hence are difficult separate. In
addition deformations from once source will probably
also influence the deformations of the others sources.
Thus, the different sources can not simply be
supperposed but must be taken into account
simultaneously. This can only be done using numerical
modelling.
Figure 3: ERS interferogram (September 2000-
July2002, Bperp = 62m) and the trace of the open
fissure (in green).
The modelling is at a preliminary stage so that only the
signal attributed to the dike is taken into account. Only
the data close the dyke were considered and sub-
sampled using a circular point pattern with an average
distance of 250m between the points (Figure 4 left).
Figure 4: Comparison between data and best-fit dyke
model for the January 2002 Nyiragongo eruption.
Subset of the ERS interferogram (September 2000 - July
2002, Bperp = 62m) with deformation attributed to a
possible dyke below the fissure.
In order to determine the best fit model, a method that
combines a 3D Mixed Boundary Element method [7]
and a neighborhood algorithm inversion [8] is used [9].
The modelling method takes the topography into
account and the inversion method uses a misfit function
that takes the data noise characteristics into account.
The inversion has two stages : Search and Appraisal.
We assume that:
- the medium is homogeneous, isotropic and elastic,
- the dyke is submitted to a constant overpressure.
We assume the dyke is a quadrangle with its top part
corresponding to the trace of the eruptive fissure and a
bottom line determined by 3 parameters, dip of the
dyke, the depth of the middle point and the angle the
line makes with the horizontal. The dyke overpressure
is a fourth parameter.
The best-fit model determined during the search stage
(Figure 5) corresponds to a sub-vertical dyke
characterized by a Dip angle of 76.8°, a bottom line
lying -590.2m below the sea level (the average altitude
of Goma is about 1400m asl) and a sub-horizontal
bottom (9.9° with respect to the horizontal). This dyke
is submitted to a 6.6 Mpa overpressure, leading to an
average opening at the surface of 2.1 m, which is
consistent with field observations (fig 5).
Figure 5: Best fit model for the dyke below the open
fissure.
The appraisal stage of the procedure involves
calculations of model marginal probability density
functions using misfits values calculated during the
search stage. The diagonal of figure 6 represents the 1D
marginal Posterior Probability density functions,
allowing the determination of confidence intervals. As
only one radar look is used, the 95% Confidence
intervals are large. It also shows that botang is not
uniquely determined. The off diagonal terms represent
2D Posterior Probability density functions indicating trade-offs between parameters. In particular, the
pressure and the dike depth do not seem independent.
Figure 6: Estimation of the uncertainties of the model
(appraisal stage).
The preliminary dyke model fits well the data close to the
dike (rms error = 9 mm). The next stages of our study will
be to correct the interferogram for possible atmospheric
effects to the southwest of Nyamulagira and model the
remaining signal.
4. THE NYAMULAGIRA NOVEMBER 2006
ERUPTION
Six months after the 2002 Nyiragongo event, the
Nyamulagira erupted in July the same year and again in
May 2004. More recently, in November 2006, an
unusual eruption took place on the south-eastern flank
from a fissure located between the two volcanoes. The
surprisingly short eruption (about a week) produced a
large amount of lava which flown southwestwards in the
direction of inhabited areas including the village of
Sake.
The eruption was forecast after an important seismic
swarm was recorded during the weeks before.
Figure 7: ENVISAT deformation interferogram (Sep.
2006 – Dec. 2006) of the November 2006 eruption of
Nyamulagira. Fringes elongated NW-SE from southern
flank of Nyamulagira up to SW flank of Nyiragongo. A
concentric deflation is also clearly detected at the East
of Nyamulagira as well as deflation fringes at the East
of Nyiragongo.
4.1. Timing of the crisis and activation of an
emergency InSAR procedure
On Nov. 27th
, shortly after the first announcement by
the Goma Volcanicological Observatory of a likely
eruption that begun the day after, contact was taken with
ESA in order to fasten the access to the ENVISAT
acquisitions routinely planned by the SAMAAV project.
Because of the heavy fights in the area and poor weather
conditions peventing helicopter flights, the location of
the vent or fracture as well as the lava flow extent and
progression rate remained unknown although strongly
required for risk assessment and management purposes.
Because of ARTEMIS failure, an image that could have
been acquired two days after the onset of the eruption in
the most interesting and richest mode was cancelled.
The next image acquired on day +8 was collected on the
ESA FTP site immediately after the notification of
availability i.e. 18 hours after the image acquisition. The
first interferogram showing deformation signals was
produced within an exceptionally short time (less than
1h30 after that notification). This was possible because
of a good knowledge of the terrain and the best
parameters for the processing of InSAR data over that
area. A refined interferogram was computed after the
delivery of preliminary orbit data 2 days later.
The computed interferogram (Figure 7) shows a fringe
pattern suggesting an inflation of 22 cm (LOS) and 17
cm (LOS) in the axis of a fracture linking the two
volcanoes. To the East of Nyamulagira, a clear
subsidence pattern is also observed.
