Nitrates fate in a small basin: combined use of kriging and physically-based model

N. Flipo (1), N. Jeannée (2), M. Poulin (1), E. Even (1), S. Théry (3), C. Viavattene (1), P. Combes (1), E. Ledoux (1)

(1) CIG, Ecole des Mines de Paris, 35, rue Saint-Honoré, 77305 Fontainebleau, France Nicolas.Flipo@ensmp.fr
(2) Geovariances, 49 bis, avenue Franklin-Roosevelt, 77212 Avon, France
(3) UMR Sysiphe, Université Pierre-et-Marie Curie Paris VI, Boite 123, 4 place Jussieu, Tour 56, Couloir 46-56, 3° étage, 75252 Paris Cedex 05, France

Abstract

Based on the experience concerning hydrogeological modelling and nitrate migration in aquifers, and on different research done on the ecological modelling of surface water, a hydrodynamic and biogeochemical model of river, ProSe, was coupled with a hydrogeological model, NEWSAM, in order to obtain the CAtchment Water Quality Simulator (CAWAQS). The hydrological part of the physically-based meta-model was fitted and validated in terms of discharge in streams and head in aquifers.

CAWAQS is being coupled with the agronomic model, STICS, in order to model nitrates migration in each component of the basin: root zone, unsaturated zone, aquifers, wetlands and rivers. STICS simulates the nitrate concentrations coming from the soil horizon to unsaturated or saturated zone. Transport and biogeochemical transformations inside the basin are simulated by CAWAQS.

The main objective of this work is to understand nitrate fate in small basin (1250 km2). The main difficulty is to obtain a satisfactory representation of aquifers nitrate contamination from punctual observations in piezometers and wells. Nitrate data are coming from punctual measurements, which may not be representative of large areas. A first problem is to extrapolate them at higher scale than this of observation. A second problem is that each data is only representative of the aquifers where it is taken from. This information is often missing. The third problem is that geological layer is composed of several aquifers and discretization of the model is based on the geological structures, not directly on the aquifers themselves. So we do not really know what a single value does mean at the mesoscale.

On this complex context, the simultaneous use of two analyzing tools, the kriging and the physically based modelling, is a good way to estimate nitrates pattern in aquifers. Preliminary exploratory Nitrate data analysis and kriging were done with the Isatis® software.

The basin is discretized in 8513 simulation cells. On the one hand kriging takes into account the spatial structure of nitrate concentrations in order to give for each cell an estimate of nitrate concentration with an associated error due to the spatial variability of the data. On the other hand, the physically based simulation with CAWAQS gives nitrate concentration in each cell due to modelled physical processes. The confrontation of the two different approaches improves our understanding of both the transport and transformation processes and the resulting spatial pattern of nitrates.

This work presents the use of these two tools in a small catchment from 1977 to 1988. The kriging is used to initialize the physically based modelling. Critics and results of the confrontation of the two approaches are given.

Poster (PDF - 1.5 Mb)
Article: Assessment of nitrate pollution in the Grand Morin aquifers (France): Combined use of geostatistics and physically based modeling

Visit VIIth IAHS Scientific Assembly website