Contexte géologique et hydrogéologique
D’un point de vue géologique, la région se situe à la jonction de deux ensembles majeurs : la province de Grenville et la plate-forme du Saint-Laurent. Le substrat rocheux du Bouclier canadien (province de Grenville) est principalement constitué de roches ignées intrusives (anorthosite, granite) et gneissiques datant de l’ère précambrienne ayant subi plusieurs épisodes de métamorphisme de haut grade (Rondot, 1989). La région de la Haute-Côte-Nord, quant à elle, présente en majorité une couverture de roches migmatiques et de roches intrusives de type gabbro (Cousineau et al., 2014, Figure 1.7). Les roches sédimentaires de la plate-forme du Saint-Laurent (grès, calcaire et shale) se retrouvent essentiellement sur les Basses Terres au sein et aux périphéries de la vallée de la Malbaie (Cousineau et al., 2014).
L’astroblème de Charlevoix, issu de l’impact météoritique survenu à la fin du Dévonien (Rondot, 1989), constitue un trait morphologique caractéristique de la région de Charlevoix.
D’un diamètre de 56 kms, celui-ci a pour point culminant le Mont des éboulements à 768 m d’altitude.
La région est également caractérisée par un dense système de failles normales et de fractures bien développées. Ce dernier serait principalement associé au cratère météoritique d’impact ainsi qu’à l’épisode de rifting à l’origine de l’ouverture de l’océan Iapétus le long des failles du Saint-Laurent et de Logan, qui représentent des linéaments majeures dans la région (Lemieux et al., 2003; Rondot, 1989).
Les dépôts de surface quaternaires recouvrant le socle rocheux résultent de la dernière glaciation wisconsinienne. Le till est constitué de particules de roches arrachées par le glacier qui se sont déposées directement sur le socle rocheux. La fonte de la glace a permis la formation de rivières où des sédiments fluvioglaciaires (sable et gravier) se sont déposés.
Lors du retrait du glacier l’avancée de la mer de Goldthwait et la mise en place de lacs proglaciaires ont également déposé des sédiments marins, glaciolacustres et glaciomarins.
Les cours d’eau actuels déposent des sédiments alluviaux qui recouvrent les sédiments plus anciens (Figure 1.8).
Le système aquifère régional est constitué de dépôts quaternaires et de roc fracturé. Le till, recouvre en surface 70% du territoire et est en discordance sur le roc (Tableau 1.3). Les propriétés du till dans cette région peuvent être très variables. Il est parfois caractérisé par une matrice sableuse à structure lâche ponctuée par endroit de blocs métriques. Ce type de till est généralement peu épais. Le till peut également être plus compact et d’épaisseur plus importante (jusqu’à 8 m) confinant les aquifères sous-jacent. Le potentiel aquifère du roc est conditionné par son degré de fracturation. Sa transmissivité varie de 10-3 à 10-7 m²/s pour une moyenne de l’ordre de 10-5 m²/s (Richard et al., 2014).
On distingue une répartition différente des dépôts entre les deux régions (Charlevoix et la Haute-Côte-Nord). En Haute-Côte-Nord, les hautes terres sont dominées par des dépôts de till et de sable le long des cours d’eau. Les basses terres correspondant à de grandes plaines fluviales, sont constituées d’argile recouverte par d’épais dépôts de sable d’origine deltaïque très perméables sur lesquels reposent souvent de grandes tourbières (Cousineau et al., 2014). La succession des dépôts est plus complexe dans la région de Charlevoix.
Associée à la moraine de Saint-Narcisse, elle présente des dépôts fluvioglaciaires et glaciolacustres en plus grande proportion (environ 20%). Ces derniers constituent un potentiel aquifère majeur dans la région. Cependant, leur étendue est généralement limitée et discontinue. Des sédiments glaciomarins se sont aussi accumulés dans les fonds de vallées (Cousineau et al., 2014, Figure 1.8).
