Explanatory note on BRGM's expert appraisal of the Deux-Sèvres substitution reserves project - Report RC-71650 (June 2022).
13 February 2023

General Framework

This appraisal was requested and financed in 2021 by the Société Coopérative Anonyme de l'eau des Deux-Sèvres (or COOP 79), the project owner for the creation of substitution reserves in the Sèvre-Niortaise-Mignon basin. The objective was to simulate the effects of a new abstraction scenario proposed by COOP 79. The study was carried out with the ultimate aim of providing input for an impact assessment.

In 2013, 2016 and 2019, BRGM had already been asked by COOP 79 to assess the impact of its successive substitution-reserve projects. In 2021, following the appeal lodged by opponents' associations against the previous prefectural decrees, the Poitiers Administrative Court requested a modification of the authorised volumes, whence the need to take a new scenario into account.

A committee was set up to monitor the BRGM study. It consisted of the following entities: DDT79, Chambre d'agriculture des Deux-Sèvres, Coordination Marais Poitevin (France Nature Environnement), Département des Deux-Sèvres, Établissement Public du Marais Poitevin, Agence de l'eau Loire Bretagne, Fédération de pêche, ARS, Syndicat des eaux du Vivier, and several mayors.

The report by BRGM is not an in-depth study, nor is it an impact assessment of all the possible consequences of the planned water abstractions. Nor is it a scientific research article subject to evaluation by the scientific community. It is a study responding to a specific order, giving rise to a technical report that answers the questions posed within the corresponding limits.

The study carried out consists in simulating the proposed water abstractions and their impact on the level of groundwater and on the flow of watercourses, by means of a model. The hydrodynamic model used by BRGM is the only one currently available to meet the contractual objectives. The scenarios tested were provided by COOP 79.

The model used is a regional model called "Jurassic" (a meshed hydrodynamic management model carried out on the Jurassic aquifer). It was developed in the context of previous BRGM public service studies (reports RP-59288- FR and RP-64816-FR). It is based on the period 2000-2011. Strictly speaking, this reference period does not allow for recent, let alone future, weather conditions to be taken into account. However, it enables us to assess what would have happened if the substitution reserves had been put in place during the period 2000-2011, bearing in mind that these years are representative of contrasting weather situations (wet and dry years). It is this model that contributed to the studies used as input for the authorisation application file under the Water Act in 2014 and again in 2016-20. COOP 79 has no authority to update the model. This would require substantial resources that have not been available in recent years. Based on public funding allotted in 2021, this update covering the period 2000-2020 is now underway and should be completed by the end of 2024. Depending on the needs expressed, it could possibly be used for new simulations including scenarios of the impact of predicted climate change.

For the assessment, the results of which were published in July 2022, two simulations were carried out: a simulation of the COOP 79 storage scenario and a simulation of the storage scenario combined with the other known projects in the study area, with a comparison to the simulated reference state (data for 2000-2011) and the results obtained in 2016 and 2019.

The simulations were therefore carried out and calibrated on the reference period 2000-2011 (the same as for the previous studies), which includes characteristic wet or dry years, a period which the monitoring committee considered to be relevant and scientifically representative (with the exception of the climate change factor).

The expected results focused solely on identifying the impact of these abstractions for storage, on groundwater levels, groundwater/river exchanges and river flow. The fate of the water in the reserves (use, evaporation, etc.) is outside the scope of the current simulation and was not one of the questions put to BRGM.

Nor is BRGM responsible for the planned use of substitution reserves, the corresponding volumes and thresholds. These elements are the responsibility of COOP 79 and the relevant government authorities, based, among other things, on available expert-reports, including that of the BRGM. Climate change, which was not simulated in the study, is nevertheless an important factor which should be taken into account. The reason is that recurring periods of winter drought could repeatedly lead to groundwater levels being below the regulatory thresholds, which would compromise the filling of reserves in certain years.

Focus on a few technical issues

BRGM considers that the model can answer the questions posed by COOP 79 at the scale of the Sèvre-Mignon-Courance catchment area. The model is based on a division of the space into 1 km x 1 km mesh elements. The study perimeter, which covers more than 2,000 surface mesh elements (2,000 km2), is valid for a global approach to the catchment area.

