Precision snow cover and glacier mapping for runoff modelling in a high alpine basin JESKO SCHAPER, KLAUS SEIDEL & JAROSLAV MARTINEC Communication Technology Laboratory, Image Science Group, Swiss Federal Institute of Technology ETHZ, Gloriastrasse 35, CH-8092 Z¨urich, Switzerland, email:
[email protected] Abstract The runoff regime of high alpine basins will be affected in response to global warming not only by the seasonal snow cover but also by glaciers. Runoff peaks due to snowmelt will partially be shifted from the summer to the winter. The glacier melt will supply a higher amount of meltwater until glaciers finally disappear, if the temperature keeps rising. In order to evaluate quantitatively this effect, a method is presented to map separately the snow cover area, the gradually decreasing snow coverage on the glacier and the area of the exposed ice. The test basin Rhˆone-Sion (3371 km2 , 491 - 4634 m a.s.l.) has been divided into 7 elevation zones in view of the large altitude range. The periodical satellite monitoring of snow and ice areas was carried out separately for each zone with high resolution optical remote sensing data which provide the capability to distinguish between snow and ice. The paper demonstrates the necessity of accurate glacier and snow cover mapping by adequate satellite sensors in order to predict snow and glacier conditions as well as the changing runoff regime in high alpine basins in the 21 century. INTRODUCTION The accuracy of runoff modelling in high alpine basins is considerably improved by evaluating separately the snow coverage over glaciers and over glacier-free areas of each elevation zone. This approach takes into account the specific melt factors of ice and the actual elevation of glaciers within the respective elevation zones. Following a test in a small experimental basin [Schaper et al., 1999], the present paper demonstrates that the method can be applied in basins of several thousand km2. Apart from the improvement of the runoff modelling, the independent computation of glacier melt is an important step towards evaluations of glacier behaviour with regard to global warming. ˆ BASIN RHONE-SION The selected test basin of the river Rhˆone at Sion has an area of 3371 km2 and ranges from 488 to 4634 m a.s.l.. The total area of glaciers (among them the large Aletsch Glacier) amounts to 580 km2 or 17 %. The situation of the basin in the Swiss Alps is shown in Fig. 1. For the purpose of runoff modelling, the basin has been divided into 7 elevation zones with respective data listed in Tab. 1. As illustrated in Fig. 1, runoff from an area of 357 km 2 , including more than 35 glaciers, is collected by galleries and stored in the Grande Dixence reservoir with an storage volume
1
Sion
Nendaz
Lac de Dixence
Fionnay
Figure 1: Basin Rhˆone-Sion with runoff diversion to hydroelectric plants. Gray appears the shaded relief, white areas indicate the glaciers and black areas indicate the water bodies.
Zone 1 2 3 4 5 6 7 Total
Altitude Range [m a.s.l.] 488 - 1100 1100 - 1600 1600 - 2100 2100 - 2600 2600 - 3100 3100 - 3600 3600 - 4634 488 - 4634
Mean Altitude [m a.s.l.] Glacier Glacier-free 796 1374 1961 1884 2410 2384 2854 2849 3349 3362 3836 3880
Area [km2] Glacier Glacier-free 0 277.0 0 390.0 4.1 562.9 48.4 812.6 238.5 571.5 237.5 137.5 51.0 40.0 579.5 2791.5
Glacier Area [%] 0.7 5.6 29.4 63.3 56.0 17.2
Table 1: Elevation zones and glacier areas of basin Rhˆone-Sion. For the purpose of runoff modelling each zone is divided into glacier and glacier-free areas.
2
of 400 10 6 m3 . From this reservoir, water is diverted outside the basin to the hydroelectric plants Chandoline, Fionnay and Nendaz. Also, there are other hydroelectric schemes in the basin which store runoff in the summer and release it in the winter. Over 80 % of the hydroelectricity is produced in the winter in order to meet the demand in Switzerland. Both the runoff diversion and the reservoir operation must be taken into account in order to rectify the runoff measured at Sion to natural runoff which can be compared with the modelled runoff. The climatic conditions with regard to precipitation and temperature are assessed in earlier publications [Funk, 1985; Sch¨uepp et al., 1978]. SEPARATE MONITORING OF THE SNOW COVER AND GLACIERS For a deterministic approach of runoff modelling in high mountain basins, the changing areas of the seasonal snow cover and of the exposed glacier ice must be evaluated throughout the snowmelt season. In 1985, 10 subsequent Landsat-TM and Landsat-MSS images were available. An example from 12 September is shown in Fig. 2.
Figure 2: Detail from Landsat-TM image 195-28, Channel 1, from 12 September 1985 showing snow cover and glaciers of basin Rhˆone-Sion in bright colors.
