Réf. Paul 2004 - T

Référence bibliographique complète
PAUL F. The new Swiss Glacier Inventory 2000 - Application of Remote Sensing and GIS. PhD Thesis, Department of Geography, University of Zurich, 2004, 198 p.

Glaciers, Swiss Alps, satellite data, GIS, DEM, evolution.

Organismes / Contact
Department of Geography, University of Zurich-Irchel, Winterthurerstr. 190, CH-8057 Zurich.
Dr. Frank Paul : fpaul@geo.unizh.ch ; Tel: ++41 1 635 5175

(1) - Paramètre(s) atmosphérique(s) modifié(s)
(2) - Elément(s) du milieu impacté(s)
(3) - Type(s) d'aléa impacté(s)
(3) - Sous-type(s) d'aléa

Pays / Zone
Massif / Secteur
Site(s) d'étude
Période(s) d'observation
Switzerland Alps       1973-1999

(1) - Modifications des paramètres atmosphériques

Informations complémentaires (données utilisées, méthode, scénarios, etc.)

(2) - Effets du changement climatique sur le milieu naturel
Since 1850, glaciers in the Alps have retreated more or less continuously (depending on size), interrupted by advance periods from 1920-1930 and 1970-1985 (Gross, 1987). From about 1970 to 1985 Alpine glaciers experienced a well-documented phase of intermittent advance, which included for a certain time 75% of all measured glaciers. Thereafter a strong decrease in glacier length and size was observed.

Set (I): Western part of Switzerland (1973-1998, including debris correction)
Glacier areas in 1973 and 1998 are tabulated for 713 glaciers and seven area. While mean values for each area class exhibit a dependence of relative area change on glacier size, the individual values express the strong increase in scatter with decreasing glacier size. The absolute loss of area is much higher for larger glaciers but glaciers larger than 5 km2 have increased their part of the entire glacier area by 4%. This is due to the large number of glaciers smaller 1 km2, which contribute 40% to the total area loss, although they cover only 15.5% of the 1973 area.

The mean size of all glaciers is decreasing from 1.42 to 1.23 km2, which is still two times higher than for the 2073 glaciers (mean size 0.63 km2). Thus, there is a bias towards larger glaciers in this selection, resulting in underestimation of the relative change in area. There is no dependence of glacier change from any of the time-independent 3D parameter variables (slope, aspect) as well as from the mean potential direct solar radiation for the months June to September or from the percentage of cast shadow cover during a day.

Set (II): Eastern part of Switzerland (1973-1999, including debris correction)
This data set is much smaller than for the western part. In comparison to set (I) no glacier is larger than 20 km2, only one glacier is larger than 10 km2, few glaciers are smaller than 0.1 km2 and comparably more glaciers with sizes between 0.1 and 0.5 km2 exist. The average glacier size in 1973 is now similar to the sample of the 2073 glaciers and average relative loss of area (-31%) is higher than for set (I). Glaciers smaller than 1 km2 are responsible for 58.6% of the total area loss, although they cover only 34% of the area.

Set (III): Western part of Switzerland (1973-1985-1992-1998, mostly debris free)
For this selection all glaciers from set (I) are evaluated again and most of the debris-covered (larger) glaciers are rejected. This increases the average loss for all 471 glaciers considered to -19.1%. The advance period of Alpine glaciers during the 1970s as documented from the annual length measurements (e.g., Patzelt, 1985), is evident. While the average change from 1973 to 1985 is only -1%, glaciers sized 5-10 km2 even show a gain in area (+1%). Also glacier sizes in the two smaller area classes remain nearly unchanged. Glaciers smaller than 1 km2 show a small loss in area.

The average area change from 1985 to 1998 is -18.2% (with respect to the 1985 area), a value also found in previous studies (Paul, 2002). The advance period shows relative change in glacier area between 1973 and 1985 vs. glacier size. Again, the increase in scatter towards smaller glaciers is most evident. For glaciers larger than 0.5 km2 the increase in area does not exceed +20%. Relative change in area vs. mean slope has been studied in order to verify an assumed steeper slope for glaciers with a gain of area. But such a dependence is not obvious. The relative decrease in glacier area is somewhat greater from 1985 to 1992 than from 1992 to 1998.

To obtain statistical data about changes in glacier length between 1973 and 1998, 683 out of the 713 glaciers from set (I) are selected. Most of the glaciers rejected had zero length in 1998. While the average of the relative retreat per length class increases with decreasing glacier length, the absolute retreat for glaciers shorter than 10 km is similar. Glaciers longer than 10 km retreat twice as much, at least on average. Glaciers shorter than 5 km account for 94% of the total retreat and the part of each length class on the entire length changes slightly towards the longer glaciers.

The plot for the relative change in glacier length vs. glacier length is similar to that for the area change. The dependence is a little weaker and scatter dominates for glaciers shorter than 2 km. There is no dependence of relative or absolute change in length from further 3D parameters. The relative change in area is usually larger than the corresponding length change. This is indeed an indication that change in glacier area is not restricted to the glacier tongue and glacier monitoring from space is especially useful in revealing changes in glacier area.

The change of selected glacier parameters from 1973 to 1998 is derived from the DEM25L2 with the respective glacier outlines. A shift of minimum elevation towards higher values with an average of +66 m and a standard deviation of 74.6 m has been observed. Some points are also below the identity line indicating that some glacier tongues in 1998 extend to lower elevations than in 1973. The glacier (or a part of its tongue) is still larger than in 1973, or it reveals an orthorectification or comparison error. On average, elevation range decreased by 97 m with a standard deviation of 100.5 m. Values of mean slope have increased since 1973 as most glaciers have been retreating to higher elevations with steeper topography.

