See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/4110680

Use of multispectral ASTER images for mapping debris-covered glaciers

within the GLIMS Project

Conference Paper · October 2004

DOI: 10.1109/IGARSS.2004.1368616 · Source: IEEE Xplore

CITATIONS READS

58 275

4 authors, including:

Roberto Ranzi Giovanna Grossi

Università degli Studi di Brescia Università degli Studi di Brescia
154 PUBLICATIONS   1,884 CITATIONS    51 PUBLICATIONS   616 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

FP7 KULTURisk - The benefits of risk prevention View project

Flood simulation and forecasting View project

All content following this page was uploaded by Roberto Ranzi on 03 June 2014.

The user has requested enhancement of the downloaded file.

 2004 IEEE International Geoscience and Remote Sensing Symposium, September 20-24, 2004 Anchorage, Alaska, Vol. II, 1144-1147

Use of multispectral ASTER images for mapping

debris-covered glaciers within the GLIMS Project
Roberto Ranzi, Giovanna Grossi, Laura Iacovelli, Stefan Taschner
Dept. of Civil Engineering
Università di Brescia
Brescia, Italy

Abstract—The problem of mapping debris-covered glaciers using
images derived from satellite-borne optical scanners is addressed
in this paper. Results using also TERRA-ASTER images on the
Belvedere and Miage Glaciers, both located in the Italian Alps,
are presented. Field measurements and energy balance modeling
indicate that for debris superimposed on ice, surface
temperatures of some degrees colder than for pure debris are to
be expected in the morning. This fact is also confirmed by
processing satellite images taking into account the thermal band.
The results of black glacier detection can be useful for the project
GLIMS (Global Land Ice Monitoring from Space) when debris-
covered glaciers are to be mapped.
Keywords: TERRA-ASTER; debris; glaciers; GLIMS; ice
I. INTRODUCTION
Monitoring the cryosphere is an important component of
the Earth’s observation since glaciers are very sensitive to
climate fluctuations. Their changes in size and physical
properties (albedo, debris cover, ice velocity,…) reflect
changes in air temperature, precipitation and geomorphology.
An inventory of glaciers updated on a regular basis (ten to
twenty years) can provide an useful information to scientists
and policy decision makers tackling with the problems raised
by fluctuations, or changes, of the Earth’s climate. In this sense
one of the most ambitious ongoing research projects in glacier
monitoring is the international project GLIMS (see also
www.glims.org), coordinated by the USGS. First outcomes of
this project are described, for instance, in [5]. The main
objective of this project is to establish a new global inventory
of land ice based on a homogeneous source of information, i.e.
optical multispectral satellite images as LANDSAT and
TERRA-ASTER. Satellite images are particularly useful to
monitor the cryosphere, as shown in [1], [2] or [7], because
land ice is mostly located in areas where field or aerial surveys
are difficult or expensive. However the effectiveness of
glaciers-classification algorithms depends also on the specific
geomorphological features of the region to be investigated, as
discussed in [6]. For instance, monitoring glaciers using remote
sensing techniques is difficult in those areas, as the southern
European Alps, where a large number of glaciers (the so called
‘black glaciers’) are partly covered by a debris layer that
prevents an appropriate use of remote sensing data, especially
in the visible and near infrared range of the spectrum.
The research team of University of Brescia is involved in
the GLIMS project, playing the role of coordinators of the
mapping activity for the glaciers located in the Italian Alps.
The status of the research can be browsed at
http://civserv.ing.unibs.it/utenti/ranzi/glimsit/index.htm.
One of the Italian test sites for studying the physics of
‘black’ glaciers is the Belvedere Glacier (45.94° N; 7.91° E),
located in the Anzasca watershed at the base of the Monte Rosa
peak (4634 m asl.), close to the Swiss border. The overall
glacierized area is 5.58 km² as reported in the World Glacier
Inventory. The glacier’s terminus reaches an altitude of only
1785 m asl., quite unusual for the southern side of the
European Alps. The Belvedere Glacier is of special interest
also from another point of view. Since 2001 every year a small
epiglacial lake, named Lago Effimero, developed with greater
extension and volume than in the past on the flat glacier’s
tongue [4]. It was well detected by the TERRA-ASTER
scanner, as reported, for instance, in [9]. This lake was a
serious risk in terms of a possible outburst and formation of a
flood wave and was therefore monitored in-situ with special
care by civil protection authorities. To better monitor this
glacier, field measurements were conducted in summer 2003
aiming at measuring incoming and net radiation and thermal
properties of a shallow debris layer superimposed on ice, as
described in the following. The results on the use of visible and
thermal information derived from TERRA-ASTER images for
the detection of this glacier are presented together with an
application to the Miage debris-covered glacier (45° 47’ N, 6°
52’ E), located in the Monte Bianco/Mont Blanc massif, in the
third section. In Fig.1. the location of both glaciers is reported.
Figure 1. Geographical location, in the Italian Alps, of the two debris-
covered glaciers mapped using ASTER images.

