Skip to Main content Skip to Navigation
Journal articles

Peut-on estimer l'effet du changement climatique sur l'écoulement à l'exutoire d'un bassin sans modèle pluie-débit ? Un test de la méthode de transfert climat-écoulement par régression dans le bassin transnational de la Meuse

Abstract : Study of the future impact of climate change on streamflow at the outlet of a catchment is usually based on a rainfall-runoff model. This paper explores pros and cons of an alternate strategy using a regression-based transfer function between climate and streamflow for assessing the impact of climatic change on flow at the outlet of a catchment. The climate-flow relationship is modeled through a PLS (Partial Least Squares) regression – RLM (Multiple Linear Regression) regression sequence. The RLM regression models are calibrated between low, medium and high streamflow indices and the best climate indices, which were preliminary identified through PLS regression according to quantitative criteria as well as an expert judgment of current hydro-climatic conditions. The regression-based climate-streamflow transfer function is then applied to climate model outputs to produce hydrological projections for the 2021-2050 and the 2071-2100 periods. Results obtained for two pilot rivers of the Meuse basin show that i) this method is efficient for some flow indices only; ii) a minimal knowledge about the climate variability (through atmospheric pressure and pressure gradient force at the ground and at the level of 500 hPa) is sufficient to depict in an acceptable way the climatic forcing on these indices; iii) an expert judgment is essential in identifying hydrologically relevant climatic predictors; iv) the strength of the climate-streamflow relationship is catchment-dependent; v) streamflow changes predicted by the climate-flow method for the next decades under CMIP5 (Coupled Model Intercomparison Project Phase 5) climate scenarios are in agreement with those of rainfall-runoff modelling. However, at the stage of our investigations, three main limits of the regression-based climate-streamflow transfer function were identified: 1) it needs long hydro-climatic time series; 2) it tends to under-estimate interannual streamflow variability; 3) there is no guarantee of its ability to extrapolate to climatic impact on streamflow indices in climatic conditions that are far from those observed in historical records.
Complete list of metadatas

Cited literature [41 references]  Display  Hide  Download

https://hal.univ-lorraine.fr/hal-01695680
Contributor : Benjamin Grelier <>
Submitted on : Tuesday, January 30, 2018 - 7:13:38 PM
Last modification on : Friday, October 23, 2020 - 4:48:46 PM
Long-term archiving on: : Friday, May 25, 2018 - 4:15:16 PM

File

03_grelier_et_al_climatologie_...
Publisher files allowed on an open archive

Identifiers

Collections

Citation

B. Grelier, Gilles Drogue, Michel Pirotton, Pierre Archambeau, Emilie Gernez. Peut-on estimer l'effet du changement climatique sur l'écoulement à l'exutoire d'un bassin sans modèle pluie-débit ? Un test de la méthode de transfert climat-écoulement par régression dans le bassin transnational de la Meuse. Climatologie, Association internationale de climatologie, 2017, 14 (Volume 14), pp.48-81. ⟨10.4267/climatologie.1232⟩. ⟨hal-01695680⟩

Share

Metrics

Record views

302

Files downloads

315