Réf. Dehn & al 2000 - A

Référence bibliographique complète
  DEHN, M., BÜRGER, G., BUMA, J., GASPARETTO, P. Impact of climate change on slope stability using expanded downscaling. Engineering Geology, 2000, Vol.55, 3, 193-204.

Abstract: Climate parameters affecting ground water and pore pressure fluctuations can, in many cases, trigger slope instability and hence landslide activity. Global warming due to the greenhouse effect and especially changes in precipitation patterns and air temperature might therefore have influences on future landslide activity. The present paper shows an assessment of climate change consequences for displacement rates of a mudslide in the Dolomites, Italy. The study is based on climate projections of a general circulation model (GCM). GCMs are able to succesfully reproduce large-scale patterns of climate, while they show a poor performance on the regional scale. Therefore, GCM output is postprocessed with a statistical downscaling technique to derive local-scale climate change information from simulated atmospheric circulation patterns of the European–North Atlantic sector. The resulting precipitation and temperature series are introduced in a hydrological tank model, which calculates daily groundwater levels. Based on the groundwater data, a visco-plastic rheological model is applied to derive displacement rates of the mudslide as final output. The climate change signal is most pronounced for air temperature, while it is weaker but still significant for yearly precipitation, which is decreasing. As a consequence, yearly displacement rates show a significant reduction. The most dramatic changes, however, occur in spring with strongly lowered groundwater levels and consequently decreasing displacement rates. This is seen as an effect of reduced storage of winter precipitation as snow and hence decreasing meltwater amounts in early spring. The presented model chain with statistical downscaling, hydrological and rheological models allows the assessment of future landslide displacement affected by the greenhouse effect. The results, however, have to be taken with caution since in all parts of the model chain there are uncertainties that are difficult to address.

Mots-clés
Climate change impact; Rheological model; Slope hydrology; Statistical downscaling

Organismes / Contacts
Department of Geography, University of Bonn, Meckenheimer Allee 166, 53115 Bonn, Germany.
Potsdam Institut für Klimafolgenforschung (PIK), P.O. Box 601203, D-14412 Potsdam, Germany.
Department of Physical Geography, University of Utrecht, P.O. Box 80115, 3508 TC Utrecht, Netherlands.
IQT Consulting s.r.l., Rovigo, Italy

(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
Precipitation and temperature   Mass movements Mudslide

Pays / Zone
Massif / Secteur
Site(s) d'étude
Exposition
Altitude
Période(s) d'observation
Italy Eastern Dolomites Alverà mudslide located near Cortina d'Ampezzo      

(1) - Modifications des paramètres atmosphériques
Reconstitutions
 
Observations
 
Modélisations

In the simulation of the 'current climate' (given by the recent 23 year average) evolution on the period ~1900-2100, the Cortina temperature begins to leave the confidence strip at about the turn of the century, and which continues to increase afterwards. Precipitation, however, has a broader confidence band and a smaller trend. Nevertheless, it is clearly visible that the curve leaves the band of current climate at about 2010, and stays below it for the rest of the scenario, without a significant trend.

The average temperature in winter (DJF) increases significantly, and, more importantly, the average winter temperatures are all above 0°C from the year 2050 onwards. Furthermore, precipitation in DJF decreases more strongly than in the other seasons.

Hypothèses
 

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

For the climate change scenario, we used data of the global coupled ocean–atmosphere GCM ECHAM4/OPYC3. We have adapted the view that the ‘control climate’ is the modelled GCM climate from the era 1964–1986. Since the subsequent downscaling model is based on anomaly information, all relevant GCM information is obtained accordingly as anomaly from the 1964–1986 GCM climate.


(2) - Effets du changement climatique sur le milieu naturel
Reconstitutions
 
Oservations
 
Modélisations

The average temperature in winter (DJF) increases significantly, and, more importantly, the average winter temperatures are all above 0°C from the year 2050 onwards. Furthermore, precipitation in DJF decreases more strongly than in the other seasons. A combined effect of these findings is a clearly reduced storage of precipitation as snow. Therefore, the release of melt water, which, under the present conditions contributes to high groundwater levels and hence strong displacement in early spring, is significantly reduced.

Hypothèses
 

Sensibilité du milieu à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)

 

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

 

Observations
 
Modélisations

The combination of three modelling steps, namely expanded downscaling, hydrological and rheological modelling to derive displacements of a mudslide from GCM experiments, is a new and promising approach. In comparison to earlier studies (Buma and Dehn) there now exists the possibility to assess climate change impact not only concerning years with or without activity but also the amounts of displacement of the simulated mudslide.

