Institute of Geography and Sustainability of the University of Lausanne
Hydro-geomorphologic hazards in alpine catchments under uncertainty and climate change perspective
The SNF Marie Heim-Vögtlin project dealt with hydro-geomorphologic modelling on two contrasting catchments: the Vallon de Nant, a natural reserve of 13 km2 and the intensively urbanised La Tintaz-Verbier catchment (1.4km2) in south-western Switzerland in order to evaluate their hazard potential under uncertainty and climate change perspective. The hydrological model, WASIM-ETH water balance model (Schulla, 1997) was calibrated in a continuous simulation mode during the 2005-2008 period at a daily time step and further validated for the period January 2009 - September 2010. Given the input data and the model structure, the parametric and predictive uncertainty was quantified by means of a Bayesian Monte Carlo Markov Chain method. The hydrologic modelling was done in an event-based mode, for the Tintaz/Verbier catchment (1.4 km2) at an hourly time step, during the spring-summer 2010, the only period with hydro-meteorological complete data. The coupling between WASIM-ETH and the infinite slope stability model enabled dynamic modelling of the stability factors that further allowed delineation of the susceptible areas for shallow slope instability phenomena such as spontaneous landslides. While most studies use a static model to estimate potential initiation areas for landslides and debris flows, this study is among few ones using a dynamic approach of shallow slope instability and is the first attempt to couple and run the WASIM-ETH hydrological model with a deterministic infinite slope instability model in order to dynamically model spatio-temporal shallow landslides initiation areas. The first results are very encouraging as basically the spatio-temporal shallow unstable zones correspond with those mapped in the landslides map of the Avançon catchment and are completely within the potentially shallow unstable regions in the indicative map of Vaud canton. Nevertheless, constraining the model with mapped/observed data from landslides inventories, aerial, geomorphologic mapping will hopefully contribute to reduce the over-estimation of the instable areas, which is usual the case when using the infinite slope instability models. Another way to cope with overestimation of potential initiation areas would be to use a multi-response calibration on simultaneous discharge and soil moisture data, this latter being one of the inputs in the slope stability model. While the statistical methodology is already well developed and thus available (Balin et al., 2010), its present implementation was not possible as soil moisture data are not available in the study catchments. The quality of the simulated results depends also on the resolution of the spatial data such as DTM model, land use and soil properties. Despite all these limitations, the new-coupled hydro-geomorphologic WASIM-ETH-FS model performs satisfactory in comparison with the static approaches, as spatial overestimation of unstable areas is much less important. This opens first the way to more valuable related danger indicative maps and further to more robust approaches for further hazard and risk evaluation. Moreover, after validation, such a deterministic approach may in the future be used to forecast changes in hazard under different land scenarios and changes in hazard caused by climatic change. The climate change scenarios have been assesses by using the climatic variables from the REMO-UBA RCM (MPI 2006, UBA 2006) model with a 10x10km2 spatial resolution and daily time resolution. It was decided to use a scaling approach instead of the well-known “delta-change” method, to correct the scenario data by applying a multiplicative factor for the precipitation and an additive one for the temperature, the main advantage being those of maintaining the rainfall temporal occurrence such as simulated by the RCM model. The calibrated WASIM-ETH model was further applied during a control period 1995-2000 and a scenario period 2045-2050 at a daily time step. Both control and scenarios runs have been done within the A2 and B1 contrasting emission scenarios for the Avancon catchment. After computing the multi-annual monthly means, the results indicate that the hydrologic regime will probably change in the future especially under A2 scenarios with the high flows period earlier than is actually known. The by-products of the WASIM-ETH such as the time series of simulated snow, glacier melt and pluvial components show clearly that the snow and glacier melt will decrease in the future and the hydrological regime will become a more pluvial one. Another potentially plausible result indicate the higher inter-annual variability of the flows with potentially drought years followed by years with potential extreme flooding especially under the A2 scenario. More emphasis should thus be put to study the impact of climate change on extremes since the long-term mean annual flood would be a less significant way to characterize the hydrological regime.