The location and the fissural type of the eruption were
clearly detected on the imagery.
4.2. Lava flow mapping and volume estimate
By comparing the averaged intensity of all the images
available prior and after the eruption, we estimated the
location of the erupting fracture and the extent of the
main lava flow (~16 km²) (Figure 8). This allowed to
estimate the minimum volume erupted i.e. 16 to 24
x106m³. These estimation will be refined after more data
becomes available.
Figure 8: Colour composite with 18 averaged images
pre-eruption (R,G) and 2 averaged images post-
eruption (B). The main lava flow contour is marked in
red and the change in intensity due to cinder deposit in
blue.
5. CONCLUSIONS
Despite the dense vegetation in the area, coherence is
preserved on recent bare lavas flows. Elsewhere, in
savanna type areas, coherence remains for reasonably
short temporal baselines. In certain cases, with short
baselines, the coherence can sometimes be preserved for
period longer than a year. This underlines the necessity
for regular acquisition planning to enhance the chance
of suitable interferometric pairs.
Moreover, to efficiently constrain the modelling of
source parameters the use of various acquisition modes
and look angles is strongly recommended.
The emergency InSAR acquisition and processing
procedure activated for the 2006 Nyamulagira eruption
has proved to be an appropriate tool for crisis
management. The deformation map has been
successfully produced within 1h30 after notification of
data availability on the ESA FTP site.
6. ACKNOWLEDGEMENTS
The data provided by the ESA CAT-1 3224 were
processed by DORIS TU Delft software.
DEOS and ESA precise orbits were used in the
processing.
This work is supported by the STEREO-II programme
of the Belgian Ministry of Science Policy and
Luxembourg National Research Fund.
The permanent GPS are provided by the National
Museum of Natural History.
7. REFERENCES
1. Ebinger, C and Tanya Furman. Geodynamical setting
of the Virunga volcanic province, East Africa. Acta
Vulcanologica Vol. 14 (1-2) 2002, 15 (1-2) 2003.
2. Simkin T., Siebert L., Blong R., Dehn J., Newhall C.,
Pool R. and Stein,T.C (1994). Volcanoes of the
World, 2nd Edition, A regional Directory,
Gazetteer, and Chronology of volcanism during the
last 10,000 years. Smithsonian Institution and
Geoscience Press, Tucon, Arizona, 349 pp.
3. Komorowski,· J-C., D. Tedesco,· M. Kasereka,· P.
Allard,· P. Papale,O. Vaselli,· J. Durieux,· P.
Baxter,· M. Halbwachs,· M. Akumbe,· B. Baluku,P.
Briole ·,M. Ciraba,· J-C. Dupin·,O. Etoy3,· D.
Garcin,· H. Hamaguchi,N. Houlié,· K. S. Kavotha,·
A. Lemarchand,· J. Lockwood,· N.
Lukaya,G.,Mavonga,· M. de Michele,· S. Mpore,·
K. Mukambilwa,· F. Munyololo,C. Newhall ·,J.
Ruch,· M. Yalire,· M. Wafula. The January 2002
flank eruption of the Nyiragongo volcano (Republic
Democratic of Congo): Chronology,evidence for a
tectonic rift trigger, and impact of lava flows on the
city of Goma. Acta Vulcanologica Vol. 14 (1-2)
2002, 15 (1-2) 2003.
4. Tedesco, D., O.Vaselli, P. Papale, S.A. Carn, M.
Voltaggio, G.M. Sawyer, J. Durieux, M. Kasereka,
F. Tassi. The January 2002 Volcano-Tectonic
Eruption of Nyiragongo volcano, Democratic
Republic of Congo. Journal Geoph. Res. Accepted
for publication.
5. Poland, M.. InSAR Captures Rifting and Volcanism in
East Africa. Alaska Satellite Facility News &
Notes, Summer 2006, Vol. 3:2.
6. Tedesco, D., Badiali, E., Boschi, E., Papale, P., Tassi,
O., Vaselli, O., Kasereka, C., Durieux, J., Denatale,
G., Amato, A., Cattaneo, M., Ciraba, H., Chirico,
G., Delladio, A., Demartin, M., Favalli, G.,
Franceschi, G., Lauciani, V., Mavonga, G.,
Monachesi, G.,Pagliuca, N.M., Sorrentino, D.,
Yalire, M. Cooperation on Congo Volcanic and
Environmental Risks. Eos, Vol. 88, No. 16, 17
April 2007.
7. Cayol, V. and Cornet, F. Effects of topography on the
interpretation of the deformation field of prominent
volcanoes – application to Etna. Geophysical Res.
Lett. 25-11, 1998
8. Sambridge, M. (1999a), Geophysical inversion with a
neighbourhood algorithm. Geophys. J. Int., 138,
479– 494.
9. Fukushima, Y., Cayol V., P. Durand. Finding
realistic dike models from interferometric synthetic
aperture radar data: The February 2000 eruption
at Piton de la Fournaise. Journal Geoph. Res.,
VOL. 110, B03206, 2005.