Réf. Horton & al. 2006 - A

Référence bibliographique complète
HORTON P., SCHAEFLI B., MEZGHANI A., HINGRAY B., MUSY A. Assessment of climate-change impacts on alpine discharge regimes with climate model uncertainty. Hydrological Processes, 2006, Vol. 20, p. 2091-2109.

Abstract: This study analyses the uncertainty induced by the use of different state-of-the-art climate models on the prediction of climate-change impacts on the runoff regimes of 11 mountainous catchments in the Swiss Alps having current proportions of glacier cover between 0 and 50%. The climate-change scenarios analysed are the result of 19 regional climate model (RCM) runs obtained for the period 2070-2099 based on two different greenhouse-gas emission scenarios (A2 and B2 scenarios) and on three different coupled atmosphere-ocean general circulation models (AOGCMs). The hydrological response of the study catchments to the climate scenarios is simulated through a conceptual reservoir-based precipitation-runoff transformation model. The results obtained show that all climate-change scenarios induce, in all catchments, an earlier start of the snowmelt period, leading to a shift of the hydrological regimes and of the maximum monthly discharges. The mean annual runoff decreases significantly in most cases. For the glacierized catchments, the simulated regime modifications are mainly due to an increase of the mean temperature and the corresponding impacts on the snow accumulation and melting processes. The hydrological regime of the catchments located at lower altitudes is more strongly affected by the changes of the seasonal precipitation. For a given emission scenario, the simulated regime modifications of all catchments are highly variable for the different RCM runs. This variability is induced by the driving AOGCM, but also in large part by the inter-RCM variability. The differences between the different RCM runs are so important that the predicted climate-change impacts for the two emission scenarios A2 and B2 are overlapping.

Climate change, regional climate models, hydrological modeling, snowmelt modeling, glacier hydrology, Alps.

Organismes / Contact
Ecole Polytechnique Fédérale de Lausanne, Hydrology and Land Improvement Laboratory, CH-1015 Lausanne, Switzerland.
Benoît Hingray, Hydram-ISTE, Station 2, EPFL, 1015 Lausanne, Switzerland. benoit.hingray@epfl.ch

(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
Temperature, Precipitation Rivers, Soil humidity    

Pays / Zone
Massif / Secteur
Site(s) d'étude
Période(s) d'observation
Switzerland Alps 11 mountainous catchments   480-4310 m a.s.l. 2070-2099

(1) - Modifications des paramètres atmosphériques
The AOGCMs predict global-mean warmings for scenario B2 from +2.4°C to +2.8°C, and from +3.0°C to +3.6°C for scenario A2 (Roeckner et al., 1999; Gordon et al., 2000; Gibelin and Déqué, 2003). The regional mean annual temperature increases predicted by the 19 PRUDENCE RCM experiments are similar for all catchments and are higher than the global-mean warming, at around +3.0°C for B2 and +4.0°C for A2. The predicted regional warming for the summer is higher than for the other seasons.

Except for a few experiments and catchments, a decrease of the annual precipitation is predicted; there is no clear gradation between the projected changes of the annual precipitation for B2 and A2. Most RCMs predict an increase of winter precipitation and a decrease of summer precipitation for both scenarios and for all catchments. For spring, the predictions of the different RCMs do not have a clear tendency towards an increase or a decrease (for all catchments and both scenarios). For autumn, precipitation is predicted to decrease for the A2 scenario, but no tendency is detectable for B2.

Informations complémentaires (données utilisées, méthode, scénarios, etc.)
Each RCM experiment consists of a simulation for the period 1961-1990 (control run) and a simulation for the period 2070-2099 (future run). For each RCM experiment, the boundary conditions were obtained from one of the three AOGCMs used in PRUDENCE (Christensen et al., 2002): ARPEGE/OPA, HadCM3 and ECHAM4/OPYC3. The AOGCM experiments were completed for the two SRES emission scenarios A2 and B2. In this study, nine RCMs are included: ARPEGE, HIRHAM, CHRM, CLM, HadRM3H, RegCM, REMO, RCAO and PROMES. Outputs from 19 RCM experiments are available in total: 12 for the A2 scenario and seven for B2.

(2) - Effets du changement climatique sur le milieu naturel
The predicted decrease of glacier surfaces between the control and the future period is considerable, with almost no ice-covered areas left in the future. For the highest catchment, the Drance catchment, the median future predicted glacier-covered area corresponds to 2% of the total catchment area for B2 (1% for A2), compared with the 39% glacier cover during the control period. The simulated complete disappearance of the Rhône glacier is in line with the results obtained by Wallinga and van de Wal (1998) with a physically based one-dimensional flow model.

Runoff changes
The simulated annual discharges show a significant decrease compared with the control period, for all catchments and for all climate experiments. The median decrease ranges between -26% (Drance de Bagnes) and -12% (Vorderrhein) under the B2 scenario, and between -30% (Drance de Bagnes) and -14% (Vorderrhein) under scenario A2. It should be noted that the simulated mean discharge decrease may be overestimated as a result of the highly simplified simulation of the glacier retreat.