L’écoulement régional des eaux souterraines est régi par le gradient topographique, soit du NW (Laurentides) vers le SE en direction du Saint-Laurent, sauf la rivière petit Saguenay qui a comme exutoire le Saguenay.
Le contexte hydrogéologique peut également être défini en termes de type d’aquifère : libre, semi-captif ou captif en tenant principalement compte de la nature des dépôts. Ainsi, les sables et graviers correspondent à des aquifères à nappe libre alors que l’argile et les dépôts organiques (de type tourbière) confinent les aquifères sous-jacents.
Les propriétés hydrauliques du till et du roc étant très aléatoires, ces derniers sont associés à des conditions semi-captives (Figure 1.9). Cette distinction permet de prévoir l’intensité spatiale de la recharge. Les dépôts perméables de surface tels que les dépôts fluvioglaciaires, constituent des zones préférentielles de recharge alors que les zones recouvertes d’argile, et de dépôts de till sont propices au ruissellement de surface.
Estimating groundwater recharge is crucial to ensuring the proper management of aquifers. In this study, net regional recharge and spatial potential recharge are estimated at four watersheds within the Charlevoix-Haute-Côte-Nord (CHCN) area, Quebec Province, Canada. The regional aquifer system is characterized by quaternary sediments (mainly composed of till) and fractured bedrock.
In this study, four methods are applied according to available data. The first two approaches are regional water budget methods. The methods differ in the estimation of vertical inflows (VI), which is estimated from two hydrological models : GR4J (Valéry, 2010) and HYDROTEL (Fortin et al., 1995, 2007; Poirier et al., 2014). The third method estimates potential recharge spatially over the study area. Finally, the streamflow data are analyzed using the Echkardt (2008) baseflow separation method to obtain an estimate of recharge, assuming that discharge is equal to recharge (Centre d’expertise hydrique du Québec (CEHQ), 2010).
According to the results of all investigated methods, the mean annual recharge for the CHCN region is approximately 183 mm, which is 18 % of the total yearly precipitation (P). The discussion highlights large uncertainties due to the assumptions of each method and the reliability of the data.
Recharge is generally defined as the amount of water that reaches the saturated zone and adds to the groundwater reservoir (Scanlon et al., 2002; Vries & Simmers, 2002).
The estimation of groundwater recharge represents a major challenge that is crucialfor both hydrogeologists and hydrologists. From a hydrogeologist’s point of view, groundwater recharge is the main input that quantifies the state of the resource. For a hydrologist, recharge is water that is temporarily lost because it does not immediately contribute to direct runoff to the river. An accurate estimation of the groundwater recharge is a key to constructing acceptable groundwater flow models. Although it is possible to assess the recharge from in situ point measurements of groundwater level fluctuations of an aquifer through extensive field surveys, available databases for groundwater levels are very sparse, particularly for remote areas. In addition, because of its high variability over time and space, calculating recharge directly from local instruments is nearly impossible.
Therefore, recharge is often deduced from indirect approaches. These approaches still require large databases and the use of many simplifying assumptions. Moreover, the uncertainty related to indirect approaches is difficult to evaluate. Therefore, the joint use of several methods to provide a reasonable estimate of recharge is justifiable (e.g., Rivard et al., 2014). Many techniques are available to quantify recharge. Recharge can be broadly categorized as water budget approaches, water table fluctuation approaches, streamflow analyses, chemical tracing, physical techniques and numerical modelling.
This paper aims to compare four methods of estimating recharge and infiltration through case studies in Charlevoix and Haute-Côte-Nord (CHCN), which are located in Quebec Province in Canada. The results are discussed in the context of uncertainties and hypotheses inherent to each method. Such comparisons allow for validation of a value and a range of values of recharge. Recommendations regarding the use of each method for estimating recharge can be provided. This study is part of a much broader research program called PACES (‟Programme d’Acquisition de Connaissances sur les eaux Souterraines”- Groundwater Knowledge Acquisition Program), initiated by the Ministère du Développement durable, de l’Environnement et des Parcs (MDDEP)- Minister of Sustainable Development, Environment and Parcs.