For a selection of 683 glaciers the hypsography is calculated at 100m intervals using an AML script and a Fortran program. The largest absolute loss of glacier area can be found in the elevation range 2700 to 3000m. This is also the elevation range, where most of the glaciers are situated, resulting in a smaller relative change (about -20%) than for lower elevations (up to -40%). Only minimal change takes place above 3200m. By using linear interpolation the median elevation of the area distribution is calculated as 2908m in 1973 and 2935m in 1998, indicating a slight shift (+27 m) of glacier distribution towards higher elevations.

Sensibilité du milieu à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)
The latest assessment of glacier areas and topographic parameters is compiled in the 1973 Swiss glacier inventory, which is based on cartographic-planimetric analysis of specially-acquired aerial photographs. The new Swiss glacier inventory 2000 (SGI 2000) shows the possibilities and limitations of a glacier inventory from satellite data using GIS (Geographic Information System) technology in combination with a digital elevation model (DEM).

Landsat 5 TM data from 1998 and 1999 are used for the SGI 2000. The large area covered on a single TM-scene (185 km on each side) combined with the high-spatial resolution (30 by 30 m per pixel), favours this sensor for glacier monitoring. Glacier classification by means of thresholded TM4/TM5 ratio images is simple, robust, and accurate (the comparison with other methods or manually derived outlines on higher-resolution satellite imagery has revealed an accuracy of better than 3% for glaciers larger than 1 km2).

In order to facilitate GIS-based processing, the glacier inventories from 1973 and 1850 are digitized. Outlines from 1973 are also used to define glacier basins for intersection with the TM-derived glacier areas. 3D glacier parameters are obtained by fusion of glacier outlines with a DEM and DEM-derived products (such as slope or aspect).

The DEM is essential for all pre-processing tasks (e.g., orthorectification of satellite images, digital terrain modelling) as well as post-processing (obtaining 3D glacier parameters, visualization of glacier changes).

The GIS is used as the central tool for data processing and integration of various data formats (vector, raster, image), namely: (1) digitizing of glacier outlines, central flow-lines and glacier basins, (2) raster-vector conversion of TM-derived glacier maps and assignment of glacier basins and IDs, (3) calculation of 3D parameters from the DEM and storage in corresponding attribute tables, (4) 2D and 3D visual representation of glacier changes and (5) export of data tables for further computations (e.g., glacier change).

(3) - Effets du changement climatique sur l'aléa

Paramètre de l'aléa
Sensibilité du paramètres de l'aléa à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)

(4) - Remarques générales

(5) - Syntèses et préconisations
Analysis of glacier change for the Swiss Alps reveals the following major findings:
The relative loss of glacier area between 1973 and 1998/99 is 18% +/-3% (about 1/5) with an assumed corresponding volume loss of about 1/4 (scaled for all glaciers).
Relative change in area between 1985 and 1992 is similar to the period 1992 to 1998/99 with respect to the 1973 glacier size (about -10% in each period). Small changes of glacier size until 1985 (-1%).
The average decadal relative loss of area from 1985 to 1998/99 is about seven times higher than from 1850 to 1973 (the samples are not exactly identical).
Glaciers smaller than 1 km2 contribute about 40% to the total loss although they cover only 15% of the area.
The relative changes in glacier size are highly individual with a fair dependence on glacier size (increasing scatter towards smaller glaciers) and no correlation to other investigated parameters. Only a large number of glaciers from all size classes will give a representative evaluation of ongoing changes.
The primary causes for area changes in valley and mountain glaciers are: separation from formerly connected tributaries, emerging rock outcrops, and melting along the perimeter.
Shrinkage of small glaciers (glacieretes) is enhanced by disintegration, with an even faster melting of the smaller parts.

These observations can be interpreted as follows:
The -20% reduction of glacier area since 1985 is already in the range of the -30% expected in previous studies for the year 2025 (Haeberli et al., 2002).
The change is also much faster than observed in the historical record. In particular the not monitored small glaciers contribute to the area loss.
The highly variable glacier retreat characteristics (often associated with arbitrary changes in geometry) makes calculation of future glacier behaviour from numerical modelling fairly questionable, at least for the majority of all glaciers.
Although changes in glacier surface elevation cannot be measured with TM, the observed changes in glacier geometry indicate a massive down-wasting since 1985 rather than a dynamic response to a changed climate.
Further glacier retreat can be expected in the future, as a probable dynamic glacier reaction to the hot decade of the 1990s is still to come.

Références citées :

GROSS, G. (1987): Der Flächenverlust der Gletscher in Östereich 1850 - 1920 - 1969, Zeitschrift für Gletscherkunde und Glazialgeologie , 23 , 131- 141.

HAEBERLI, W., MAISCH, M. AND PAUL, F. (2002): Mountain glaciers in global climaterelated observation networks, WMO-Bulletin , 51 (1), 18-25.

PATZELT, G. (1985): The period of glacier advances in the alps, 1965 to 1980, Zeitschrift für Gletscherkunde und Glazialgeologie , 21 , 403-407.

PAUL, F. (2002): Combined technologies allow rapid analysis of glacier changes, EOS, Transactions, American Geophysical Union , 83 (23), 253, 260, 261.