The considerable decrease of the mean annual discharges has several reasons, the most evident being the decrease in mean annual precipitation. Another significant part of the discharge reduction is due to the predicted increase of evapotranspiration. For the currently glacierized catchments, this increase is partly a consequence of the significant glacier retreat, which results in an increase of the ice-free portion of the catchment where evaporation is subtracted from precipitation. Additionally, the substantial temperature increase throughout the seasons enforces the total evapotranspiration on ice-free areas; accordingly, all catchments show a strong increase of total evapotranspiration. This increase is expected to be less important in summers than in winters: even if there is a pronounced temperature increase in summer, summer evapotranspiration will be limited due to reduced precipitation.

The important variability of annual precipitation changes simulated by the different climate models and the significant variability of simulated changes in evapotranspiration explain the different magnitudes of mean annual runoff decrease. Note that if the maximum predicted discharge decrease is significantly higher for the A2 scenario than for the B2 scenario, then the median and minimum decreases obtained for both scenarios are quite similar. The variability highlighted is mainly due to inter-model differences.

Hydrological regime changes
All simulations under the future regional climate scenarios show the same significant changes of the hydrological regimes and the same trends of these changes. Summer discharge is found to reduce significantly, winter discharge increases, and the snowmelt-induced peak is shifted to earlier in the year and decreases in most cases. Owing to the substantial reduction of the glacierized areas, pure glacial discharge regimes tend to disappear. The changes are very similar for catchments having the same present regime, independently of their geographical situation. The changes predicted for the B2 scenario tend to be enhanced for A2. The median simulated shift of the maximum monthly or semi-monthly discharge is, for example, around half a month for B2 and an entire month for A2. The results are, however, highly variable for the different RCM experiments. For many catchments the main difference compared with the present regime types was the appearance of a small secondary peak in autumn that does not exist in present intra-alpine regimes, but only in south-alpine regimes.

It should be noted that the shift of a glacier- or snowmelt-driven hydrological regime to a future regime more driven by snowmelt and/or rainfall will be associated with a modification of the year-to-year variability of the discharges. For high-elevation catchments (catchments with current proportions of glacier cover higher than 15%), the variation coefficient of annual discharges is, for example, predicted to increase by around 85% for scenario A2. The interannual variability of seasonal discharges also changes significantly due to an important modification of the temporary water storage properties of these mountainous catchments.

For catchments with high elevations, the regime modifications are highly conditioned by the changes of the seasonal temperatures. The simulated maximum monthly discharge is, for example, highly correlated to the change in mean spring temperature. For catchments at lower altitudes, the influence of precipitation modification is more pronounced and the variability of the climate-change impacts is mainly due to the large range of predicted regional precipitation changes. In terms of impact on the hydrological regime, the results suggest that the driving AOGCM may induce more impact prediction variability than the inter-RCM variability for a given AOGCM. Nevertheless, the results show that the inter-RCM variability is far from being negligible.

For both scenarios, the Rhône catchment (glacial regime) and the Weisse Lütschine catchment (glacio-nival regime) tend to have a future regime essentially driven by spring snowmelt. The variability of the results obtained for the different climate models is, however, high and the corresponding regimes can be quite different. The snowmelt-induced peak occurs, for example, between late May and July. The attenuation of this maximum monthly discharge is also highly variable from one scenario to another.

The changes obtained for the Verzasca catchment (nivo-pluvial meridional regime) are slightly less variable. The shift of snowmelt is about half a month for all experiments. The rainfall response peak in autumn does not shift in time, but does vary in amount. This catchment is, thus, still expected to exhibit the same nivo-pluvial meridional regime, as it does presently.

For the Minster catchment (nival of transition regime), the peak observed for the control run in May-June shifts to April-May under B2 and to even earlier for A2. The seasonality of discharges is still significant, but much less pronounced than for the control period; for A2, the seasonality tends to disappear. For both emission scenarios, the predicted changes vary significantly between the RCM experiments. For the two extreme climate experiments, the Minster regime even becomes exclusively driven by rainfall.

Sensibilité du milieu à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)
With temperature increase and precipitation decrease, annual discharges are expected to deacrease. Summer discharge is found to reduce significantly, winter discharge increases, and the snowmelt-induced peak is shifted to earlier in the year.
The 11 catchments analysed (Drance de Bagnes, Saaser Vispa, Lonza, Rhône at Gletsch, Weisse Lütschine, Minster, Tamina, Vorderrhein, Dischmabach, Rosegbach and Verzasca) were selected to represent the seven different hydro-climatic zones of the Swiss Alps (Laternser and Schneebeli, 2002) and the different mountainous hydrological regimes occurring in these zones. The catchments selected have different proportions of glacier cover (between 0 and 50%) and altitude ranges. Their total areas vary between 39 and 185 km2, and their mean altitudes vary from 1340 to 2940 m a.s.l. All catchments selected (except the Verzasca catchment) show typical discharge patterns characterized by a single peak occurring between May and August.