The goal of this program is to survey groundwater quantity and quality throughout Quebec in populated territories and to obtain an overall understanding of this resource. The PACES projects intend to gather and compile data on groundwater for establishing groundwater geodatabases (Chesnaux et al., 2011). The PACES projects also aim to estimate the availability of groundwater for better management of this resource. These PACES projects have conducted many field campaigns for hydrogeological data acquisition. The PACES program is being conducted separately for different areas of the province, including in the Charlevoix and Haute-Côte-Nord regions.
Several techniques are available for assessing the recharge of aquifers. Each technique possesses strengths and weaknesses.
Water budget approaches are widely used to estimate groundwater recharge. These approaches include most physically-based hydrological models, such as SWAT, HELP and MOHYSE (e.g., Arnold et al., 2000; Croteau et al., 2010; Rivard et al., 2014). Seepage meters or lysimeters allow for direct measurements of real time vertical infiltration, but their use is limited to point estimations. These installations may also perturb the local soil and modify the rate of infiltration. In addition, such measurements require costly material, careful maintenance and extended field campaigns, which reduce their implementation to a few points in a watershed that can cover thousands of square kilometers.
Investigations of the fluctuations in the water table are among the most popular methods for estimating groundwater recharge and are mostly suitable for areas with a shallow water table. Such methods are easy to apply and inexpensive (e.g., Healy, 2010; Yin et al., 2011). Difficulties in applying these methods are related to determining the specific yield, which depends on the fluctuation of the level in the water table. Streamflow time series analysis can also be used to assess groundwater discharge to a river. Hydrogram separation and recession analysis are two approaches that rely on streamflow time series analysis (e.g., Meyboom, 1961;Domenico & Schwartz, 1998; Rutledge, 2007). In this framework, it is assumed that the discharge is equal to the recharge (Bredehoeft, 2007). Analytical models can also be used to describe groundwater flow and
to estimate recharge. For example, Chesnaux (2013) estimated the recharge of an unconfined fractured bedrock aquifer in the Canadian Shield based on the assumption of a Dupuit-Forchheimer flow type (Dupuit, 1863). Numerical modeling of groundwater flow in the vadose zone can determine infiltration by solving Richard’s equation (van Dam & Feddes, 2000; Gogolev, 2002) for a 1D model of vertical infiltration. UNSATH, VS2DT, SWAP and HYDRUS 2D (Scanlon et al., 2002) are among the most popular commercial software used for this purpose. Extensive field studies are necessary to obtain the data required by numerical models. Consequently, the study area must be small, and many uncertainties remain, particularly regarding the relationship between hydraulic conductivity, water content and pore pressure.
Finally, the methods based on tracer tests can also be used to estimate groundwater recharge (e.g., Flint et al., 2002; Scanlon et al., 2002; Vries & Simmers, 2002). Several tracer-based methods have been proposed: isotopic analysis (Allison et al., 1994), heat tracers (Flint et al., 2002), historical tracers (e.g., tritium/helium couple or Chlorofluorocarbons), (Delin et al., 2007; Coes et al., 2007) and environmental tracers, e.g., chloride mass balance (CMB, Flint et al., 2002). All of these methods are generally costly and only manageable for a small controlled study area (Scanlon et al., 2002). In this study, four methods are selected and compared for estimating recharge based on available data in the PACES groundwater database. The first two methods are based on a global water budget (WB1 and WB2), but they differ in the vertical inflows (VI) estimations (Fortin et al., 1995; Perrin et al., 2003; Valéry, 2010; Poirier et al. 2012). The third method is a local water budget (WB3) (Lim et al., 2006; Monfet, 1979). The fourth method is based on hydrogram separation (Eckhardt, 2005, 2008; CEHQ, 2010). These methods are applied on different time and space scales depending on the availability of climatic data (temperature, T; precipitation, P) and hydrological data (Streamflow, Q). WB1, WB2 and the hydrogram method provide a net recharge assessment for four watersheds over the 1975-1995 period, whereas WB3 provides a local assessment of potential recharge (infiltration) over three watersheds for the 1989-1992 period. This paper is organized as follows: Section 2 provides a description of the study area, and the methodology is detailedin Section 3. The results are presented and discussed in Section 4. Conclusions and recommendations regarding future work are presented in the final section.
