Content uploaded by Benoit Smets
Author content
All content in this area was uploaded by Benoit Smets on Aug 09, 2018
Content may be subject to copyright.
This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
A new map of the lava flow field of Nyamulagira (D.R. Congo) from satellite imagery
Smets Benoît
a,
⇑
, Wauthier Christelle
a,b
, d’Oreye Nicolas
c
a
Royal Museum for Central Africa, Geology Department, Leuvensesteenweg 13, B-3080 Tervuren, Belgium
b
University of Liège, Department ArGEnCo, Sart Tilman B52, B-4000 Liège, Belgium
c
National Museum of Natural History, Geophysics/Astrophysics Department, rue Josy Welter 19, L-7256 Walferdange, Luxembourg
article info
Article history:
Available online 10 August 2010
Keywords:
Nyamulagira
Nyamuragira
Map
Lava flow
Volume
Remote sensing
Multispectral
Radar
InSAR
Coherence
abstract
Nyamulagira (3058 m a.s.l.), a volcano of the Virunga volcanic province in the western branch of the East
African Rift, is Africa’s most active volcano with one eruption every 2–4 years. It represents a hazard for
the Virunga National Park and its vicinity. Despite such a frequent activity, Nyamulagira remains poorly
studied. The only existing volcanological map was produced in the sixties by Thonnard et al. (1965). The
occurrence of 19 eruptions since its publication makes it obsolete. In the present study we mapped the
Nyamulagira lava flows from 1938 up to the last eruption to date in 2010 using optical (Landsat, ASTER)
and radar (ENVISAT-ASAR, ERS, JERS) imagery. The results are integrated into a Geographical Information
System (GIS) and coupled with additional data sources. GIS use makes the new database a flexible – and
easy-to-update – tool for scientific purposes as well as for risk, environmental and humanitarian manage-
ment. Here a new lava flow map was produced. Volumes of the successive lava flows and affected areas of
the Virunga National Park were estimated.
Ó2010 Elsevier Ltd. All rights reserved.
1. Introduction
Nyamulagira (or Nyamuragira in Rwandese) is a 3058 m high
shield volcano with a 2 2.3 km caldera located south of the
Virunga National Park (SE part of N-Kivu, D.R. of Congo) (Fig. 1).
It is one of the two active volcanoes in the Virunga, a volcanic
province in the western branch of the East African Rift (EAR). This
volcano is Africa’s most active volcano with, on average, one
eruption every 2–4 years. Nyamulagira’s lava field covers over
1100 km
2
and contains more than 100 flank cones. Lavas are basic,
SiO
2
-undersaturated and K-rich (Kampunzu et al., 1982; Aoki and
Yoshida, 1983; Aoki et al., 1985). Such composition results in low
viscosity lavas able to flow for tens of kilometres. Comprehensive
descriptions of the Nyamulagira geology and eruptive history are
given by Pouclet (1976) and Hamaguchi (1983).
The only existing volcanological map of Nyamulagira was estab-
lished by Thonnard et al. (1965). This map was constructed by
interpreting three independent sets of aerial photographs. The
mosaics were assembled with the help of nine triangulated points
and theodolite field measurements (Thonnard, 1961; Thonnard
and Denaeyer, 1965). Comparison of the digitized and geocoded
map with recent field GPS measurements revealed local distortions
in the map, which induce errors of location reaching sometimes
more than 2 km in places. Contour lines on the 1965 map only give
a general idea of the relief as they lack absolute or relative eleva-
tion values. Nyamulagira also erupted 19 times since the produc-
tion of the 1965 map, which makes the latter totally outdated.
Partially updated maps are found as figure in publications (e.g.
Ueki, 1983; Zana, 1984; Louant, 1987; Zana and Hody, 1989) and
recently, Colclough (2005) and Kitagawa et al. (2007) used radar
remote sensing techniques to map the 1996, 1998 and 2001 lava
flows. In fine, no comprehensive and up-to-date compilation of his-
torical lava flows at Nyamulagira has been published yet.
An updated map with recent lava flows is essential as a tool for
scientific purposes, as well as for risk, environmental, urban and
humanitarian management. Nyamulagira’s eruptions can pose a
non-negligible hazard for local populations and their environment.
Coupled with deforestation, political and social unrest as well as
important demographic and urban growth, the potential risk/im-
pact from Nyamulagira’s volcanic hazards increases accordingly.
Villages, roads and the protected forest of the Virunga National
Park (VNP) are under the threat of volcanic eruptions. Given the
VNP is a UNESCO World Heritage since 1979 with ‘‘endangered sta-
tus” since 1994 (UNESCO, 1992–2009), impacts of the Nyamulagi-
ra’s volcanic activity upon forest and its unique ecosystem should
be assessed in future.
In order to meet these needs, lava flows from eruptions in 1938–
2010 were mapped with both optic and radar satellite imagery and
compiled into a GIS database together with data from other sources
1464-343X/$ - see front matter Ó2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jafrearsci.2010.07.005
⇑
Corresponding author. Tel.: +32 (0) 2 769 54 48; fax: +32 (0) 2 769 54 32.
E-mail address: benoit.smets@africamuseum.be (B. Smets).
Journal of African Earth Sciences 58 (2010) 778–786
Contents lists available at ScienceDirect
Journal of African Earth Sciences
journal homepage: www.elsevier.com/locate/jafrearsci
Author's personal copy
Fig. 1. Location of Nyamulagira and of the Virunga volcanic province.
