1. Introduction

Mass changes of glaciers are considered to be a sensitive indicator for climatic variations (e.g. Reference HoughtonHoughton and others, 2001). Understanding the evolution and spatial variability of glaciers may contribute to better knowledge of impact, trends and rates of ongoing climate change. However, there are only a limited number of direct mass-balance observations available covering a relatively short period (SCNAT– ETHZ, 1881–2002; Reference Dyurgerov and MeierDyurgerov and Meier, 2005). The directly related information on volume changes, i.e. net balance (Reference PatersonPaterson, 1994, p. 29), provides a good opportunity to quantify glacier changes over a larger area and time-span.

Thickness or volume changes may be derived from repeated comparison of the surface topography of different years. Digital elevation models (DEMs) describing the glacier surface geometry can be extracted by applying modern remote-sensing methods (Reference Baltsavias, Favey, Bauder, Boesch and PaterakiBaltsavias and others, 2001) and evaluation of contour lines on topographic maps.

Most mountain glaciers and small ice caps showed significant mass loss during the second half of the 20th century (Reference Dyurgerov and MeierDyurgerov and Meier, 2000; Reference Ohmura, Sparling and HawkesworthOhmura, 2004). A change in trend was observed in the time series in the mid-1970s, followed by a more rapid loss of ice volume in the last two decades. Glacier retreat in the Swiss Alps since the end of the Little Ice Age around 1850 is estimated at 700km² in area and 50–60km³ in volume (Reference Müller, Caflisch and MüllerMüller and others, 1976; Reference Maisch, Wipf, Denneler, Battaglia and BenzMaisch and others, 1999; Reference Paul, Kääb, Maisch, Kellenberger and HaeberliPaul and others, 2004). Significant thickness changes have been reported from both mountain glaciers and polar ice sheets (Reference EchelmeyerEchelmeyer and others, 1996; Reference Arendt, Echelmeyer, Harrison, Lingle and ValentineArendt and others, 2002; Reference Rignot and ThomasRignot and Thomas, 2002; Reference Rignot, Rivera and CasassaRignot and others, 2003).

In the course of ongoing glacier monitoring in Switzerland (SCNAT–ETHZ, 1881–2002) glaciers have repeatedly been surveyed in detail, and high-resolution glacier maps have been produced (e.g. Reference MercantonMercanton, 1916; Reference Kasser and RöthlisbergerKasser and Röthlisberger, 1966). These investigations are complemented by the well-documented glacier surface geometry on historical maps (the so-called Siegfried maps) generated in the second half of the 19th century and the first complete survey of official national maps 1 : 50 000 in the first half of the 20th century.

Since the 1960s, aerial photographs have been taken at regular intervals in Switzerland in order to update the national maps. Additionally, high-resolution aerial photographs exist for selected glaciers, some at annual intervals (SCNAT–ETHZ, 1881–2002). Several projects carried out by the Versuchsanstalt für Wasserbau, Hydrologie und Glaziologie (VAW) of ETH Zürich in collaboration with public administrations and private companies have contributed substantially to the documentation of glacier surface geometry (many unpublished reports).

In this study, we present a compilation of ice volume changes for a selection of glaciers, with detailed information on the surface topography for five to six specific points in time within the last 100–150 years. The spatial distribution as well as temporal variations of the thickness change were analyzed. The main goal was to determine decadal net volume changes and to both identify and interpret differences on a regional scale in the Swiss Alps. Similar investigations have been conducted in the Austrian Alps for the three decades 1950–79 (Reference Finsterwalder and RentschFinsterwalder and Rentsch, 1981).

2. Methodology

The comparison of repeated surveys of glacier surface topography has traditionally been applied to evaluate mass balance indirectly (e.g. Reference HoinkesHoinkes, 1970; Reference PatersonPaterson, 1994). This method cannot be used to obtain mass balance of annual resolution and therefore replace in situ measurements (Reference Fountain, Jansson, Kaser and DyurgerovFountain and others, 1999; Reference BauderBauder, 2001). Glacier mapping has been mainly used to confirm results from direct mass-balance measurements (Reference HaakensenHaakensen, 1986; Reference Cox and MarchCox and March, 2004). Due to significant progress in remote-sensing technologies, DEM generation has become a standard tool.

We limited our investigations to available aerial photographs or special glacier maps with clearly identifiable production procedures and survey dates. In order to extend our time series further back in time, we resorted to older maps mainly from the Siegfried survey between 1860 and 1890. Where possible we used originals (so called plane-table sheets) from the field surveys.

3. Results

As one example of the 19 glaciers (see Table 1) that have been studied, the results of Grosser Aletschgletscher, the largest glacier in the Alps, are presented. We have chosen the sub-period 1980–99 to highlight the comparison of the two successive DEMs.

Table 1. Compilation of glaciers used in this study with area, evaluated period, number of individual DEMs, cumulated total ice-volume change and mean net balance

In Figure 1 the spatial distribution of the thickness change is shown. A typical pattern was found. Thickness changes are small in the accumulation area, and the largest changes occur at the glacier snout. In the ablation area, there is a gradual increase starting from the equilibrium line, cumulating at the current glacier snout and decreasing again toward the former snout. A much larger scatter is found in the accumulation area. The observed general pattern of thickness change is typical for both a retreating and advancing glacier.

Fig. 1. Spatial distribution of thickness change (in mice) of Grosser Aletschgletscher in the period 1980–99.

The evolution of the mean thickness change for the investigated 120 year period (1880–1999) shows a significant ice loss of >50m mean thickness (see Fig. 2). Also given in Figure 2 are the rates of mean thickness change for each sub-period.

Fig. 2. Evolution of cumulative thickness change (solid line) and rate of thickness change (dashed line) of Grosser Aletschgletscher.