Description of the study area
The four watersheds considered in this study are located on the north shore of the St. Lawrence River in Quebec Province, Canada (Figure 2.1). The surface areas of these watersheds range from 712 to 3 085 km². The river channel network is generally dense.
Figure 2.1 shows that this portion of Quebec comprises two major rivers: the Saguenay (Fjord) and St. Lawrence Rivers. The topography of the region is dominated by the Laurentides Range, the meteoritic impact crater of Charlevoix, and deep valleys, all of which create a steep landscape.
The climate of the region is cold and humid. The mean annual precipitation (water equivalence) is approximately 1020 mm/y. The forest is composed of mixed broadleaved trees and evergreens and cover 80 % of the study area. The average temperature ranges from -25°C in winter and 25°C in summer (Figure 2.2). In winter, precipitation is solid, and the streamflow is low. When the temperature begins to rise, the snow cover melts and causes a spring flood. In summer, the high temperature triggers evapotranspiration, inducing low flows in rivers. Heavy rainfall events during autumn occasionally induce high flows.
From a geological point of view, the region of the Charlevoix –Haute-Côte-Nord is located between two major units: Grenville Province and the St. Lawrence Platform of the Canadian Shield. The substratum is mainly composed of crystalline granitic and gneissic rock. The bedrock in the northern part of the study area is mainly represented by migmatites. Rock outcrops are observed across a large portion (10 %) of the territory, particularly on highlands (Table 2.1). Quaternary sediments result from the last glaciation and deglaciation episodes. Mapping of these surface deposits has been performed over the CHCN region (Cousineau et al., 2014). These sediments have glacial, glaciofluvials and glaciomarine origins. Following the glacier withdrawal, the Goldthwait Sea progressed inland and deposited marine sediments. The Till glacial sediments cover approximately 70 % of the study area (Figure 2.3 and Table 2.1).
The regional aquifer system is characterized by quaternary sediments and fractured bedrock. A layer of till covers most of the area, and it is in direct contact with the bedrock.
The layer was formed by rock particles that were dragged and carried by glaciers and then deposited directly onto the bedrock. The till is characterized by variable and complex structures. Some till it is composed of a sandy matrix with larger metric blocks. Till can also be represented by a compact matrix, which confines the underlying bedrock aquifer.
The potential aquifer property of the bedrock depends on the density of its fracture network. However, no significant associations between transmissivity and structural features, such as faults or the meteoritic impact and geologic formations have been observed in this study area. The average transmissivity of the crystalline bedrock is on the order of 10-5 m²/s (Richard et al., 2014b).
The spatial sequence of quaternary deposits is complex throughout the region. The Haute-Côte-Nord region is dominated by till deposits at the highest elevations, with sand along the streams. Lowlands correspond to large fluvial plains where clay sediments are covered by thick deltaic sand deposits confined by peatland. With the Saint-Narcisse moraine, the Charlevoix area comprises glaciofluvial and glaciolacustrine deposits (approximately 20 %) in the valleys. These deposits represent important potentially productive aquifers. However, the deposit extent is limited and discontinuous.
Glaciomarine sediments also accumulated at the bottom of steep-sided valleys, such as those along the Du Gouffre and La Malbaie Rivers (Cousineau et al., 2014). This quaternary history has resulted in the formation of different types of aquifers with various confining conditions (Figure 2.3).