One major problem still limits the practical application of the method in this stage of development. Besides inherent problems of the two slope models including extrapolation of results of one borehole, great differences sometimes occur between observations, ANA (downscaled processes stemming from the analysed circulations), and the years 1964–1986 of SCA (modelled circulations). This is valid for all three examined parameters, precipitation, groundwater and displacement. While the differences between ANA and observed data can be attributed to insufficiencies of EDS (Expanded downscaling), the differences between the control period of the GCM and observations have to be interpreted as a cumulative error of EDS and the GCM. These initial shortcomings in downscaling and GCM experiment are consequently traced through all stages of the approach. Therefore, following the experiences of two of the authors (Buma and Dehn) it is highly recommended to use more than one GCM experiment for assessing climate change impacts. In this way, at least GCM uncertainty can be narrowed to some degree. A further improvement would be the use of more than one downscaling technique. In the present study, due to computing and time resources, [the authors] were only able to use the ECHAM4/OPYC3 GCM, which was downscaled with EDS.

Uncertainties of the model chain stemming from the two slope models have already been mentioned. One part of this uncertainty is due to the limited length of the measurement period used for model calibration. Further, both slope models are simple lumped approaches that are unable to adequately represent the spatial variability of geotechnical and hydrological parameters of the slope.

The decrease in displacement from March to May can be attributed to two changing parameters. The average temperature in winter (DJF) increases significantly, and, more importantly, the average winter temperatures are all above 0°C from the year 2050 onwards. Furthermore, precipitation in DJF decreases more strongly than in the other seasons. A combined effect of these findings is a clearly reduced storage of precipitation as snow. Therefore, the release of melt water, which, under the present conditions contributes to high groundwater levels and hence strong displacement in early spring, is significantly reduced. This effect can be suggested as a conceptual explanation for the strong reduction of future landslide displacement in spring. The physics of this effect have not been investigated in detail, however.

Hypothèses

Climate parameters affecting ground water and pore pressure fluctuations can, in many cases, trigger slope instability and hence landslide activity. Global warming due to the greenhouse effect and especially changes in precipitation patterns and air temperature might therefore have influences on future landslide activity.


Paramètres de l'aléa
Sensibilité du paramètre de l'aléa à des paramètres climatiques et du milieu
Informations complémentaires (données utilisées, méthode, scénarios, etc.)
Slope stability

Slope hydrology

The monitoring equipment [of the mudslide] consists of four inclinometric tubes, 11 piezometers and 11 extensometers all equipped with electric pressure transducers connected to an automatic recording system. Furthermore, a meteorological station has been collecting data since 1989.

Observed climate records from the Italian Meteorological Survey consist of daily precipitation for Cortina d'Ampezzo (1922–1996), Misurina (1922–1975) and San Vito (1921–1987). Misurina is located 10 km ENE and San Vito 12 km SE of Cortina d'Ampezzo. The data of Cortina were tested on a monthly basis for homogeneity against the other two stations, which did not show any problems. Daily minimum and maximum air temperature were available from the same source for Cortina only.

The submodels expanded downscaling, a hydrological model and a visco-plastic rheological model are linked to derive slope displacement from GCM experiments.

The model calibration consists of two parts. The first part concerns the ability of the hydrological and rheological model to reproduce observed conditions and was already presented in the previous section. The second part focuses on the ability of the downscaling technique to simulate the local weather correctly.

For the same periods, the mean monthly displacement of the mudslide and the seasonal pattern are captured well, while the amplitude of the annual cycle is overestimated based on the downscaled climate. Displacement based on downscaling is underestimated during summer and autumn, while it is overestimated in spring.

Simulation of future landslide activity is based on the ECHAM4/OPYC3 experiment described above and the downscaled climate variables for the Cortina region.


(4) - Remarques générales

The presented assessment of future climate-induced slope displacement shows several features:

• Coupling of the various models is feasible and allows the estimation of future landslide displacement as affected by the greenhouse effect.

• Several model biases are visible, while it is not easy to find their sources.

• Climate change signal is weak but significant.

• Due to errors and a weak signal, the uncertainties of the results are still high.

• Further climate change impact assessments should try to include various GCM experiments and different downscaling techniques.

• Finally, changing boundary conditions like vegetation pattern or possible enhanced development of shrinkage cracks in the root zone due to higher temperatures should be included in further approaches.

The shortcomings of the presented study can mainly be assigned to biases in the GCM at a large scale, to biases of the scale tranformation between the large and local scale caused by EDS and biases of the slope model at the borehole scale. The scale transformation between the borehole scale and the landslide scale also introduces uncertainty and must be achieved by upscaling the slope models.


(5) - Syntèses et préconisations

[...] following the experiences of two of the authors (Buma and Dehn) it is highly recommended to use more than one GCM experiment for assessing climate change impacts.

Références citées :

Buma, J. and Dehn, M., 1999. A method for predicting the impact of climate change on slope stability. Environ. Geol. 35, pp. 190–196.

Dehn, M. and Buma, J., 1998. Modelling future landslide activity based on general circulation models. Geomorphology submitted for publication.