Vulnérabilité de l’eau souterraine à la contamination par les nitrates sur le bassin versant de la rivière Noire (Montérégie, Québec) Groundwater vulnerability in the Noire River watershed (Montérégie, Québec)

Résumé

La rivière Yamaska est l’un des affluents du Saint-Laurent les plus contaminés par les activités agricoles. Cette problématique touche particulièrement le sous-bassin de la rivière Noire où les dépôts de surface du Quaternaire sont discontinus, de faible épaisseur et souvent perméables. L’objectif de cette étude est de déterminer la vulnérabilité de l’eau souterraine sur une partie du sous-bassin de la rivière Noire (100 km²). La méthodologie utilisée comprend la caractérisation de l’aquifère, l’analyse des concentrations en nitrates et des contenus en δ¹⁸O, l’étude de la vulnérabilité en utilisant l’approche AQUIPRO et la modélisation hydrogéologique. Les résultats montrent une dégradation significative et d’origine anthropique de l’eau souterraine : plusieurs concentrations excèdent 1 mg N-NO₃/L et quelques-unes excèdent 10 mg N-NO₃/L. Les puits contaminés sont situés sur les crêtes topographiques où le silt argileux est absent et le till discontinu, et où la plus grande vulnérabilité AQUIPRO a été identifiée. Une diminution des concentrations avec la profondeur de prélèvement s’explique par un écoulement souterrain peu profond entraînant les nitrates vers un ruisseau et vers la rivière Noire. La vulnérabilité de l’eau souterraine est ainsi beaucoup plus grande dans les couches superficielles du roc fracturé. Le temps moyen de séjour de l’eau souterraine est estimé à 20 ans. Les concentrations mesurées permettent d’établir un lien direct entre la vulnérabilité, les dépôts de surface, la stratigraphie du substrat et les directions de l’écoulement souterrain. Elles démontrent aussi la présence d’une contamination de l’eau souterraine par les nitrates qui est susceptible d’augmenter si aucune mesure préventive n’est mise en place.

Mots clés:

• vulnérabilité,
• eau souterraine,
• nitrates,
• contamination,
• agriculture,
• AQUIPRO,
• modélisation hydrogéologique,
• Québec

Abstract

Many studies have shown a link between intensive agricultural practices and groundwater pollution by nitrates. In Québec, recent studies have shown that the Yamaska River is highly contaminated by agricultural activities. Maize and pork production are particularly intensive in the Noire River sub-basin. In this area, quaternary deposits are discontinuous, of limited thickness and are generally permeable, leading to high groundwater vulnerability. The objective of this study was to determine groundwater vulnerability to nitrate contamination on a small agricultural sub-watershed of the Noire River. The methodology included aquifer characterization, analysis of nitrate concentrations and δ¹⁸O composition, as well as a vulnerability evaluation and groundwater flow modelling.

Located on the south shore of the Saint Lawrence River, the Noire River (1579 km²) is located in the southeastern portion of the Yamaska basin. A small part of the basin (100 km²) was the focus of this study. The Noire River flows in the centre of the study area whereas the Aulnages creek is a small tributary to the Noire River. The study area was limited to the east and west by topographic highs. It is located between the limit of the St. Lawrence Lowlands and the first Appalachian ridges. The bedrock, Cambrian to Ordovician in age, is mainly made of limestone in the western zone and is composed of terrigenous siliciclastic facies in the eastern zone. The substratum forms elongated crests, due to the tectonic grain, surrounded by recent surface deposits. These quaternary deposits are discontinuous and are of limited thickness. The hydrological potential of the fractured rock aquifer is interesting but spatially variable.