Recharge assessment methods
In this study four methods have been selected and applied on four watersheds into the CHCN. These methods are applied on different time and space scales depending on the availability of climatic data (temperature, T; precipitation, P; and the streamflow, Q).
This method consists of applying the principle of mass conservation to the hydrological cycle. This principle states that the difference between the input and output fluxes of water in an aquifer system is equal to the change in water storage. Recharge is thus obtained by summing all the other terms in the water balance (Equation 1):
Global water budget (WB1 and WB2)
The global values of recharge provided by WB1 and WB2 are computed for the period 1975-1995 and for four watersheds where data are available. These two methods differ in the estimation of VI. For WB1, VI is calculated from the global conceptual hydrological model, GR4J, which was developed by Perrin et al. (2003) using CemaNeige (Valéry, 2010) as the snowmelt module. For WB2, VI is calculated from physically-based distributed hydrological model HYDROTEL (Fortin et al., 1995). The VI from HYDROTEL was provided by the CEHQ (Centre d’Expertise Hydrique du Québec; Poirier et al., 2014). Both models were run with a daily time step. The vertical inflows produced from these two models use gridded climatic data (precipitation and temperature) were generated by the Centre d’expertise hydrique du Québec (Poirier et al. 2012). The interpolation has been performed at a daily time step, at a 0.1 degree resolution over the southern Quebec territory (Figure 2.1). Simple isotropic kriging was performed, with monthly mean variograms for precipitation, minimum temperature and maximum temperature. CemaNeige, which is used for WB1, is a daily degree snowmelt module that simulates the daily evolution of snow cover. The climatic interpolated data are averaged over each watershed before being added to GR4J with CemaNeige, since GR4J is a global model. The model comprises a total of six free parameters that must be calibrated against the observed streamflow using a portion of the hydroclimatic data reserved for this purpose. Two of these parameters, kf and Ctg, are only related to snowmelt. These last parameters represent the snowmelt parameter and the correction of the thermic state of the snow cover, respectively. One of the particularities of CemaNeige is the division of each watershed into five altitudinal zones to calculate the median altitudes of each zone as inputs to CemaNeige. According to the recommendation of Valéry (2010), five parameters are fixed: the mean annual quantity of snow (QNBV, in mm), the altitudinal correction of precipitation, the altitudinal gradient of temperature (-0.6°C/100 m), the snow melt temperature (0°C) and Vmin, which is the percentage of Kf that corresponds to the minimal snowmelt speed (10 %). CemaNeige separates the total precipitation into rain and snow according to the median altitude of the watershed for each altitudinal zone. Snowmelt in each zone is delayed according to the thermic state of the snow cover. Finally, the quantity of water from snowmelt (Sm), which is added to the rainfall (Pl), is calculated according to Equation 2.
Local water budget (WB3)
The third water budget variant estimates the spatial infiltration (on a 250 m × 250 m grid) by a computation implemented in ArcGIS (ESRI ArcGIS 10.1, 2012). This method has been developed with the PACES team for of calculating the DRASTIC index (Aller et al., 1987) over the study area. This last water budget (WB3) is determinated from the VI provided by the CEHQ. The AET is estimated from the Budyko relation (Budyko, 1974) and is only applied for the summer (used for WB2). The surface runoff computation is based on the Soil Conservation Service (SCS) Curve Number, (CN) method (Cronshey, 1986). This approach aims to account for the characteristics of each mesh of the watershed grid (topography, soil coverage and deposits). Nevertheless, it does not allow for the estimation of subsurface runoff. Consequently, the resulting maps correspond to an infiltration map or a map of a potential recharge zone. In the present study, this method is only performed for the Du Gouffre, La Malbaie and Petit-Saguenay watersheds because the soil coverage information is not available for the Portneuf River watershed.