B. Smets et al. /Journal of African Earth Sciences 58 (2010) 778–786 779
Author's personal copy
(maps, catalogues of geodata, publications, unpublished notes and
field data) using the ESRI ArcGIS
Ò
Software. The present paper de-
scribes the methodology used to generate the new lava flow map,
highlights scientific information that can be extracted from it and
discusses the contribution of the map to volcanic risk management.
2. Remote sensing data and methods
2.1. Remote sensing mapping of lava flows
Since the emergence of moderate-high spatial resolution space
imagery (15–30 m pixel size), satellite data have been used to map
volcanic terrains (e.g. Francis and Baker, 1978; Francis and Wells,
1988; Carn, 1999; Lu et al., 2004). Remote sensing is a powerful
tool for studying large structures, often impossible to appreciate
on the ground especially when they are remote and/or situated
in areas difficult to access, such as some volcanoes in developing
countries (Kervyn et al., 2007; Ernst et al., 2008). If multispectral
data like Landsat, ASTER or SPOT imagery enable accurate lava flow
mapping, cloud cover often limits their use, mainly in tropical
zones (Carn, 1999; Lu et al., 2004). In addition, frequent eruptions
generate lava flow overlap for which the flows boundaries are not
always clearly visible on optical images.
Radar sensors allow data acquisition at any time of day or night
and through cloud cover (Hanssen, 2001; Kervyn, 2001). Further-
more, Synthetic Aperture Radar (SAR) data can be processed to ob-
tain different and complementary products (amplitude, phase and
coherence) usable for lava flow mapping (Lu et al., 2004). SAR
amplitude images give information on the backscattering proper-
ties of the ground, which vary spatially with the nature and struc-
ture of the targeted terrain. Coherence images, which correspond
to the degree of similarity of the backscattering properties between
the two images used to form an interferogram (Massonet and Feigl,
1998), depict the changes of ground surface characteristics with
time (Hanssen, 2001). SAR interferometry (InSAR) also allows the
production of Digital Elevation Models (DEMs). Lava flow area
and volume can be measured by subtracting pre- and post-erup-
tion DEMs (e.g. Rowland et al., 1999; Lu et al., 2004). However
these InSAR tools may yield poor results when applied over vege-
tated areas because of the loss of coherence (or decorrelation) that
vegetation induces. In order to compensate for their respective
limitations, lava flows at Nyamulagira were mapped using a com-
bination of optical and radar-derived datasets.
2.2. Optical remote sensing data
Archive data from Landsat satellites, covering a time span of
three decades, has become progressively available at no charge
during 2008 (http://earthexplorer.usgs.gov or http://glovis.usgs.
gov). A time-series of images can be easily created. For Nyamulagi-
ra, a set of 8 Landsat 4–5 TM, 12 Landsat 7 ETM+ and 3 ASTER
images was used. We used Landsat spectral bands 5, 4 and 3 to en-
sure the best contrast between lava and vegetation, and between
the different overlapping lava flows, as suggested by Lu et al.
(2004). Band 5 is sensitive to soil and vegetation moisture whereas
bands 4 and 3, usually used to calculate a vegetation index, are
respectively sensitive to biomass and chlorophyll absorption. To
benefit from the highest spatial resolution of ASTER (i.e. 15 m),
we combined the spectral bands 3, 2 and 1. Details of these images
are summarized in Tables 1 and 2.
2.3. Radar remote sensing data
In the framework of volcano monitoring projects using InSAR
(Kervyn et al., 2005; d’Oreye et al., 2008; van Overbeke et al.,
in press) hundreds of SAR images, including ENVISAT ASAR,
ERS, JERS images (see Table 3 for data specifications and charac-
teristics), acquired over Nyiragongo and Nyamulagira in the last
15 years were processed. At the time of writing, over 2500 geo-
coded interferograms and associated coherence maps were com-
puted providing us with a huge amount of data for lava flow
mapping. This was done by tracking the decorrelation on syn-
eruptive coherence images, or by detecting the improvement of
coherence on post-eruptive coherence images. Nyamulagira’s lava
flows proved to be too thin (typically <5 m thick) to enable accu-
rate mapping by comparison of pre- and post-eruptive InSAR-de-
rived DEMs.
2.4. Mapping lava flows at Nyamulagira
With a rich optical dataset of 22 images over a 21 years period,
impact of cloud cover and lava flow superposition were minimized
Table 1
Specifications of satellites, sensors and optical products considered in this study. All these satellites are in a 705 km altitude sun-synchronous orbit with a 16-days repeat cycle.
Launch year Landsat 4 and 5 Landsat 7 Terra
1982 and 1984 1999 1999
Sensor used TM (Thematic Mapper) ETM+ (Enhance Thematic Mapper) ASTER (Advanced Spaceborne Thermal
Emission and Reflection radiometer)
Spectral bands 1 (B) = 0.45–0.52
l
m 1 (B) = 0.45–0.52
l
m V1 (G) = 0.52–0.60
l
m
2 (G) = 0.52–0.60
l
m 2 (G) = 0.53–0.61
l
m V2 (R) = 0.63–0.69
l
m
3 (R) = 0.63–0.69
l
m 3 (R) = 0.63–0.69
l
m V3 (NIR) = 0.76–0.86
l
m
a
4 (NIR) = 0.76–0.90
l
m 4 (NIR) = 0.78–0.90
l
m S4 (SWIR) = 1.6–1.7
l
m
5 (SWIR) = 1.55–1.75
l
m 5 (SWIR) = 1.55–1.75
l
m S5 (SWIR) = 2.145–2.185
l
m
6 (TIR) = 10.4–12.5
l
m 6 (TIR) = 10.4–12.5
l
m S6 (SWIR) = 2.185–2.225
l
m
7 (SWIR) = 2.08–2.35
l
m 7 (SWIR) = 2.09–2.35
l
m S7 (SWIR) = 2.235–2.285
l
m
8 (Pan) = 0.52–0.90
l
m S8 (SWIR) = 2.295–2.365
l
m
S9 (SWIR) = 2.260–2.430
l
m
T10 (TIR) = 8.125–8.475
l
m
T11 (TIR) = 8.475–8.825
l
m
T12 (TIR) = 8.925–9.275
l
m
T13 (TIR) = 10.250–10.950
l
m
T14 (TIR) = 10.950–11.650
l
m
Spatial resolution 30 m 30 m 15 m (visible and NIR)
120 m (band 6) 60 m (band 6) 30 m (SWIR)
15 m (Panchromatic) 90 (TIR)
Scene size 185 172 km 185 172 km 60 60 km
a
Two V3 bands exist for DEM production: V3 N (nadir view) and V3B (backward view).
