Exploration des séries chronologiques d’analyse de la qualité des eaux de surface dans le bassin de la Tafna (Algérie) Exploration of chronological series of surface water analyses in the Tafna basin (Algeria)

Résumé

La qualité des eaux de surface est sujette à de nombreuses fluctuations et par conséquence les gestionnaires sont souvent confrontés à des pics de pollution problématiques. À cet effet, nous présentons dans cette étude les résultats d’une exploration des séries d’analyses de la qualité des eaux de surface dans le bassin de la Tafna (Algérie). Nous avons retenu trois descripteurs de la qualité des eaux; la matière en suspension (MES), la conductivité et la demande chimique en oxygène (DCO). On dispose pour cette étude de séries de prélèvements instantanés d’une fréquence au moins mensuelle, couvrant une période de sept ans (avril 1995-mars 2002). Les stations de prélèvements choisies se situent sur les principaux cours d’eau du bassin de la Tafna (Mouillah, Isser et Tafna) de manière à couvrir l’ensemble du bassin versant.

Afin d’aboutir à des conclusions exploitables, nous avons commencé, dans une première étape, par une vérification de la cohérence des résultats d’analyse, suivie d’une analyse conjointe de tous les paramètres pour explorer la structure globale des données et appréhender les principaux facteurs qui contrôlent l’évolution de la qualité physico-chimique des eaux.

Nous proposons dans cette étude deux modèles mathématiques prévisionnels : le premier est un modèle de simulation du comportement de la qualité des eaux basé sur les relations concentration-débit. Le second est une approche prédictive qui permet de mettre en évidence les saisonnalités et les tendances d’évolution des paramètres étudiés.

Mots Clés:

• eaux de surface,
• qualité,
• débit,
• MES,
• DCO,
• conductivité,
• pollution,
• séries chronologiques,
• analyse statistique,
• modélisation,
• comportement,
• prévision

Summary

A river is a complex system in interaction with the whole of its catchment area. River water acquires its physicochemical composition under the combined action of several factors: weathering of rocks, decomposition of terrestrial organic matter, diffuse and point-source pollution, and atmospheric precipitation.

The large catchment area of Tafna (7,245 km² surface area), located at the extreme northwest sector of Algeria and extending into Moroccan territory (23%), is subdivided into four sub-basins. However, the downstream portion of the basin, which represents the sub-basin of maritime Tafna, was not retained in this study; its principal river Tafna is often dry because several hydraulic constructions have been installed in the upstream part of the basin. The three sub-basins concerned then with this study are the upper Tafna, the Mouillah and the Isser.

In the basin of Tafna, surface water quality is being continually degraded. In addition to the phenomenon of erosion, which has been estimated to be 400 t/km²/year in the Mouillah sub-basin and 1,000 t/km²/year in the Isser and upper Tafna sub-basins, domestic and industrial wastes increasingly threaten the quality of the surface water.

The basin of Tafna has a surface water quality network composed of 41 sampling stations. Three stations have been selected for this study. They are distributed in the three sub-basins studied, and are located upstream of dams, so as to cover the whole of the basin: Station ST160202 in the Mouillah sub-basin, ST160426 in the upper Tafna sub-basin and ST160614 in the Isser sub-basin. This work presents the results of an exploration of a series of instantaneous samples collected with at least monthly frequency over a period of seven years (April 1995-March 2002). The descriptors used in this study are the total concentrations: suspended matter concentration (SS), conductivity, which expresses the dissolved matter concentration, and the organic matter concentration as estimated with the chemical oxygen demand (COD).

To be able to interpret the influence of the characteristics of the catchment area on the physicochemical quality of surface water, we started by describing the principal characteristics of the sub-basins studied. Given the great variability of the results of the analysis, we considered it to be useful and necessary to evaluate the coherence of the measurements with ion balance and statistical analysis (Box and Whisker plots). A joint analysis of all parameters was then applied to explore the total structure of the data in order to understand the principal factors that control the evolution of the physicochemical water quality.

This study proposes two models. The first model is a simulation of the behaviour of surface water quality based on concentration-flow relations. The second is a predictive approach that makes it possible to highlight the seasonal effects and the temporal trends of the studied parameters.

Based on this modelling, it appears that the concentration variance explained by the concentration-flow models varies between 50% and 72%, which means that the reconstitution of the concentrations is only approximate. To improve the reliability of these models, we propose several solutions.

As for the forecasting models, a comparison of the values calculated to those measured shows that the results for conductivity and the COD are close to the measured values, with average relative variations of the order of 15%, but the forecasts of SS are considered to be less precise because they produce greater average relative variations. This is explained by the fact that the suspended matter concentrations do not follow a seasonal trend but are strongly related to flow variations.

Finally, it should be noted that the reliability of the established models remains very sensitive to the size of the measurement series one has. Better information contained in longer series of analyses would undoubtedly make it possible to reduce this uncertainty.

The methodology has been suggested on a purely exploratory basis and was limited to the three descriptors of the surface water quality: SS, conductivity and COD. The work will be extended by applying it to other parameters, which should make it possible to establish a complete model for the behaviour of surface water quality in the catchment area of the Tafna.

Key Words:

• surface water,
• water quality,
• flow,
• suspended solids,
• COD,
• conductivity,
• pollution,
• time series,
• statistical analysis,
• modelling,
• forecast