The CN method has been adapted by Monfet (1979) for humid southern Quebec to match meteorological conditions, the geological context and the nature of the soils. The CN is a quantitative parameter that depends on soil properties (nature, grain size, texture and infiltration capacity), the type of soil coverage (vegetation and land use), the topography of the watershed and the hydrologic conditions, which must be determined from a field study. Such a large investigation would not be feasible for the entire study area. Consequently, the mean hydrologic conditions for all studied watersheds are applied.
A streamflow hydrogram can be divided into three main types of flows: sub-surface or hypodermic flow, baseflow and surface runoff (Domenico & Schwartz, 1998). The hydrogram separation is possible assuming that the flow component presents different time lags (Eckhardt, 2008). Runoff is related to rapid flow: its stream contribution stops earlier. Baseflow has a long time lag, and it continues after the hypodermic flow stops.
During low flow, the streamflow mostly consists of baseflow. This baseflow component corresponds to groundwater discharge into a river.
Several authors focused on streamflow records to estimate groundwater recharge (e.g., Meyboom, 1961; Rorabaugh, 1964;Mau & Winter, 1997; Rutledge, 1998; Lee et al., 2006). In this context, recharge is assumed to be equal to the baseflow. Various methods can be used for hydrograph separation, including graphical approaches (Linsley et al., 1982; Brodie & Hostetler, 2005), digital filtering with programs such as HYSEP or PART (Arnold et al., 1995; Brodie & Hostetler, 2005; Rutledge, 2007) and recession-curve displacement methods (Rorabaugh, 1964).
In this study, baseflow data computed with recursive digital filters (Eckhardt, 2005; 2008) are directly supplied by the Centre d’expertise hydrique du Québec (CEHQ, 2010). Such methods are based on the generalized Equation 12. The following paragraphs present some details of the calculation procedure applied by the CEHQ.
The vertical inflows
The daily results of GR4J-CemaNeige are presented in Figure 2.7 for 1990 in Du Gouffre watershed. Additionally, the value of the parameters after calibration and the performance index of each watershed are reported in Table 2.5, showing very good NS (>0.8) for all of the watersheds.
Figure 2.7 shows that during winter, VI are very low or null, since the snowmelt is generally negligible. During spring, VI is high due to snowmelt. Therefore, a lot of runoff is produced. In summer, VI is equal to precipitation because the latter is liquid. Part of this water is lost by evapotranspiration.
The runoff (Ru)
The mean annual total runoff (Ru= QT-QB)varies between 265 mm (La Malbaie) and 375 mm (Portneuf); the mean annual total runoff is 321 mm/y over the entire CHCN region or 31 % of the total yearly precipitation. A large portion of the total runoff occurs during spring (54 %). This is attributed to snowmelt, which rapidly saturates the soil and causes surface and sub-surface runoff. The highest values of runoff are observed for Portneuf and Petit-Saguenay (35 % and 33 %, respectively), which present higher proportions of rock and till deposits than the two other watersheds (see Table 2.1).
Infiltration maps or potential recharge zones
The percentages of each component of the water budget in regards to precipitation (P) is computed and presented in Figure 2.9. The regional mean annual surface runoff (RuS) and infiltration are 181 mm and 208 mm, corresponding to 20 % and 22 % of the total precipitation. As mentioned earlier the subsurface runoff can’t be estimated by this method. The results are detailed in Table 2.6. The spatial variations (independent of the quantity of precipitation) depend on the physical characteristics of the watersheds, particularly the nature of surface deposits and slopes. Sand, gravel and low slope zones or valleys favor maximum infiltration (532 mm/y) and minimum runoff (4 mm/y), whereas the impervious deposits, including clay, till and rock, and steep landscapes correspond to the lowest infiltration rates (113 mm/y) and highest surface runoff (312 mm/y). The ranges of infiltration and surface runoff according to the nature of surface deposits are reported in Table 2.6.