780 B. Smets et al. /Journal of African Earth Sciences 58 (2010) 778–786
Author's personal copy
and most of the lava flows were mapped. Radar products were
needed to complement mapping of the 1996, 1998, 2000, 2001
and 2004 flows. A manual mapping method was preferred to auto-
mated methods (e.g. supervised and unsupervised classifications)
which yielded confusion among the frequently overlapping lava
flows and required important post-processing. The volcanological
map (scale 1/50,000) by Thonnard et al. (1965), geocoded and
orthorectified with ASTER images, was also used to help localize
and identify the 1938–1940, 1948, 1951, 1954, 1956, 1957 and
1958 lava flows.
The general location of each event was found in papers describ-
ing eruptive activity of the last century (Hamaguchi, 1983; Burt
et al., 1994; Smithsonian Institution, 1996–2009) and served to
identify the source of each flow. Frequent eruptions along the
NNW–SSE fracture network crossing the caldera make the location
of some recent eruptive vents and corresponding lava flows diffi-
cult. SAR coherence images were used to locate them (Fig. 2). Once
located, these lava flows were mapped using optical images. Next,
SAR data (both coherence maps and amplitude images) were used
to complete the lava flow mapping, especially for the 1996, 1998,
2000, 2001 and 2004 lava flows, which were only partially mapped
with optical images. Fig. 3 shows the lava flows from 1938 to
present.
The accuracy of lava flow mapping is directly proportional to
the spatial resolution of the images. Mapping errors depend mainly
on vegetation regrowth over fresh lava and ability of the operator
to visually discriminate the ground coverage. According to Pouclet
(1976), vegetation growth makes lava flow boundaries on images
Table 2
List of optical images used for lava flow
mapping. The path and row of the images are
respectively 173 and 61.
Satellite and
sensor
Acquisition date
Landsat 5 TM July 19, 1986
Landsat 5 TM August 7, 1987
Landsat 4 TM August 4, 1989
Landsat 4 TM August 20, 1989
Landsat 4 TM September 21,
1989
Landsat 4 TM May 19, 1990
Landsat 5 TM October 13, 1994
Landsat 5 TM January 17, 1995
Landsat 7 ETM+ September 17,
1999
Landsat 7 ETM+ October 3, 1999
Landsat 7 ETM+ December 6, 1999
Landsat 7 ETM+ October 3, 2000
Landsat 7 ETM+ December 8, 2000
Landsat 7 ETM+ February 26, 2001
Landsat 7 ETM+ August 5, 2001
Landsat 7 ETM+ December 11,
2001
Landsat 7 ETM+ January 28, 2002
Landsat 7 ETM+ December 30,
2002
Landsat 7 ETM+ January 15, 2003
Landsat 7 ETM+ January 31, 2003
Terra ASTER January 23, 2006
Terra ASTER June 19, 2007
Table 3
Specifications of satellites and SAR sensors considered in this study. All these satellites are in a sun-synchronous orbit. As the dataset contains lots of images and SAR-derived
products, only the time spanned by data from each satellite is given.
ERS 1 and 2 ENVISAT JERS
Country EU EU Japan
Launch year 1991 and 1992 2002 1992
Altitude 782–785 km 800 km 568 km
Repeat cycle 35 days 35 days 44 days
Sensor used SAR (Synthetic Aperture Radar) ASAR (Advanced Synthetic Aperture Radar) SAR (Synthetic Aperture Radar)
Spectral bands C-band C-band L-band
Frequency = 5.3 MHz Frequency = 5.3 MHz Frequency = 1.275 MHz
Wavelength = 5.66 cm Wavelength = 5.66 cm Wavelength = 23.5 cm
Polarization = VV Polarization = VV, HH, VV/HH, HV/HH, VH/VV Polarization = HH
Spatial resolution 30 m 30–150 m 18 m
Scene size 100 100 km 100 100 km 75 75 km
Time spanned by data 21/02/1996 to 16/12/2004 25/12/2002 to now 19/10/1996 to 28/02/1997
Fig. 2. Example of lava flow mapping assisted by InSAR products: the May 8–28, 2004 eruption. Comparing pre- (A) and syn-eruptive (B) coherence images from ENVISAT
interferograms allows the localization and identification of new lava flows (black flow on panel C). By modifying the ground coverage, the new lava flow changes the
backscattering conditions and induces decorrelation (black zones on A and B) clearly visible in contrast to the still highly coherent former flows (white areas on A and B).