Thanks to the good spatial resolution of the Landsat sensors (80 m for MSS and 30 m for TM) and advanced methods of satellite data processing [Ehrler & Seidel, 1995], it was possible to distinguish between snow and ice and periodically determine the respective areas in each elevation zone. To this effect, the scenes have been classified with multivariate statistics as snow, ice, and snow- and glacier-free. In a GIS analysis, snow cover units have been generated in order to complement the images in areas obscured by clouds. This method is described in detail elsewhere [Ehrler et al., 1997]. From this periodical map3
ping, depletion curves of the snow cover in glacier-free areas as well as of the snow cover superimposed on glaciers were derived. Glacier extension is known (see Tab. 1), so the increasing areas of exposed ice can be modelled for glacier melt computation. The variables snow cover and glacier ice constitute the direct input into the SRM-ETH model which was used in this study. SNOWMELT AND GLACIER MELT The degree-day factor, which is used to compute melt depths, includes the radiation component. Consequently it increases with aging of snow and sinking of albedo. In this study, the degree-day factor for snow was determined using snow density measurements since these are good indicators for the albedo [Anderson, 1973; Rango & Martinec, 1995]. The seasonally changing values range from 0.2 to 0.65 cm Æ C 1 d 1 . Glacier ice has a still lower albedo than old wet snow (0.17 - 0.22 as compared with 0.35) according to Bezinge [Bezinge, 1978] and therefore higher degree-day ratios are indicated. Using recent measurements of glacier ablation by Turpin [Turpin, 1997], the degree-day factor for ice was computed, resulting in values which are ranging from 0.61 to 0.79 cm Æ C 1 d 1 . The same range can be derived from an empirical formula for glacier ablation [Kotlyakov & Krenke, 1982]: A
= 1:33(T + 9:66)2:85
(1)
where A
T
= total glacier ablation [mm] = average temperature in degree-days for June, July and August.
For T = + 6Æ , a = 0.61 cm Æ C 1 d 1 , for T = + 2Æ , a = 0.79 cm Æ C 1 d 1 . In this study, the daily glacier melt depth were computed by a uniform ag = 0.7 cm Æ C 1 d 1 . Naturally, this factor was applied only to the gradually increasing snow-free area of glaciers. When new snow temporarily covered the ice, the degree-day factor for snow was substituted until the ice became again exposed. DERIVATION OF DEPLETION CURVES OF THE SNOW COVERAGE Usually the conventional depletion curves (CDC) of the snow coverage are derived by interpolating between points obtained from the periodical snow cover mapping. Consequently the snow coverage keeps decreasing even when no snowmelt takes place which occurs during occasional cold spells in the summer. In the SRM-ETH version, the interpolation is carried out in the so called modified depletion curves (MDC) which relate the snow coverage not to time, but to the cumulative computed snowmelt depth. For example, Fig. 3 shows the modified depletion curves derived for the glacier-free part and the glacierized part of the elevation zone 4. These curves stop declining on days with temperature 0 Æ C because no snowmelt depth is added on the x-axis. Since the dates are known on which the respective melt depth totals have been attained, the conventional depletion curves can be automatically derived by the computer program. For example, the snow coverage of the glacier in the zone 4 (area 4g ) was reduced to 50 % by a cumulative snowmelt depth of 90
4
1 Glacier 0.9
Glacier-free
0.8
Snow Cover
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0
50
100
150
200
250
300
350
400
450
500
550
Melt Depth (cm)
Figure 3: Modified depletion curves relating the snow coverage to the cumulative snowmelt depth, glacier-free and glacier-covered part of the zone 4.
1 Glacier 0.9
Glacier-free
0.8
Snow Cover
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
Apr
May
Jun
Jul
Aug
Sep
Figure 4: Conventional depletion curves of the snow coverage in the glacier-free and glacier-covered part of the zone 4.
5
cm (see Fig. 3). This cumulative snowmelt depth was reached on 4 July so that the CDC has S = 0.5 on this date. Conventional depletion curves for the zone 4 thus derived are shown in Fig. 4. These curves indicate the areas of snow and ice being melted which are multiplied by melt depth in order to obtain the daily volumes of the meltwater production. RUNOFF MODELLING With the separate computation of glacier melt, the SRM formula may be rewritten as follows:
Q +1 =
X
f[ a T S A + a T (1 S ) A + s;i
n
i
s;i
s;i
s;i
i
s;i
s;i
all zones
+ a T S A + a T (1 S ) A ℄ + + P (A + A ) g 10000 (1 k +1 ) + Q k +1 86400 s;i
i
r;i
s;i
g;i
g;i
g;i
g;i
r;i
i
g;i
n
g;i
s;i
n
n
(2)
where
Q n as ag T
= = = = =
s r
,
= = = = =
Pr
=
k
= =
Ss As Sg Ag
10000 86400
average daily discharge m3 s 1 . index indicating the sequence of days. the degree-day factor for snow [cm Æ C 1 d 1 ]. the degree-day factor for ice [cm Æ C 1 d 1 ]. number of degree-days extrapolated to the mean hypsometric altitude of glacier-free and glacier areas in each zone [ Æ C]. ratio of the snow covered area to the total area As . area of the glacier-free part of a zone [km 2 ]. ratio of the snow covered area to the total area Ag . glacier area in a zone [km2 ]. runoff coefficients expressing the losses as a ratio (runoff/ precipitation) with the index referring to snow and rain. precipitation as rain according to the critical temperature at the mean hypsometric altitude of a zone or new snow falling in the summer on snowfree area [cm d 1 ]. n+1 in a period without snowmelt or rainfall. recession coefficient = QQn conversion from [cm km2 d 1 ] to [m3 s 1 ].