Baseflow and global recharge
Figure 2.10a and 2.10b show the seasonal fluctuation in the baseflow obtained after the hydrogram separation. However, only 30 % of the baseflow occurs during summer, whereas the main part of the baseflow is observed during the spring melt period.
This directly depends on the hydrogram separation method, which considers that the baseflow increases when the streamflow rises. The same conclusion is proposed by Croteau et al. (2010) and Larocque et al. (2010). For this reason, the baseflow is also calculated separately for the June-February period, when the streamflow is lowest and is mostly from the discharge of groundwater into the river. The mean annual values for the baseflow are presented in Table 2.7. The mean yearly baseflow (over 1975-1995) accounts for 30 % of the mean total annual precipitation, whereas the mean baseflow computed over the June- February period accounts for 18 % of the mean total annual precipitation.
The accuracy of the water budget method is directly dependent on the uncertainties in each component. These uncertainties are particularly high when the recharge is low.
The larger part of the water budget is represented by the AET (48 % of the total annual precipitation). This component impacts the recharge to a large extent. The AET is defined as a proportion of the PET, which changes each year for all of the watersheds. However, this AET/PET ratio could also vary seasonally as it is indeed not always constant throughout the year. Additionally, PET and AET estimations performed in this study only rely on temperature and precipitation. Because many of other parameters impact the AET, such as the soil coverage or the air saturation, more sophisticated approaches could be implemented using an energy budget. However, this requires a lot of data that are not often available.
Runoff estimations based on the hydrogram separation (QT – QB) can also induce errors. In fact, a part of the baseflow (especially during spring) can also be attributed to subsurface runoff. In this case, the runoff calculated from QT minus QB could be underestimated.
The strong assumption that VI is the only supply of AET, runoff and recharge represents a coarse simplification. In fact, water can be added or subtracted by deep groundwater flow or by soil water storage. This fact can explain the negative value of recharge that is occasionally obtained some years (see Figure 2.11b).
Climatic data can also contain some errors. Approximatively one-third of the precipitation is solid, but this quantity is often underestimated due to the wind, which prevents snow to be captured by the snow gauge. The total precipitation measured by gauges thus represents a minimum value. Most operational agencies use snow surveys to correct SWE values throughout winter; such is the case with VICEHQ in this study. Moreover, the number of climatic stations is very low (especially in the Haute-Côte-Nord; see Figure 2.1) and is non- uniformly distributed over the area.
The hydrogram separation method seems to be the most realistic method of recharge estimation. Indeed, streamflow records represent an observable hydrologic signal that characterizes the aquifer system (Bredehoeft, 2007). Nevertheless, this principle triggers large discussions, and two main points of view are debated. Some authors support the fact that basin recharge is approximatively equal to basin discharge when dynamic equilibrium conditions are satisfied. According to Rutledge (2007), this method would be valid when the database is sufficiently long (ten years or more). Bredehoeft (2007), also encourages the scientific community to concentrate on developing methods to compute the discharge rather than the recharge. However, this analogy between discharge and recharge is too simplistic. Pumping, deep ground water flow, hydraulic connections between aquifers (e.g., Richard et al., 2014a), wetland and lake discharge, bank storage or even runoff can have significant impacts and can disrupt the equilibrium (e.g., Halford & Mayer, 2000; Scanlon et al., 2002; Stephens, 2009). In the study presented here, wetland and lake discharge could be considerable. In fact, the percentage of the watershed area covered by lakes ranges from 1 % (Du Gouffre) to 7 % (Portneuf). Additionally, the flow from springs or outcrop face seepage can also introduce water from deep fractures, as observed in steep rock valleys, such as La Malbaie.
All of these uncertainties are significant, but their quantification is beyond the scope of this paper. Thus, the daily recharge is not reliable because the water budget is not closed over one day or one season. Nevertheless, the annual mean value seems reasonable.