Interferometric products are produced from ENVISAT images in ascending orbit, swath 2. The coherence maps on A and B span December 5th 2002–January 14th 2004 and
December 5th 2002–May 3rd 2006 respectively.
B. Smets et al. /Journal of African Earth Sciences 58 (2010) 778–786 781
Author's personal copy
indistinct after few decades. Thus, optical images used were cho-
sen as close as possible to eruption dates in order to minimize this
effect. Errors from both vegetation recovery (highly variable in
time and space) and manual mapping do not exceed one pixel
width on optical data, which corresponds to the accuracy of the
images.
There are two additional sources of error: (1) local misinterpre-
tation in near-vent area related to tephra deposits, and (2) quality
of SAR data (noise, effects of relief on coherence, etc.). These errors
are not expected to strongly influence the final result. As errors due
to tephra deposits cannot be assessed without field measurements
and lava flows were only mapped locally with SAR data, these two
Fig. 3. Map of most recent (1938-present) lava flows from Nyamulagira (darker flows are the most recent ones). The Nyiragongo lava field is also identified with its two most
recent flows from 1977 and 2002 eruptions. The new volcanological map based on this lava flow mapping contains additional information such as topography, petrology,
fractures and faults, adventive cones, VNP boundaries, roads, and urban areas.
782 B. Smets et al. /Journal of African Earth Sciences 58 (2010) 778–786
Author's personal copy
error sources were neglected for surface and volume calculations,
even if they must be taken into account for the interpretation of
the results.
2.5. GIS integration
The Geographic Information System (GIS) has become an essen-
tial tool in any cartographic project. It enables comparison and
overlap of any kind of data with geographic coordinates (i.e. lay-
ers), rapid production or update of thematic maps and quantitative
characterization of geographic features (e.g. length, area). GIS is the
most suitable technique for processing geographic data in areas
with constant changes such as active volcanic zones.
In addition to the lava flow layers, our GIS database contains
information related to the petrography of lava flows, the geological
and geomorphological structures (cones, faults, fissures and linea-
ments), the topography (DEMs and resulting contour lines),
hydrography (rivers and lakes), land use (e.g. urban areas, Virunga
National Park) and geography (e.g. roads, localities, political
boundaries).
3. The new map of Nyamulagira
The new map of Nyamulagira combines data from the two remote
sensing techniques and from the literature. Table 4 summarizes the
different geographic data (layers) integrated in the new map. The
lava flows are represented by a specific colour for each eruption.
Lighter and darker colours correspond respectively to older and
younger events. Lava petrology with corresponding location on the
map is from Thonnard et al. (1965), Aoki and Yoshida (1983) and
Aoki et al. (1985). Urban areas and lakes were mapped from 2001
to 2003 Landsat ETM+ and 2006–2010 ASTER imagery. The name
and location of localities were obtained from RDC-Humanitaire
NGO’s Map & GIS Data Center (http://www.rdc-humanitaire.net/
spip.php?rubrique30). The road and hydrographical networks result
from Landsat image interpretation and from the BEGo project
(2005). Contour lines were derived from the Shuttle Radar Topogra-
phy Mission (SRTM) DEM (http://seamless.usgs.gov/).
The map is frequently updated when new data are available
(e.g. very high spatial resolution images) or when a new eruption
occurs (e.g. 2010 eruption). The map is only available at the Royal
Museum for Central Africa and will be available for purchase soon,
along with an explanatory booklet in English or French (see http://
www.africamuseum.be/research/natural-sciences/earth-sciences
for future information).
4. Contribution to scientific research and risk management
The new map and the GIS database could have great potential
for scientific research. They could also prove valuable as basis for
hazard mapping and risk assessment (De La Crùz-Reyna et al.,
2000; Newhall, 2000). As an example, we use them to estimate
the volume of the different lava flows and to assess the impact of
historical eruptions on the area.
4.1. Lava flow volume estimation
Estimation of erupted volumes through time at a volcano pro-
vides volcanologists with information on the volcano dynamics
(e.g. effusion rate) and can be used for statistical analysis or sto-
chastic modelling of the eruption history (Burt et al., 1994).
As already explained, unlike the surfaces, the flow heights could
not be determined by DEM comparison due to their limited thick-
ness. Previous volume estimations at Nyamulagira typically used a
mean lava flow thickness of 3 m (e.g. Pouclet, 1976; Pottier, 1978;
Brousse et al., 1979, 1981, 1983; Caron et al., 1982; Ueki, 1983;
Burt et al., 1994; Colclough, 2005; Kitagawa et al., 2007). According
to Burt et al. (1994), this mean value is probably affected by an
uncertainty of only a few tens of percent. In the absence of new
field measurements, still prevented by local insecurity and fre-
quent civil unrest, we use the same mean thickness to estimate
lava flow volumes.
As previously discussed, the uncertainty in lava flow surface
mapping corresponds to the image accuracy that is ±1 pixel for
the location of lava flow limits. Thus, the error on the estimated
areas (E
s
) can be calculated and is given by:
E
s
¼Pp
where Pis the lava flow perimeter and pis the pixel size. The real
lava flow area (S) is then expressed as:
S¼S
e
E
s
where S
e
is the estimated area. We postulated that the mean thick-
ness (H) corresponds to 3 m for each lava flow. Consequently, the
volume (V) corresponds to:
V¼SHE
v
where E
v
, the error on estimated volume, corresponds to:
E
v
¼E
s
H
E
v
is directly proportional to E
s
and only an indicative error estimate
due to the uncertainty on Hthat is not taken into account. Hence it
serves mainly for comparing our results with other estimates using
the same mean lava flow thickness.