The deposits were analyzed at 50 observation sites and 18 typical samples were sieved or submitted for density analysis (GEOTERAP). Data from the Système d’Information Hydrogéologique (SIH) were used to complete the field information in establishing the stratigraphy of the area. Soil hydraulic conductivity was measured using a Guelph permeameter and two short-term pumping tests were performed. Monthly water levels were measured in 18 private wells from June 2001 to June 2002. Water was sampled bimonthly from 35 sites (25 deep wells, two shallow wells, two sites in the Noire River, four in the Aulnages stream, and two in a temporary lake, located in a gravel pit). In October 2001 and April 2002, water was sampled for δ¹⁸O composition. Analyses were performed at the GÉOTOP-UQÀM-McGill laboratories. Aquifer vulnerability was determined using the AQUIPRO approach, a simple method that considers clay and till thickness, in addition to well depth. A groundwater flow model was developed using MODFLOW and MODPATH to simulate groundwater flow, flow paths and residence times.

Characterization of the quaternary deposits confirmed the following sequence, from the base to the top: till; clayey silts; sand and sandy gravel. The thickness of these deposits was variable, and there were bedrock outcrops, mainly on the western and eastern sides of the study area. A north-south esker (partially exploited) is present on the western side of the Noire River. Measured soil hydraulic conductivities show that the clayey silts and till deposits have low permeability whereas the sand is more permeable. Clayey silts present in the centre of the study area, as well as occasional till deposits, offer partial protection to the aquifer, whereas elsewhere infiltration and contaminants can reach the aquifer through recharge. Pumping tests showed high transmissivities for the rock formation, probably resulting from the important rock fracturing in the upper 10 to 15 m of the aquifer. Groundwater flowed towards the river from the eastern and western limits of the study area, and water depth was on average 2.5 m. The Noire River is generally in contact with rock and drains the aquifer.

Measured nitrate concentrations exceeded 10 mg  N‑NO₃/L (guideline for potable water) in the two sampled surface wells. Concentrations were between 1 and 10 mg N-NO₃/L at least once during the study period in ten of the 25 deep sampled wells, indicating a groundwater contamination problem of anthropogenic origin. The highest concentrations were measured in recharge areas and nitrate concentrations were found to decrease generally with well depth. Nitrate concentrations were higher in the Aulnages creek than in the Noire River, probably because the creek intercepts drainage water and shallow groundwater flow. The isotopic composition of surface wells, deep wells and Aulnages creek water remained relatively stable between sampling times. This indicates an important mixing of fresh recharge with in situ groundwater. In the Noire River, δ¹⁸O compositions reflect the yearly variations in precipitation.

AQUIPRO aquifer vulnerability was highest in the eastern and western parts of the study area where the highest nitrate concentrations were measured. AQUIPRO vulnerability was lowest where the clayey silts provide some protection near the Noire River. The results showed an important spatial heterogeneity in the thickness of the clayey silt layer, underlining the generally high aquifer vulnerability in the study region. The groundwater flow model developed using field data simulated the measured heads adequately. Flow directions simulated with MODPATH confirmed the presence of a shallow groundwater flow from the eastern and western limits and towards the Noire River. This flow is probably responsible for the decreasing nitrate concentrations observed with increasing water sampling depth and confirms that aquifer vulnerability varies with depth in the aquifer. The average groundwater residence time is 20 years. This means that recharge will take on average 20 years to travel through the aquifer and to the Noire River. It also indicates the time frame required for the aquifer to eliminate a large-scale nitrate contamination, after the initiation of control measures.

This work showed that groundwater nitrate contamination is related to groundwater vulnerability, which is a function of quaternary deposits, substrate stratigraphy and groundwater flow directions. These factors must be considered when studying groundwater vulnerability as they directly affect contaminant transport. It therefore appears necessary to use a combination of various approaches to better understand aquifer vulnerability and to design preventive measures. This work also demonstrated groundwater contamination by nitrates in the study region. Because of the generally high vulnerability of the aquifer, increased nitrate contamination can be expected in the future if no preventive measures are undertaken to protect the groundwater resource.

Key words:

• vulnerability,
• groundwater,
• nitrates,
• contamination,
• agriculture,
• AQUIPRO,
• groundwater modelling,
• Province of Quebec