However, the model as it stands (without a finer mesh) cannot address specifically local issues such as a wellfield on a few mesh elements, or the impact of a single abstraction. But it is possible to use the results of the model on the scale of the Sèvre-Mignon-Courance basin, which provides a relevant and useful approach to inter-basin relations and the cumulative effect of abstractions.

The tool best suited to answering the question of the cumulative impact of abstractions on a basin is the meshed hydrodynamic model, which was used for BRGM’s expert appraisal. It allows for the location of abstractions and the corresponding volumes to be taken into account. The results are analysed at two scales of information:

  1. the piezometer (water level measuring station, used to estimate the impact at a given point and for a given period);
  2. mapping of the difference in water levels between two simulations, which provides an estimate of the impact of a project on the whole model, but for a given date.

Cross-referencing this information gives an overall view of the results of the simulations.

The report does not make a small-scale analysis of the results (i.e. finely at given locations); if the results are discussed with respect to piezometer measurements, it is because these are the only measured points that allow a time-specific visualisation of the results. The results included in the synthesis and in the conclusion are mainly those at the outlet of the catchment area, as they integrate the phenomena at the scale of the whole catchment area.

The model takes into account the main hydrographic system. The flows of this main network integrate the flows of the small streams that flow into the main network. The watercourses, represented in a simplified way (an intrinsic principle of modelling) are in proportion to the water table (the groundwater-river exchange surface is fully integrated).

To integrate elements at a more local scale, and therefore with a finer mesh, and for this to be justified, it would be necessary to:

  • have finer spatial and temporal data and, in particular, to have suitable measurement records on which to base them;
  • know with the same level of precision the different physical processes at work (e.g.: functioning of wetlands or peat bogs).

There are questions about the margin of error of the model and how it is taken into account in the assessment carried out. The 2022 report does not comment on the discrepancy between two simulations and the actual observed data in the records, nor is this an uncertainty inherent in the model. Indeed, one should not confuse uncertainties in the strict meaning of the term (margin due to uncertainties on input parameters) with calibration quality (difference between simulated and observed data), or comparison between simulated scenarios.

One sentence in the 2022 report may have been confusing: "It is difficult to calculate the margin of error of simulation models. However, in this study it can be considered that a difference in piezometric load between two simulations of less than 2 centimetres is within the margin of error of the model. Similarly, a difference in streamflow between two simulations of less than 5 L/s falls within the model's margin of error”.

The "2 cm" and "5 L/s" mentioned represent the difference between the two simulations, above which BRGM considers that the model reacts to a change in parameters, in this case a change in abstractions. In other words, a difference of less than 2 cm would not be significant. This value is empirical and hence limits the interpretation of the results. If the model reveals a difference between two simulations, regardless of the quality of the calibration, this difference represents a trend, especially if the piezometric dynamics are well represented, although the values should be taken with caution. The quantification was done by comparing the simulation and the reference, through graphs and maps.

This statement is also independent of the mesh size of the calculation and the type of model.

The model is currently considered to be well calibrated with a good reproduction of the observed piezometry values. Water budgets (the relationship between the inflow and outflow of water in a given system over a given time interval) were not included in the report as this level of detail was not expected. However, the budgets and convergences were checked in the most recent update.

A model updated up to 2020 or 2021 should be available by the end of 2024 (reduction of the mesh size, extension of the simulation period, modification of the calculation time steps, reworking of the geometry).

Increasing the spatial resolution of the model (mesh size) implies completely reworking the geological model on which the hydrogeological model is based, retracing the watercourses, etc. BRGM has planned 18 months of work on these subjects in the updating programme.

Integrating new years in the simulations implies integrating all the abstractions corresponding to this period. It should be noted that these data are only available one year later at best. This requires precise cross-referencing of the various data sources. It is possible to increase the temporal resolution of the model (time step) but this requires knowing how to distribute the abstractions over the period. For the new update, this aspect will be taken into account (daily water budgets and more precise hydrodynamic calculation time steps).

It will then be necessary to perform a new calibration stage (to verify the piezometry and flow data that have already been assimilated in the model, to make adjustments if necessary, to add and calibrate new points and new knowledge).

BRGM has also provided all or some of the model’s parameters to various organisations, under corresponding agreements. The model used in this study is also available online on the AQUI-FR platform.