In the last 70 years, single-event erupted volumes are found to
have an average volume of 72 10
6
m
3
. The largest volume,
201.55 10
6
m
3
, was emitted during the 29 month-long eruption
in 1938–1940 and the smallest one was erupted in the 1956 erup-
tion that lasted only 2 days and was limited to the summit caldera
(1 10
6
m
3
).
Table 5 summarizes the date, location and estimated lava flow
area and volume. Fig. 4 compares the estimated volumes from
the present study with previous estimates from Burt et al.
(1994), Colclough (2005) and Kitagawa et al. (2007). Results from
the latter two studies, also based on remote sensing techniques,
do not differ by more than 10% from ours (see Fig. 4). Results from
Table 4
List of information visible on the new volcanological map of
Nyamulagira. Cones, fractures and lava fields come from remote
sensing and field data. The petrology comes from Thonnard et al.
(1965) and Aoki et al. (1985). Geographical layers come from
remote sensing and BEGo (2005).
Geographical information
Name Type
Main towns and localities Point
Road network Polyline
Urban areas Area
National boundaries Area
Boundaries of the Virunga National Park Area
Hydrographic network Polyline
Lakes Area
Contour lines (every 200 m) Polyline
Colored shaded relief Image
Volcanic fractures of the Nyamulagira and
Nyiragongo volcanic fields
Polyline
Adventive cones of Nyamulagira and
Nyiragongo
Area
Nyamulagira’s lava flows of 1938–2006
eruptions
Area
Nyiragongo’s lava flows of 1977 and 2002
eruptions
Area
Petrography of lavas from Nyamulagira Point
B. Smets et al. /Journal of African Earth Sciences 58 (2010) 778–786 783
Author's personal copy
Burt et al. (1994) show larger discrepancies (i.e. 10–30%) with our
estimated volumes. This can probably be explained by the fact that
Burt et al. (1994) used data from previous works based on the 1965
map and did not have the advantage of modern techniques such as
differential GPS or high spatial resolution spaceborne remote
sensing.
4.2. Impact of lava flows on VNP and human infrastructures
The GIS database enables the cross-correlation of mapped lava
flows with other geographic data. For instance, we observe that
the total Nyamulagira lava field (mapped using Landsat images)
covers at least 1117 km
2
, more than 85% of which are within the
VNP (see Fig. 1). The multiple lava flows from the 25 mapped erup-
tions that occurred between 1938 and 2006 occupy 40% of the to-
tal lava field, which attests to the rapid re-coverage of the area by
new lavas. Table 6 summarizes the surfaces and proportion of fea-
tures of natural or human interest (roads, park or urban areas) that
are covered or within the lava field of Nyamulagira and Nyiragon-
go. These observations of past events illustrate the lava flow hazard
and its recurrence in association with volcanic activity in the VNP.
The fact that Nyamulagira caused damages to crops, livestock and
Table 5
Summary of the eruptive history of Nyamulagira, from 1938 to present. The dates come from Burt et al. (1994), Smithsonian Institution (1996–2009), Bluth and Carn (2008) and
field observations for the last eruption.
Start date End date General location Cone name Lava flow
area (10
6
m
2
)
±Error Lava flow
volume (10
6
m
3
)
±Error
28/01/1938 25/06/1940 or after Summit, SE & SW flanks Tshambene 67.18 4.76 201.55 28.54
1/03/1948 15/07/1948 or after SW flank Gituro and Muhoboli 18.95 1.29 56.86 7.75
16/11/1951 16/01/1952 NNW fissure zone Shabubembe and Ndakaza 5.81 0.63 17.42 3.80
21/02/1954 28/05/1954 SSE fissure zone Mihaga 22
a
66
a
17/11/1956 18/11/1956 Summit caldera 0.33 0.06 1.00 0.37
28/12/1957 29/12/1957 Summit caldera & SSE fissure zone 1.82 0.37 5.47 2.24
7/08/1958 21/11/1958 N flank Kitsimbanyi 47
a
141
a
23/04/1967 3/05/1967 N flank Gakararanga 32
a
95
a
24/03/1971 28/04/1971 or after WNW flank Rugarama 30
a
90
a
23/12/1976 April 1977 SSW flank Murara and Harakandi 21
a
62
a
30/01/1980 23/02/1980 N flank Gasenyi 22.31 1.73 66.94 10.37
25/12/1981 14/01/1982 SE flank Rugarambiro 41.89 2.74 125.67 16.47
23/02/1984 14/03/1984 NW flank Kivandimwe 22.48 1.46 67.45 8.76
16/07/1986 20/08/1986 South flank 18.41 1.37 55.22 8.20
30/12/1987 4/01/1988 N flank Gafuranindi 2.05 0.35 6.14 2.09
23/04/1989 After 4/08/1989 Summit, SE & E flanks 31.74 1.99 95.22 11.94
20/09/1991 8/02/1993 NE flank Mikombe and other 43.68 1.82 131.04 10.90
4/07/1994 24/07/1994 West flank Kimera 14.32 0.92 42.97 5.49
1/12/1996 16/12/1996 or after N flank 16.58 2.83 49.75 8.49
17/10/1998 29/10/1998 Summit & NW flank 23.06 3.49 69.18 10.48
27/01/2000 11/02/2000 or after SE flank Ngerageze 15.77 2.13 47.30 6.40
6/02/2001 5/04/2001 or after N & SSE flanks 46.71 3.78 140.14 11.33
25/07/2002 Between 9/08 and 27/09/2002 Summit caldera, N & S flanks 18.89 1.45 56.67 8.69
8/05/2004 28/05/2004 or after Summit & NNW flank 22.83 1.97 68.48 11.84
27/11/2006 5/12/2006 or after South flank Ushindi 14.73 1.13 44.19 3.39
2/01/2010 27/01/2010 SE flank 17.65 5.98 52.96 17.94
a
Volumes from Burt et al. (1994), not calculated in this study.