ISBN CD-ROM 0-7803-8743-0 (C) 2004 IEEE

 2004 IEEE International Geoscience and Remote Sensing Symposium, September 20-24, 2004 Anchorage, Alaska, Vol. II, 1144-1147
II. FIELD MEASUREMENTS AND MODELLING
In the frame of the GLIMS activities 2003 an energy
balance station was installed on July 29 on the main glacier
tongue of the Belvedere Glacier at an altitude of 1880 m asl.
and remained there till October 15. Besides standard
meteorological measurements of air temperature, humidity,
wind speed and direction, global shortwave and net all-wave
radiation, also the heat flux and the temperature of the debris at
the surface and at 3 different depths down to the underlying ice
were recorded on a hourly basis. Fig. 2 shows the installation
of the monitoring station. The energy supply was provided by a
20 W solar panel and the quality and continuity of the data
acquisition was controlled remotely on a regular basis through
a modem connection. The debris cover depth at the station was
30 cm, quite close to the average debris layer depth for the
whole Belvedere Glacier, as confirmed by the extensive field
surveys conducted by [3] and by the Brescia research team.
For black glacier detection using satellite images the
thermal band is important to distinguish between debris
superimposed on ice and pure debris or rocks. Visible or near
infrared images are almost useless in this case. A shallow
debris layer, with a depth less than 30-40 cm is such that its
surface temperature is influenced by an almost constant
temperature boundary condition at 0°C (during the melt season,
at least) at the underlying ice surface. As a consequence the
surface temperature is colder than for debris not superimposed
on ice. Field surveys conducted in summer 2003 indicated an
average temperature 4.5°C (± 3 °C) colder for debris layers
with depth ranging from 5 to 30 cm over ice than in the debris
deposits located at the same altitude. These measurements were
conducted in morning hours and are representative of the
surface thermal conditions at the time of the passage of the
TERRA-ASTER and LANDSAT satellites. The resulting
thermal signature can be detected by most of the satellite-borne
scanners, as the LANDSAT-ETM+ or the TERRA-ASTER.
In areas where no field verification is available, energy
balance modeling can help to estimate surface temperature
thresholds which are expected for debris-covered glaciers. For
Figure 2. Energy balance station, measuring meteorological variables and
debris-on ice thermal properties from July 29 – October 15 2003 on the main
glacier tongue of the Belvedere Glacier at an altitude of about 1880 m asl. On
the left the Little Ice Age lateral moraine can be seen.
ISBN CD-ROM 0-7803-8743-0 (C) 2004 IEEE
the purpose of verifying this approach an energy-balance
model was applied to the basin monitored by the Pecceto
hydrometric station, gauging an area of 43.5 km2, including a
total glacierized area of 10.8 km2. The model was also used to
simulate the runoff inflow to Lago Effimero, draining a surface
of 9.7 km2 with 4.7 km2 of glaciers. The energy balance model
used albedo estimates of snow, ice and the surrounding terrain
derived from a TERRA-ASTER image dating from July 19
2002 (see next section). The verification of the model was
conducted by comparing simulations and measurements of:
- temperatures and conductive heat fluxes in the debris
cover at the energy balance station (only for the melt
period 2003);
- hourly records of simulated runoff and river levels at the
Pecceto hydrometric station (melt period 2002 and
2003);
- cumulated inflow volumes at the Lago Effimero (melt
period 2002 and 2003).
One of the results of the comparison at the energy balance
station is reported in Fig.3 where the cumulated ice melt
simulated by the model is compared with the observed one in
year 2003, that was derived from the measurement of thermal