The simulation of catchment behaviour for the different time periods was carried out using a hydrological model, for the precipitation-runoff transformation, and a land-cover evolution model. The only land-cover change that is explicitly accounted for is the modification of the glacier-covered surface. In this study, the glacier surface for future time periods is updated through a conceptual modelling approach based on the so-called accumulation area ratio (AAR) method. The mean annual AAR value for the present situation was assumed to remain constant for the future period.

The glacio-hydrological model needs three input time-series, namely daily precipitation, daily PET and mean daily temperature. For all catchments, the precipitation and temperature data are obtained from nearby meteorological stations. The hydrological discharge simulation is carried out at a daily time-step through a conceptual reservoir-based model called GSM-SOCONT (Schaefli et al., 2005). For temperature, regional altitudinal gradients are estimated based on the observed temperature series. A mesoscale annual precipitation gradient of 80 mm per 100 m, estimated by Kirchhofer and Sevruk (1991) for the entire Swiss alpine region, is applied. The aggregation state of precipitation (liquid, solid or mixed) is determined through a fuzzy temperature threshold approach (Klok et al., 2001). Snow and ice melt are simulated through a simple degree-day approach (Rango and Martinec, 1995). Actual evapotranspiration is estimated as a function of the potential evapotranspiration (PET) and the filling rate of the slow component reservoir. The PET data are derived from the monthly PET series estimated by New et al. (2000) for the 1961-1990 control period, according to the Penman–Monteith version given by Burman and Pochop (1994).

The model has seven parameters to calibrate (four for catchments without a glacier): two degree-day factors for the ice and snowmelt computation, four time constants of the reservoirs, and the maximum storage capacity of the slow reservoir. Calibration and validation were undertaken for two consecutive 10-year periods (1971-1980 and 1981-1990 respectively, for the majority of the catchments). The series of mean daily discharges for model calibration and validation are provided by the Swiss Federal Office for Water and Geology. The calibrated model performs well for all catchments for both the calibration and the validation period.

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

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

(4) - Remarques générales

(5) - Syntèses et préconisations
Considering the various hydrological regimes and the predictions associated with the different climate models, there is a general trend in the simulated climate-change impact for both emission scenarios: the predicted climate change results in a significant decrease of the total annual discharge and in a shift of the monthly maximum discharge to earlier periods of the year due to the temperature increase and the resulting impact on the snow melting processes. An increase of the discharge year-to-year variability is also expected. Considering all RCM experiments, the resulting ranges of hydrological regimes for both emission scenarios are overlapping. The large prediction variability induced by the 19 RCM experiments is partly due to the underlying driving AOGCMs. The results of the present study clearly show that the use of several RCM experiments with the same driving AOGCM can result in impact differences that are comparable to the results from the same RCM driven by different AOGCMs.

Références citées :

Burman R, Pochop LO. 1994. Evaporation, Evapotranspiration and Climatic Data . Elsevier: Amsterdam.

Christensen JH, Carter TR, Giorgi F. 2002. PRUDENCE employs new methods to assess European climate change. Eos, Transactions of the AGU 83 : 147.

Gibelin AL, Déqué M. 2003. Anthropogenic climate change over the Mediterranean region simulated by a global variable resolution model. Climate Dynamics 20 (4): 327–339.

Gordon C, Cooper C, Senior CA, Banks H, Gregory JM, Johns TC, Mitchell JFB, Wood RA. 2000. The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments. Climate Dynamics 16 (2–3): 147–168.

Kirchhofer W, Sevruk B. 1991. Mean annual corrected precipitation depths 1951–1980. In Atlas Hydrologique de la Suisse . Service Hydrologique et Géologique National: Bern.

Klok EJ, Jasper K, Roelofsma KP, Gurtz J, Badoux A. 2001. Distributed hydrological modelling of a heavily glaciated alpine river basin. Hydrological Sciences Journal 46 (4): 553–570.

Laternser M, Schneebeli M. 2002. Spatial grouping of snow stations, snow and avalanche climatology of Switzerland . ETH Zürich.

New M, Hulme M, Jones P. 2000. Representing twentieth-century space–time climate variability. Part II: development of 1901–96 monthly grids of terrestrial surface climate. Journal of Climate 13 (13): 2217–2238.

Rango A, Martinec J. 1995. Revisiting the degree-day method for snowmelt computations. Water Resources Bulletin 31 (4): 657–669.

Roeckner E, Bengtsson L, Feichter J, Lelieveld J, Rodhe H. 1999. Transient climate change simulations with a coupled atmosphere–ocean GCM including the tropospheric sulfur cycle. Journal of Climate 12 (10): 3004–3032.

Schaefli B. 2005. Quantification of modelling uncertainties in climate change impact studies on water resources: application to a glacier-fed hydropower production system in the Swiss Alps . Doctoral thesis, Ecole Polytechnique Fédérale de Lausanne.

Wallinga J, van de Wal RSW. 1998. Sensitivity of Rhonegletscher, Switzerland, to climate change: experiments with a one-dimensional flowline model. Journal of Glaciology 44 (147): 383–393.