Fig. 4. Histogram comparing the lava flow volume estimates from Burt et al. (1994) in black, Colclough (2005) in light grey, Kitagawa et al. (2007) in dark grey and from this
study in white. The volume estimates calculated in this study are in close agreement with those from former studies by Colclough (2005) and Kitagawa et al. (2007). Relative
differences between this study and these previous estimates for the 1996, 1998 and 2001 flow volume are 9.5%, 2.6% and 5.3%, respectively. Results from Burt et al. show
slightly lower agreement (see text).
784 B. Smets et al. /Journal of African Earth Sciences 58 (2010) 778–786
Author's personal copy
infrastructure at least nine times since 1938 (Smithsonian Institu-
tion, 1996–2009) highlights that this volcano represents a direct
hazard that must be taken into account for a safe and sustainable
development of the area. Concerning Nyiragongo, the present re-
sults are also consistent with the hazard assessments by Favalli
et al. (2009) and Chirico et al. (2009), which suggest that Nyiragon-
go eruptions can lead recurrently to fatalities and substantial infra-
structure damage, particularly in the densely populated city of
Goma and its environs (see Table 6).
5. Conclusions
The combined use of optical and radar remote sensing tech-
niques allowed us to successfully map the lava flows of the last
26 eruptions of the Nyamulagira. Integrated in a GIS database with
complementary datasets, a new thoroughly updated volcanological
map is now available and easily updatable and adaptable to the
needs of different users. The lava flow mapping was urgently
needed for scientific studies as well as for enhanced risk, environ-
mental and humanitarian management. As examples, accurate sur-
face estimation of lava-covered areas allowed us to calculate lava
volumes for successive eruptive phases and provide us with essen-
tial information for hazard assessment. It shows in particular that
Nyamulagira poses a lava flow hazard in the region, especially for
the protected forests of the Virunga National Park. Although most
of the lavas flow toward the northern uninhabited regions, the
Nyamulagira’s lavas also potentially threaten several tens of thou-
sand people in the southern region of Sake. Indirect hazards should
also be considered in addition to the lava flow hazard. Contamina-
tion by gas and ash as well as loss of crops and livestock can ma-
jorly affect the larger population living up to several hundreds of
kilometres away from the lava fields.
These new remote sensing results must, however, be validated
in the field, which is currently prevented by social and political un-
rest of the region.
5.1. Note added in proof
As mentioned in the present work Nyamulagira is the most ac-
tive volcano in Africa. It actually erupted in January 2010 during
the reviewing procedure of the current paper, leading us to add a
preliminary mapping of its last lava flows in Fig. 3 and update
Tables 5 and 6. The preliminary mapping of the new lava flows
was performed using the TIR bands of an ASTER image taken on
29th January 2010. This mapping is less accurate than for previous
eruptions due to the lower spatial resolution (i.e. 90 m) of the TIR
bands used and the possible effect of cloud cover on some pixel
values.
Acknowledgements
The current study was carried out in the frame of the projects
‘‘SAMAAV: Study and Monitoring of Active African Volcanoes” (funded
by the Belgian Federal Science Policy) and ‘‘GORISK: A multi-ap-
proach tool for the volcanic risk management of the Goma region
(North Kivu)” (funded by the Belgian Federal Science Policy (SR/
00/113) and the National Research Fund of Luxembourg (FNR/STE-
REOII/06/01)). SAR data are provided in the frame of the European
Space Agency (ESA) Category-1 Project No. 3224 and ESA-JAXA
(Japanese Space Agency) Project No. 3690. Precise orbits are pro-
vided by the Delft Institute of Earth Observation and Space Systems
(DEOS) and ESA. Interferograms are computed with the DORIS
(TUDelft) and the ROI-PAC (JPL/Caltech) open source softwares.
SRTM digital elevation models are provided by the US Geological
Survey. We thank J. Lavreau, F. Kervyn and M. Kervyn for helpful
comments. We also thank S.A. Carn and G.G.J. Ernst for their review
of the manuscript.
References
Aoki, K.-I., Yoshida, T., 1983. Petrological and geochemical studies on the 1981–
1982 lava from Nyamuragira volcano. In: Hamaguchi, H. (Ed.), Volcanoes
Nyiragongo and Nyamuragira: Geophysical Aspects. The Faculty of Science,
Tôhoku University, Sendai, Japan, pp. 91–96.
Aoki, K.-I., Yoshida, T., Yusa, K., Nakamura, Y., 1985. Petrology and geochemistry of
the Nyamuragira volcano, Zaire. Journal of Volcanology and Geothermal
Research 25 (1–2), 1–28.
BEGo (Synoptics, Keyobs, Royal Museum for Central Africa, Catholic University of
Louvain), 2005. Parc National des Virunga, République Démocratique du
Congo, 1/260,000. Map realized in the framework of the BEGo (Building
Environment for the Gorilla’s) project, co-financed by ESA and UNESCO-WHC,
Belgium.
Bluth, G.J.S., Carn, S.A., 2008. Exceptional sulfur degassing from Nyamuragira
volcano, 1979–2005. International Journal of Remote Sensing 29 (22), 6667–
6685.