flux in the debris layer at the energy balance station. The
agreement is surprisingly good confirming the correct setup of
the key model’s parameters after field and laboratory
measurements of the physical properties of the debris. Instead
the amplitude of daily fluctuations of the debris temperature is
underestimated by a factor of 1.5÷2 during the monitoring
period. The simulated net radiation is in good agreement with
point measurements.
The cumulated inflow volumes at Lago Effimero in the
snowmelt season starting with April 1st 2003 is modeled with
about 2.7 x106 m3 (276 mm) until June 18 2003, when the Lake
suddenly emptied after the formation of a subglacial channel
system on its northern side. This value is in agreement with the
lake’s volume estimated on the basis of the surveys conducted
during the emergency. For the April 1st -June 30 2002 period
the simulated cumulated inflow is 3.4 x106 m3 (350 mm), a
value also in accordance with the estimates of the civil
protection authorities who started, at that time, to artificially
empty the lake with pumping stations.
Figure 3. Cumulated melt - modelled and measured for the location of the
energy balance station installed on the Belvedere Glacier from the end of July
til the end of September 2003.
 2004 IEEE International Geoscience and Remote Sensing Symposium, September 20-24, 2004 Anchorage, Alaska, Vol. II, 1144-1147
III. ASTER IMAGE PROCESSING
The mutual synergy between energy-balance modeling and
remote sensing will be briefly described in this section.
Radiance at sensor maps of the July 19 2002 TERRA-ASTER
bands VNIR1, 2 and 3, were used to estimate the albedo in the
investigated area. Point global radiation measurements at
different altitudes, a radiation model including the effect of
shadows, reflected radiation, terrain view factor, and a
simplified two-paths atmosphere attenuation scheme were used
to derive the albedo map reported in Fig. 4. For the debris
cover an average value of 0.2 is estimated, that is also
confirmed by point measurements. The albedo map was
resampled to the 100 m spatial resolution of the snow and ice-
melt model GIS. It is one of the key input components of the
model, since net shortwave radiation is the dominant term of
the energy balance in the investigated area.
The field measurements and the energy balance model
results, on the other side, were used for a better interpretation
of remote sensing data and for interpreting the thermal
signature of debris-covered glaciers. A GIS based classification
algorithm for black glacier detection was developed taking into
account the thermal band of the TERRA-ASTER or the
LANDSAT-ETM+ radiometers. The method is mainly based
on an extended band ratio algorithm, as described in [9] where
the results achieved in the detection of the Belvedere Glacier
outlines are discussed in detail.
For glacier ice detection with LANDSAT ETM+ data the
band ratio TM4/TM5 is often utilized [6]. However this
method could also be applied for TERRA-ASTER images by
using the ratio VNIR3/SWIR4. Both ratios are effective in
filtering slope, aspect and shadow effects and in discriminating
snow from clouds. This band ratio is however insufficient to
recognize debris superimposed on ice from rocks or alluvial
debris fans. For this purpose the use of the thermal information
of the ASTER sensor (3 bands in the spectral range from 8.1 -
9.3 µm and 2 bands from 10.2 – 11.6 µm) is introduced. After
excluding vegetated areas by the help of a derived NDVI
(Normalized Digital Vegetation Index) map, the thermal band
is applied by taking into account the fact that colder
temperatures have to be expected for debris superimposed over
glacier ice.
Figure 4. Albedo map derived from TERRA-ASTER image dating from July
19 2002. The limits of the Belvedere Glacier, as derived by [9] applying the
extended band ratio algorithm are representd in red.
ISBN CD-ROM 0-7803-8743-0 (C) 2004 IEEE
