1. Introduction
Minimum extreme flows are sensitive to environmental changes induced by climate variability and human activities (deforestation, reforestation, urbanization, agriculture, dams,
etc.). However, the impacts of such environmental changes vary between climate regions and between watersheds within the same climate region (e.g., [
1,
2,
3,
4,
5,
6,
7]). In the current climate-warming context, in the province of Quebec (Canada) for instance, a decrease in minimum streamflow is expected to result from increased evapotranspiration and lower springtime precipitation (lower infiltration) due to climate warming [
8,
9]. Human activity could, however, prevent such a decrease. For instance, in a climate-warming context, according to [
8], deforestation would lead to a significant increase in minimum flows in summer in the Famine River watershed, a tributary of the Chaudière River. Using several general circulation models coupled with a hydrological model, [
9] analyzed the impact of increased agricultural surface area, associated with increased temperature, on the evolution of summer minimum flows, among other things, in the Chaudière River watershed. Such predictions are often marred by relatively high uncertainty concerning the response of extreme hydrological events to climate warming (e.g., [
10,
11]). However, analysis of the spatial variability of minimum extreme flow characteristics during the growing season (May to October) revealed a decrease in magnitude of these flows in agricultural watersheds. The other characteristics (timing, frequency, variability) were not affected by agriculture [
12]. Aside from deforestation and farming, the impacts of numerous dams and reservoirs in Quebec on spatial variability of extreme minimum flows have been analyzed [
13]. These impacts depend on the type of dam management mode.
However, all these studies are restricted to the analysis of seasonal or monthly extreme minimum flows. Moreover, most of the studies focus on only one characteristic, namely magnitude, even though all characteristics (magnitude, frequency, duration, timing and variability) of extreme minimum flows affect fluvial ecosystem function (e.g., [
14,
15]). The analysis of all five characteristics is therefore important to better constrain the impacts of human activities and climate on the spatial and temporal variability of AMEF (annual minimum extreme flows) characteristics.
In light of the foregoing, the three objectives of this study are the following:
To compare the spatial and temporal variability of AMEF characteristics (magnitude, duration and timing) as a function of land use and dam management mode.
To compare the relationship between climate variables (temperature and precipitation) and AMEF characteristics as a function of land use and dam management mode.
To analyze the impacts of dam management mode on the relationship between AMEF characteristics and climate variables downstream from dams.
4. Discussion and Conclusion
The impacts of land use and dam management modes on the spatial and temporal variability of AMEF characteristics (magnitude, duration and timing) as a function of climate variables were constrained as part of this study. To do so, two contiguous watersheds with similar physiographic and hydrogeological characteristics were selected, the L’Assomption River and Matawin River watersheds. The former is an agricultural watershed comprising a dam characterized by a natural-type management mode (maximum flows in the spring and minimum flows in winter), while the latter watershed is entirely forested and comprises a dam with an inversion-type management mode (maximum flows in winter and minimum flows in the spring).
Comparison of the mean values of AMEF timing revealed that these flows occur frequently in August and September in both watersheds. However, while their timing is synchronous, AMEF in the forested Matawin River watershed are higher and last longer than in the agricultural L’Assomption River watershed. Differences in precipitation cannot account for this difference between the two watersheds, because precipitation is higher in the latter (agricultural) watershed than in the former (forested). Like precipitation, annual and summer temperatures are higher in the agricultural watershed than in the forested watershed, and this difference in temperature may account for the higher magnitude and duration of AMEF in the forested watershed, as relatively higher temperature promotes evapotranspiration, leading to reduced AMEF magnitude and duration in the agricultural watershed. This factor was mentioned by [
12] for other Quebec watersheds. According to these authors, the decrease in AMEF magnitude in agricultural watersheds does not result from reduced infiltration in these watersheds since peak flood flows in agricultural watersheds are not significantly different from those observed in forested watersheds. Such a decrease in the magnitude of minimum flows with a higher proportion of agricultural land in a watershed has been observed in many watersheds in the United States [
6], among other places. In contrast, in Great Britain, an increase in agricultural area tends to lower the magnitude of minimum flows [
5].
The lack of a statistically significant correlation between temperature and flow characteristics may be explained by the fact that, unlike rainfall, temperature does not directly affect minimum flow characteristics; its influence being affected by evapotranspiration and/or runoff [
25]. The study has highlighted the predominant influence of dam management mode on the spatial variability of AMEF characteristics in regulated rivers. Thus, downstream from the Matawin River dam, characterized by an inversion-type management mode, AMEF magnitude is much lower than downstream from the Ouareau River dam, characterized by a natural-type management mode. Furthermore, downstream from the former dam, AMEF last longer and occur earlier in the year than downstream from the latter. The inversion-type management mode is characterized by water storage in reservoirs in springtime during snowmelt, and water release in winter to supply hydroelectric power plants located downstream. Water storage in the spring produces long-lasting AMEF that frequently occur early (April) in the year. Thus, downstream from this type of dam, AMEF are not affected by evapotranspiration, but rather by large-scale water storage in reservoirs in the spring, during snowmelt.
As far as the temporal variability of AMEF characteristics is concerned, the only difference observed between the non-regulated rivers is the date and nature of the shift in mean values of AMEF duration. In the forested watershed, this shift occurred later and is gradual compared to the shift in mean observed in the agricultural watershed. In the latter watershed, this shift is synchronous with the break in minimum temperature. However, it was not possible to draw a causal link between the two variables. For regulated settings, little change is observed in the temporal variability of AMEF characteristics downstream from the two dams studied. The only significant change is a shift in mean AMEF timing downstream from the Matawin River dam, a shift that is not observed downstream from the other dam, nor in natural settings.
Analysis of correlation between the climate variables and the three AMEF characteristics revealed generally low coefficients of correlation, none of these coefficients exceeding 0.600, implying a weak linear relationship between AMEF characteristics and climate variables. Be that as it may, in both the agricultural and forested watersheds, AMEF magnitude is better correlated with precipitation (particularly as rain) than with temperature. This positive correlation is explained by the occurrence of AMEF during the warm summer season, when aquifers are exclusively fed by rainwater infiltration. As for the timing of AMEF, while it is similar in the two watersheds, this characteristic is not significantly correlated with the same climate variables in the two watersheds: in the agricultural watershed, it is positively correlated with maximum temperature, while in the forested watershed, it is positively correlated with maximum summer temperature, and negatively correlated with precipitation. It follows that precipitation and temperature seem to have opposite effects on AMEF timing. As for AMEF duration, it is positively correlated with summer rain in the agricultural watershed, this correlation being absent in the forested watershed. Finally, these correlations change downstream from the dams, the largest difference being observed downstream from the inversion-type Matawin River dam.
In conclusion, human activity affects AMEF characteristics to varying degrees. Agriculture reduces the magnitude and duration of AMEF compared to the forested watershed. In the case of dams, the extent of changes in AMEF characteristics depends on the mode of management, with greater changes observed downstream from inversion-type dams.