Calculated decadal net balance rates of all investigated glaciers for each sub-period are contrasted in Figure 3. Varying lengths of sub-periods, with an increasing resolution in the last 30 years, are clearly visible. The general overall trend of negative rates is interrupted by one or two sub-periods of mass gain around 1960–85. Exceptions to this are the large glaciers (e.g. Grosser Aletschgletscher, Unteraargletscher) and glaciers where this specific time period is not sufficiently resolved by individual sub-periods. The corresponding values of total ice volume change and derived mean net balance can be found in Table 1.

Fig. 3. Decadal net balance rates. The vertical dashed lines indicate the periods since 1920 used for the analysis of mean rates.

The overall error in total volume change is difficult to estimate. There are three dominant contributions: (1) error in map survey, (2) accuracy of contour line evaluation or photogrammetric analysis, and (3) error due to interpolation on a regular grid. The first contribution is difficult to quantify, but limited analyses obtain values in the orders of meters to tens of meters depending on the quality of the maps. The second type of error is evaluated to be 0.3 m for the photogrammetric analysis and 0.5–2.5m depending on map scale, respectively. The third type of error is the smallest at 0.05 m, but detailed analysis showed significant differences between the relatively smooth glacierized surface and complex topographic structures. The total error of the estimated mean volume change is within 5% for most cases. Exceptions may result for small glaciers with inaccurate old maps or short periods of successive DEMs. The accuracy is better in the ablation area than in the accumulation area where changes are bigger and the texture is more distinct. In addition, errors in an individual DEM will be corrected by earlier or later DEMs in the time series and do not affect the total volume change.

4. Discussion and Conclusions

The DEMs were chosen not only to reach far back in time but also to resolve observed advancing and recession periods (SCNAT–ETHZ, 1881–2002). Whereas the period of steady to positive evolution during the 1960s–80s and the subsequent period of distinct mass loss are clearly displayed, our dataset is not yet detailed enough to clarify previous periods more specifically. All of the investigated glaciers show a similar evolution of the net balance rate, except Glacier du Giétro, where mass loss by break-off plays an important role until the 1970s. This glacier is situated in a lateral hanging valley. In the past, it repeatedly built up unconnected ice dams at the bottom of the main valley during advance periods (Reference Raymond, Wegmann and FunkRaymond and others, 2003). The glacier snout remained constant at the valley edge, and during advance ice broke off and no significant thickness change occurred in the ablation area. The resulting evolution of volume changes completes this picture.

The quality of the 19th-century maps (Siegfried survey) varies significantly. Many of these maps had to be rejected because the surface topography was insufficiently accurate. In some cases, a visual inspection revealed discrepancies, and comparison of successive DEMs provided an unambiguous test by evaluating the resulting spatial distribution of the thickness changes. Maps dating between 1920 and 1940, belonging to the first survey of the official Swiss map series, are of the required accuracy and are of unique value for glaciological reconstruction of former surface topography. The scale of the maps used in this study varies between 1 : 2000 and 1 : 50 000, resulting in large differences in the contour-line interval and accuracy of the DEM that may be obtained.

Many maps of the official series at scales 1 : 50 000 and 1 : 25 000 published by the Swiss Federal Office of Topography (swisstopo) exist, and have recently become available as digital products. However, a lot of care in the application is needed because updates are usually only made where changes are large, and the glaciers are not completely surveyed. In addition, the data of the mapped state is hardly known. So for our purpose the available accuracy is often insufficient.

The assumption of a constant ice density is reasonable if no major change in the firn layer is expected. Our results clearly show that thickness changes are generally small in the accumulation area, and we expect that the firn-layer thickness will remain constant even over long timescales. This is supported by direct point-based observations on Grosser Aletschgletscher or Claridenfirn (SCNAT–ETNZ, 1881–2002). The shifting of the firn/ice transition and the firn line was limited in the past but may become more important as complete meltout repeatedly occurs due to rapid downwasting.

Our datasets cover an area of 255.7 km² compared to an estimated total area of 1285.7 km² in the 2000 inventory of the Swiss Alps (Reference Paul, Kääb, Maisch, Kellenberger and HaeberliPaul and others, 2004). The calculated total volume loss of 12.96 km³ since the mid-19th century is underrated due to sparse datasets of the 19 selected glaciers, but relative values will give an indication on the changes involved. From the 1920s onward, a period covered by most of the individual time series, a volume loss of 8.73 km³ resulted, of which 3.45 km³ (40%) has occurred since 1980. For the periods 1920–60, 1960–80 and 1980–present, the decadal mean net balance rates are –0.29, –0.03 and –0.53ma^–1w.e., respectively. These values clearly illustrate the temporal variation of the thickness change evolution. Our results agree with a similar study of selected Austrian glaciers for the period 1950–79 (Reference Finsterwalder and RentschFinsterwalder and Rentsch, 1981) and a recent compilation of averaged observations of mass-balance time series for the whole Alps since 1960 (Reference Dyurgerov and MeierDyurgerov and Meier, 2005).

The selected glaciers in our study are geographically distributed over the major glacierized region of the Swiss Alps and reflect the spatially uneven distribution of large and small glaciers. Although there are marked differences between individual glaciers, there is no geographic pattern in the volume loss visible.

To complete our effort to establish time series of volume changes since the end of the Little Ice Age in the second half of the 19th century, a more extensive analysis of inadequate maps is needed. Our results indicate fewer problems on larger glaciers where bigger changes occur. The smaller glaciers, located mostly in remote areas, were given less careful attention by the surveyors.

Our analyses provide the baseline for further investigations of a reconstruction of a mass-balance time series of annual to seasonal resolution. A combined approach using an accumulation and melt model and validation with cumulative total volume changes seems to be a promising approach.