Thus a threshold value can be found that narrows the
fraction of candidated “icecored debris” pixels. Further low
pass and contextual filtering [2] is applied considering the
already detected glacierised areas.
To test the procedure the algorithm was also applied to
classify the Miage Glacier with a TERRA-ASTER image
dating from July 2nd 2000. The Miage Glacier is located in the
Italian Alps in the Monte Bianco/Mont Blanc group and is
comparable to the Belvedere Glacier in terms of several
geomorphological characteristics. The main tongue, dividing in
the lower part in two large lobes, is debris covered below an
altitude of 2500 m asl. and terminates at 1775 m asl. (Fig. 5).
Near the terminus, the debris thickness exceeds in general 0.5
m and is often more than 1 m deep [8] [10]. Moreover this zone
of thick debris cover is also characterized by sparse vegetation,
particularly small trees (fir and larch). Another characteristic of
interest is Lago del Miage, a small glacial lake located between
the glacier tongue and the lateral moraine. It is well known for
its ice cliff.
The debris covered part of the Miage Glacier cannot be
discriminated, using the visible bands only, from e.g. a debris
fan. Thus the glacier’s limits classification would be difficult.
However applying the extended band ratio algorithm the
glacier’s recognition is more feasible. The debris covered part
of the Miage Glacier, in fact, exhibits surface temperatures
which are correlated to the debris depth. Close to the glacier’s
termini (black circle), where the debris layer is often more than
1m deep, surface temperatures are warmer than those observed
close to the transient snowline, but in general still colder than
on the debris alluvial deposits on the left side of Fig. 6 (black
oval). Colder temperatures of Lago del Miage are also evident
in Fig. 6 (dark blue circle), indicating these water bodies as a
suitable control target for the calibration of the algorithm. The
glaciers boundaries mapped in Fig. 6 (thin red line) are
consistent with those derived after field surveys (light blue
line), apart from very near end of the termini (debris cover
more than 1 m)
Figure 5. The Miage Glacier (TERR-ASTER July 2nd 2000, RGB band
3,2,1). The debris-covered tongue of the glacier (right), Lago del Miage
(center), and a debris fan (left) are indicated by the white arrows.
 2004 IEEE International Geoscience and Remote Sensing Symposium, September 20-24, 2004 Anchorage, Alaska, Vol. II, 1144-1147
Figure 6. Thermal signature of the Miage Glacier as measured by TERRA-
ASTER band 14 (July 2nd 2000). Indicated in light blue are the glacier’s
outlines due to field surveys while the thin red line represents the result of the
glacier classification by means of remote sensing. Shown in dark blue is the
location of Lago del Miage and in black the debirs coverd glacier (right) and
an alluvial fan (left).
IV. CONCLUSIONS
Field measurements and energy balance modeling simulating
the 2002 and 2003 melt seasons for the Belvedere Glacier show
that around 10-12 UTC the surface temperatures of a debris
layer, some tens of centimeters deep, superimposed on a
glacier’s ice are 4.5° C colder, on average, than for other debris
deposits. This enables the identification of “black” glaciers by
satellite-borne thermal radiometers, at least when the debris
layer is less thick than 40-50 cm.
For the Belvedere Glacier albedo maps were derived from a
TERRA-ASTER image, useful to simulate the cumulated
inflow to the “Lago Effimero” epiglacial lake which caused
alert to the civil protection authorities in summer 2002 and
2003 for the possible formation of a dangerous flood wave. The
cumulated melt estimated for the basin upstream the lake from
April 1st was 350 and 276 mm, respectively, until June 30 2002
and June 18 2003, a value in agreement, for both years, with
the volumes of the lake measured after surveys.