Finally, the methods used in this study are only based on hydro-climatic data (precipitation, temperature and streamflow). The infiltration is estimated locally to account for spatial variations in the topography, the soil coverage and deposits. Nevertheless, hydraulic conductivity, soil properties (soil saturation and water content) or the water table level are not considered. In another study, it would be interesting to take additional measurements and to investigate the saturated and unsaturated zones. Additionally, the implementation of a groundwater model, such as HELP or SWAT, would confirm the recharge values that were calculated in this study.
The comparison between ReWB1 and ReWB2 indicates the spatio-temporal variations in the sublimation of snow (see Figure 2.11b). ReWB2, which indirectly and partly accounts for sublimation through snow survey corrections, is consistently lower than ReWB1, which does not consider sublimation. The spatial analysis here is rather coarse and refers to the varying difference between ReWB2 and ReWB1 from one watershed to another.
Estimating recharge remains a major challenge in hydrology and hydrogeology. It is particularly difficult in Quebec because the soil is covered by snow six months per year. In addition, a number of regions such as CHCN only have a few weather stations, which are non-uniformly distributed over the area.
This study assessed the regional recharge and potential zones of recharge. Overall, the mean annual recharge estimated from the water budget (WB1 and WB2) and baseflow (June to February) is 183 mm on average over 1975-1995; this value represents 18 % of the mean total annual precipitation.
The hydrogram separation method is appropriate as a first approximation when the low flow period (June to February) is considered. This method is easy to apply, inexpensive and only requires streamflow data. However, this method must be validated by other methods to provide a reasonable estimate of recharge.
The water budget method is also easily implemented. Nevertheless, the method’s accuracy is hampered by underlying simplistic assumptions and by the uncertainties related to precipitation (caused by the sublimation and the underestimation of snow cover). If snow-water equivalent corrections based on a snow survey are not available, computing winter evaporation over large spatial and time scales and considering the result an estimate of sublimation are recommended.
A local approach accounts for the physical characteristics of the study area to evaluate the potential recharge zones according to the nature of deposits and the slope. The CN method has been adapted for the CHCN region to obtain surface runoff that is representative of surface deposits and the topography of the region.
Finally, applying other methods in the vadose zone would help account for the soil water content, the hydraulic conductivity and the water table in much greater detail and would help to obtain more accurate results than those presented here.
Table des matières
TABLE DES MATIÈRES
LISTE DES FIGURES
LISTE DES TABLEAUX
LISTE DES ANNEXES
1.1 PROBLEMATIQUE GENERALE
1.2 REVUE DE LITTERATURE
1.2.1 Définition de la recharge
1.2.2 Intérêt de l’étude de la recharge
1.2.3 Les différentes méthodes d’estimation de la recharge
1.3 PROBLEMATIQUE SPECIFIQUE
1.4 PRESENTATION DE LA ZONE D’ETUDE
1.4.1 Sélection des méthodes et des sites spécifiques d’étude
1.4.2 Physiographie et hydrologie
1.4.3 Contexte hydro-climatique et occupation du sol
1.4.4 Contexte géologique et hydrogéologique
2. PRÉSENTATION DE L’ARTICLE
2.3 DESCRIPTION OF THE STUDY AREA
2.4 RECHARGE ASSESSMENT METHODS
In this study four methods have been selected and applied on four watersheds into the CHCN.
These methods are applied on different time and space scales depending on the availability of climatic data (temperature, T; precipitation, P; and the streamflow, Q)
2.4.1 Water budget
2.4.2 Hydrogram separation
2.5.1 The vertical inflows
2.5.2 The actual evapotranspiration (AET)
2.5.3 The runoff (Ru)
2.5.4 Infiltration maps or potential recharge zones
2.5.5 Baseflow and global recharge
3. SYNTHÈSE ET CONCLUSION
ANNEXE A : GR4J ET CEMANEIGE
ANNEXE B : LA METHODE DES CURVE NUMBER : DE LA VERSION ORIGINALE A SON ADAPTATION
POUR LA REGION CHCN