Brousse, R., Cocheme, J.-J., Pottier, Y., Vellutini, P.-J., 1979. Eruption et nature de la
lave du Murara: un nouveau volcan (décembre 1976–avril 1977), au Kivu
(Zaïre). Comptes rendus des séances de l’Académie des Sciences 289 (Série D),
809–812.
Brousse, R., Caron, J.-P.H., Kampunzu, A.B., Lubala, R.T., Musengie, M.K., Vellutini,
P.-J., 1981. Eruption et nature de la lave du Gasenyi: un nouveau volcan
(janvier–février 1980) au flanc Nord du Nyamulagira (Kivu, Zaïre). Comptes
rendus des séances de l’Académie des Sciences 292 (Série II), 1413–1416.
Brousse, R., Caron, J.-P.H., Kampunzu, A.B., Lubala, R.T., 1983. Le Rugarambiro:
nouveau cône adventif du Nyamulagira actif de décembre 1981 à janvier 1982
(rift valley d’Afrique Centrale, Kivu). Comptes rendus des séances de l’Académie
des Sciences 296 (Série II), 813–818.
Burt, M.L., Wadge, G., Scott, W.A., 1994. Simple stochastic modelling of the eruption
history of a basaltic volcano: Nyamuragira, Zaire. Bulletin of Volcanology 56,
87–97.
Carn, S.A., 1999. Application of synthetic aperture radar (SAR) imagery to volcano
mapping in the humid tropics: a case study in East Java, Indonesia. Bulletin of
Volcanology 61, 92–105.
Caron, J.-P.H., Kampunzu, A.B., Lubala, R.T., Miruho, K.B., Musengie, M.K., Vellutini,
P.-J., Brousse, R., 1982. A propos de l’éruption adventive du Nyamulagira de
janvier-février 1980: nature de la lave du nouveau Gasenyi. Annales de la
Société Géologique de Belgique 105, 31–40.
Chirico, G.D., Favalli, M., Papale, P., Boschi, E., Pareshi, M.T., Mamou-Mani, A., 2009.
Lava flow hazard at Nyiragongo Volcano, DRC: 2. Hazard reduction in urban
areas. Bulletin of Volcanology 71 (4), 375–387.
Colclough, S.J., 2005. Investigations of Nyamuragira and Nyiragongo Volcanoes
(DRC), using Interferometric Synthetic Aperture Radar. Ph.D. Thesis, University
of Cambridge, United Kingdom.
De La Crùz-Reyna, S., Meli, P.R., Quaas, W.R., 2000. Volcanic crises management. In:
Encyclopedia of Volcanoes. Academic Press, pp. 1199–1214.
d’Oreye, N., Kervyn, F., Calais, E., Cayol, V., Fernández, J., Frischknecht, C., Gonzales,
P., Heleno, S., Oyen, A., Wauthier, C., 2008. Systematic InSAR monitoring of
African active volcanic zones: what we have learned in three years, or a harvest
beyond our expectations. In: Proceedings of 2008 Second Workshop on USE of
Remote Sensing Techniques for Monitoring Volcanoes and Seismogenic Areas
Table 6
Example of information resulting from collected data and GIS processing. Results directly depend on the accuracy of collected data. According to them, Nyamulagira lava flow
activity threatens important roads, the town of Sake and at least 12% of the Virunga National Park. Similar GIS computation performed for Nyiragongo shows that the latter is
mainly hazardous for urban areas, especially for Goma, rather than for the Virunga National park.
Virunga National Park Roads (km) Urban areas (km
2
) Goma Sake
Nyamulagira lava field 954.2 km
2
(12% of the park) 72.9 4.2 – 2 km
2
(93% of the town)
1938–2010 Lava flows from Nyamulagira 437.2 km
2
(5.7% of the park) 8.4 0.5 – 0.5 km
2
(22% of the town)
Nyiragongo lava field 165.5 km
2
(2% of the park) 239.2 39 37 km
2
(100% of the city) –
1977 Lava flows from Nyiragongo 7.4 km
2
(0.1% of the park) 17.4 1.6 –
2002 Lava flows from Nyiragongo 3.8 km
2
(0.05% of the park) 11 3.7 3.7 km
2
(10% of the city) –
B. Smets et al. /Journal of African Earth Sciences 58 (2010) 778–786 785
Author's personal copy
(USEReST 2008), Naples, Italy, November 11–14, 2008, 1-4244-2547-1/08/
$20.00 Ó2008 IEEE, pp. 57–62.
Ernst, G.G.L., Kervyn, M., Teeuw, R.M., 2008. Advances in the remote sensing of
volcanic activity and hazards, with special consideration to applications in
developing countries. International Journal of Remote Sensing 29 (22), 6687–
6723.
Favalli, M., Chirico, G.D., Papale, P., Pareshi, M.T., Boschi, E., 2009. Lava flow hazard
at Nyiragongo Volcano, DRC: 1. Model calibration and hazard mapping. Bulletin
of Volcanology 71 (4), 363–374.
Francis, P.W., Baker, M.C.W., 1978. Sources of two large volume ignimbrites in the
Central Andes: some landsat evidence. Journal of Volcanology and Geothermal
Research 4, 81–87.
Francis, P.W., Wells, G.L., 1988. Landsat Thematic Mapper observations of debris
avalanches deposits in the Central Andes. Bulletin of Volcanology 50, 258–278.
Hamaguchi, H., 1983. Volcanoes Nyiragongo and Nyamuragira: Geophysical
Aspects. The Faculty of Science, Tôhoku University, Sendai, Japan. 130 p.