A further verification of the energy-balance model
implemented confirmed a very close agreement between
simulated and measured cumulated melt (320 mm for the
August to September 2003 months) at an energy balance
station installed on the debris-covered glacier.
After these encouraging results the surface temperature
differences for debris deposits superimposed over ice and over
ground were used for mapping the glacier’s margins of the
Belvedere and the Miage Glacier, in the Italian Alps. For the
Miage Glacier one of the very first images acquired over the
Alps and useful for glaciological purposes, the July 2nd 2000
ISBN CD-ROM 0-7803-8743-0 (C) 2004 IEEE
View publication stats
TERRA-ASTER image, was processed, showing the capability
and limitations in discriminating the two debris-covered lobes
of the glacier from other surfaces.
These results were obtained for a glacierized region which
is quite easily accessible for field verification. They confirm
the potentials of remote sensing for mapping black glaciers
with similar geomorphological features worldwide, thus
contributing to the ongoing GLIMS project aiming at
producing an inventory of the Earth’s land ice.
ACKNOWLEDGMENT
This study was funded by the MURST 60% - Glaciers
monitoring grant and the research contract CNR/ASI ARS
1/R/192/02 sponsored by the Italian Space Agency through
CNR-IREA, under the supervision of Anna Rampini. Claudio
Smiraglia is thanked for having shared his experience during
the field surveys on the Belvedere Glacier. USGS, NSIDC and
the EDC are thanked for having made the TERRA-ASTER
images available within the Project GLIMS-Global Land Ice
Measurements from Space. Part of the meteorological data
were provided by Regione Piemonte.
REFERENCES
[1] K.J., Bayr, D.K., Hall, and W.M, Kovalick, “Observation on glaciers in
the Eastern Austria Alps using satellite data”. International Journal of
Remote Sensing, 1994, 15, pp. 1733-1742.
[2] E., Binaghi, P., Madella, M. P., Montesano and A. Rampini, “Fuzzy
contextual classification of multisource remote sensing images”, IEEE
Trans. on Geoscience and Remote Sensing, 1997, GE - 35 (2), pp. 326 -
339.
[3] G. Diolaiuti, C. D'Agata and C. Smiraglia, ”Variations in Belvedere
Glacier (Monte Rosa, Italian Alps) tongue thickness and volume in the
second half of the 20th century,” Arctic, Antarctic and Alpine Research,
vol. 35 (2), 2003, pp. 255-263.
[4] W., Haeberli, A., Kääb, F., Paul, G., Mortara, M., Chiarle and A.,
Mazza, “A surge-type movement at Ghiacciaio del Belvedere and a
developing slope instability at the east face of Monte Rosa,” Norsk
Geografisk Tidskrift, vol. 56 (2), 2002, pp.104-111.
[5] J. S. Kargel, H. Kieffer, R. Barry, M. Bishop, D. MacKinnon, K.
Mullins, B. Raup, G. Scharfen, J. Shroder, and R. Wessels, "Initial
Glacier Images from ASTER and Test Analysis for GLIMS," Eos
Transactions, American Geophysical Union, vol. 81, no. 48, 2000,
supplement.
[6] F. Paul, C. Huggel, and A. Kääb “Combining satellite multispectral
image data and a digital elevation model for mapping of debris-covered
glaciers,” Remote Sensing of Environment, vol. 89 (4), 2004, pp. 510-
518
[7] A. Rampini, P. A. Brivio, F. Rota Nodari and E. Binaghi, “Mapping
Alpine Glacier Changes from Space,” Proc. IGARSS 2002, IEEE, ISBN
0-7803-7537-8, Toronto, 2002, 3 pp.
[8] C. Smiraglia, G. Diolaiuti, D. Casati and M.P. Kirkbride, “Recent areal
and altimetric variations of Miage Glacier (Monte Bianco massif, Italian
Alps),” IAHS Pub. 264, 2000, pp. 227-233.
[9] S. Taschner and R. Ranzi, “Comparing the Opportunities of LANDSAT-
TM and ASTER Data for Monitoring a Debris Covered Glacier in the
Italian Alps within the GLIMS Project,” Proc. IGARSS 2002, IEEE,
ISBN 0-7803-7537-8, Toronto, 2002,, 3 pp.
[10] M.H. Thomson, M. Kirkbride and B.W. Brock, “Twentieth century
surface elevation change of the Miage Glacier, Italian Alps,” IAHS Pub.
264, 2000, pp. 219-225.