What can hydrological time series variations tell about karst dynamics? A coupled statistical/conceptual modeling analysis - Normandie Université Access content directly
Conference Papers Year : 2015

What can hydrological time series variations tell about karst dynamics? A coupled statistical/conceptual modeling analysis

Abstract

In this research we study the capability of time series analysis approaches to extract meaningful components of karst spring hydrographs. In this aim we compare these statistical components to the internal components of a conceptual precipitation/discharge model based on the physical knowledge of the site studied. We used the conceptual modeling software KARSTMOD developed by the INSU/CNRS National Karst Observatory to model discharge at a small karst spring in Normandy (France). The model comprised four reservoirs E, L, M and C (interpreted as epikarst, high- inertia/highly capacitive matrix, fissure network and conduits), consistent with previous works showing the existence of a triple porosity in chalk of Normandy. KARSTMOD internal flow components were analyzed with correlation and Fourier spectral analysis, and compared to statistical components extracted from spring discharge by wavelet multiresolution analysis and Ensemble Empirical Mode Decomposition (EEMD). We could also analyze how the hydrological signal acquired its red noise statistical characteristics while water flow propagates into the conceptual model. The trend of the discharge signal, given by the residue of EEMD, appeared quite similar to the variation in reservoir L and well correlated to the variation of the water level within the aquifer. Exchanges between fissured matrix and conduits (reservoirs M and C) could be also investigated: a high frequency pressure pulse-controlled flow from C to M (intermittent recharge from the conduits) was identified, as well as fissured matrix flow likely to take place in the surroundings of the conduit network. Flow from reservoir M to reservoir C could be recovered by recombining wavelet components of spring discharge. This study demonstrated that statistical components extracted from a discharge signal of a karst spring can provide meaningful hydrological information. Comparison with a physics-based model would however be required in order to complement this study.
No file

Dates and versions

hal-01742273 , version 1 (24-03-2018)

Identifiers

  • HAL Id : hal-01742273 , version 1

Cite

Nicolas Massei, L. Duran, Matthieu Fournier, Abderrahim Jardani, N. Lecoq. What can hydrological time series variations tell about karst dynamics? A coupled statistical/conceptual modeling analysis. AGU Fall Meeting, Dec 2015, San Francisco, United States. ⟨hal-01742273⟩
52 View
0 Download

Share

Gmail Facebook X LinkedIn More