Hanssen, R.F., 2001. Radar Interferometry – Data Interpretation and Error Analysis.
Kluwer Academic Publishers, The Netherlands. 171 p.
Kampunzu, M.M.A.B., Sebagenzi, M.N., Caron, J.P.H., Vellutini, P., 1982. Comparaison
pétrochimique entre les laves du champ Sud-Kivu (Bukavu) et du Nord-Kivu
(Virunga), Zaïre. Annales de la Société Géologique de Belgique 105, 25–29.
Kervyn, F., 2001. Modelling topography with SAR interferometry: illustrations of a
favourable and less favourable environment. Computers & Geosciences 27,
1039–1050.
Kervyn, F., d’Oreye, N., Fernandez, J., the SAMAAV team, 2005. The SAMAAV project:
study and analysis and monitoring of active African volcanoes using ERS and
ENVISAT data. In: Proceedings of the FRINGE 2005 Workshop, Frascati, Italy.
Kervyn, M., Kervyn, F., Goossens, R., Rowland, S.K., Ernst, G.G.J., 2007. Mapping
volcanic terrain using high-resolution and 3D satellite remote sensing. In:
Teeuw, R.M. (Ed.), Mapping Hazardous Terrain using Remote Sensing.
Geological Society, Special Publication 283, London, pp. 5–30.
Kitagawa, S., Fukui, K., Takagi, A., 2007. The December 1996 eruption at Nyamuragira,
Africa, viewed from JERS-1 InSAR. Progress of Crustal and Deformation Research
by Space Geodesy and Remote-sensing Techniques, Kyoto University, Japan,
January 18–19, 2007.
Louant, B., 1987. Correction to July 1986 eruption site. Scientific Event Alert
Network 12:08.
Lu, Z., Rykhus, R., Masterlark, T., Dean, K.G., 2004. Mapping recent lava flows at
Westdahl Volcano, Alaska, using radar and optical satellite imagery. Remote
Sensing of Environment 91, 345–353.
Massonet, D., Feigl, K.L., 1998. Radar interferometry and its application to changes
in the Earth’s surface. Reviews of Geophysics 36 (4), 441–500.
Newhall, C.G., 2000. Volcano warnings. In: Sigurdsson, H. (Ed.), Encyclopedia of
Volcanoes. Academic Press, pp. 1185–1197.
Pottier, Y., 1978. Première éruption historique du Nyiragongo et manifestations
adventives simultanées du volcan Nyamulagira (chaîne des Virunga – Kivu –
Zaïre: Déc. 76–Juin 77). Musée Royal de l’Afrique Centrale, Tervuren
(Belgique), Département de Géologie et Minéralogie, Rapport Annuel 1977,
pp. 157–175.
Pouclet, A., 1976. Volcanologie du rift de l’Afrique Centrale, le Nyamuragira dans les
Virunga, essai de magmatologie du rift. Ph.D. Thesis, University of Paris-Sud
Orsay, France (in French).
Rowland, S.K., MacKay, M.E., Garbeil, H., Mouginis-Mark, P.J., 1999. Topographic
analyses of Kilauea Volcano, Hawaii, from interferometric airborne radar.
Bulletin of Volcanology 61, 1–14.
Smithsonian Institution, 1996–2009. Global Volcanism Program: Worldwide
Holocene Volcano and Eruption Information. Smithsonian Institution. <http://
www.volcano.si.edu>.
Thonnard, R.L.G., 1961. Utilisation de la photogéologie dans la chaîne des Virunga
(Congo). Bulletin de la Société belge de Géologie 70, 144–149.
Thonnard, R.L.G., Denaeyer, N.E., 1965. Carte volcanologique des Virunga (Afrique
centrale), introduction générale et notice explicative de la feuille no. 1. Centre
National de Volcanologie (Belgium), Publication no. 32, 14pp.
Thonnard, R.L.G., Denaeyer, N.E., Antun, P., 1965. Carte volcanologique des Virunga
(Afrique centrale), feuille no. 1, 1:50,000. Centre National de Volcanologie
(Belgium), Publication no. 32.
Ueki, 1983. Recent volcanism of Nyamuragira and Nyiragongo. In: Hamaguchi, H.
(Ed.), Volcanoes Nyiragongo and Nyamuragira: Geophysical Aspects. The
Faculty of Science, Tôhoku University, Sendai, Japan, pp. 7–18.
UNESCO, 1992–2009. UNESCO World Heritage Centre Website. United Nations.
<http://whc.unesco.org>.
van Overbeke, A.-C., Bagalwa, M., Durieux, J., Kavotha, D., Kies, A., Kervyn, F., Lukaya,
F., Mitangala, P., d’Oreye, N., Osodundu, E., Smets, B., Tedesco, D., Wauthier, C.,
Yalire, M., in press. Monitoring of volcanic activity in the Goma region (N-Kivu,
Democratic Republic of Congo) and mitigation of related risks by both
spaceborne and ground-based techniques: experience of the GORISK project.
In: Proceed. ECGS Cahiers Bleus – 26th ECGS Workshop on Active Volcanism &
Continental Rifting, with Special Focus on the Virunga (North Kivu, DR of
Congo).
Zana, N., 1984. Large lava flows and tephra ejection from flank fissure. Scientific
Event Alert Network 09:03.
Zana, N., Hody, H.-L., 1989. Inflation precedes fissure eruption; large lava flows.
Scientific Event Alert Network 14:05.
786 B. Smets et al. /Journal of African Earth Sciences 58 (2010) 778–786