In Fig. 5, the computed runoff is compared with the natural measured runoff. In line to previous studies [Funk, 1985; Sch¨uepp et al., 1978], special attention was paid to the distribution of precipitation and to the temperature lapse rate (0.65Æ per 100 m in the summer, 0.5Æ per 100 m in the winter). The runoff coefficient s varied seasonally between 0.8 and 0.9, r between 0.65 and 0.8. Eq. 2 corresponds to a time-lag of 18 hours, which was shortened to 10 hours (automatic program) in view of the glacier runoff collection by galleries and the rapid flow in pressure pipelines. The critical temperature (rain/snow) was preselected in the range of +1 to +3Æ C. Fig. 6 shows, by cumulative curves, the proportions of the runoff components. This gives a quantitative comparison of the contributions to runoff. 6
500 R2: 0.97 ∆V: -2.1 %
Natural Runoff measured Runoff simulated
Runoff Q [m3s-1]
400
300
200
100
0
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Figure 5: Natural runoff in the basin Rhˆone-Sion in the hydrological year 1985 compared wih the computed runoff. V = -2.1% means less runoff simulated than measured.
100
13% Icemelt
Percentage of Total Runoff
90 80
27% Rain
70 60
10% Newsnow
50 40 30
50% Snowmelt 20 10 0
Apr
May
Jun
Jul
Aug
Sep
Figure 6: Contributions to runoff in April – Sept. from the seasonal snow cover, new snow, rain and glacier.
7
CONCLUSION Accurate runoff modelling in a large high-alpine basin with a significant glacier area has been achieved by the following factors: 1. Advanced snow cover and glacier mapping with a spatial resolution of 30 m as available from Landsat-TM and pixel classification to the categories snow- and glacier-free, snow, ice. 2. Complementation of satellite images partially obscured by clouds. 3. Deriving depletion curves of the snow coverage taking into account daily temperatures. 4. Separate computation of snow- and ice melt. Enhanced monitoring of the spatial distribution of seasonal snow cover are also needed in view of the climatic warming [WMO-UNEP, 1990], in order to predict snow and glacier conditions as well as the changing runoff regime in high alpine basins in the 21st century. ACKNOWLEDGEMENTS This study was carried out with the support of the Swiss National Science Foundation (Grant 21-53539). REFERENCES Anderson, E. A. (1973). Techniques for predicting snow cover runoff. In The role of Snow and Ice in Hydrology, Proceedings of the Banff Symposium, 840–863. IAHS Publ. No. 107. Bezinge, A. (1978). Glacial torrents, hydrology and bedload (Torrents glaciaires, hydrologie et charriages d’alluvions). In Glaciology Symposium, Annual Assembly of the Swiss Society of Natural Sciences, Brig, Switzerland. Ehrler, C. & Seidel, K. (1995). Mutual effects of the climate change and the alpine snow cover and their influence on the runoff regime evaluated with the aid of satellite remote sensing. In Stein, T. I. (Ed.), IGARSS’95; Quantitative Remote Sensing for Science and Applications, Florence, Italy, 3: 1973–1975. Ehrler, C., Seidel, K., & Martinec, J. (1997). Advanced analysis of the snow cover based on satellite remote sensing for the assessment of water resources. In Baumgartner et al. (Eds.), Remote Sensing and Geographic Information Systems for Design and Operation of Water Resources Systems, 5th Scientific Assembly of the IAHS, Rabat, Morocco., 93–101, IAHS Publ. No. 242. Funk, M. (1985). R¨aumliche Verteilung der Massenbilanz auf dem Rhˆonegletscher und ihre Beziehung zu Klimaelementen. Z¨urcher Geographische Schriften 24. H. Lang et al. (Eds.). Kotlyakov, V. M. & Krenke, A. N. (1982). Investigations of the hydrological conditions of alpine regions by glaciological methods. In Hydrological Aspects of Alpine and High-Mountain Areas (Proceedings of the Exeter Symposium, 1982), 31–42. IAHS Publ. No. 138. Rango, A. & Martinec, J. (1995). Revisiting the degree-day method for snowmelt computations. Water Resources Bulletin, 31(4):657–669. Schaper, J., Martinec, J., & Seidel, K. (1999). Distributed Mapping of Snow and Glaciers for Improved Runoff Modelling. Hydrological Processes, 13(12–13):2023–2031. Sch¨uepp, M., Bou¨et, M., Bider, M., & Urfer, C. (1978). Le Valais. In SMA (Ed.), Klimatologie der Schweiz, Band II, Regionale Klimabeschreibungen, 1. Teil, 88–114. Turpin, O. C. (1997). Glacier melt measurements at Findelen Glacier, Valais, from July - August 1993. personal communication. WMO-UNEP (1990). Intergovermental Panel on Climate Change (IPCC), Potential Imppacts of Cllimate Change. Report prepared for IPCC by Working Group II, World Meteorological Organization, Geneva.
8