Figure 1. A conceptual approach to RRR operation illustrating a peak flow of a river i.e., red phase that can be potentially stored by RRR and a corresponding regulated flow release back into the river i.e., bule phase. a) illustrates a hypothetical image of RRR and a river; b) presents a conceptual hydrograph of the river flow starting from upstream of the HPP and progressing downstream of RRR. Google Earth 9.185, (2020) Petäjäskoski, 66° 16’ 10” N, 25° 20’ 17” E elevation 49m. [Online] Available at: https://earth.google.com [Accessed 5 March 2023].
Hydrological and Ecological Criteria for Hydropeaking Mitigation
The hydrologic alteration in a river can be governed on spatial and temporal scales through a change in magnitude, rate of flow change, frequency, and duration of flow. Any deviation from the natural states of these parameters is associated with a discrete environmental impact (Haghighi et al., 2014). Thus, to minimize the impact, variations in these parameters should be restricted to thresholds. However, despite extensive research on hydropeaking, only a few European countries (i.e., Switzerland and Austria) have national regulations defining hydropeaking thresholds (Moreira et al., 2019). While current literature mainly focuses on qualitative targets, setting thresholds and targets for the aforementioned factors is still considered a challenge. There is still a lack of consensus to specify thresholds for the mentioned parameters (Costa et al., 2017).
The main adverse impact of down ramping is fish stranding and changes in habitat locations, which results in major ecological pressure on the river (Nagrodski et al., 2012). Besides down ramping rates, minimum flow and peak flow magnitudes affecting spawning and intra-gravel life stages are also important factors to consider (Moreira et al., 2019). Up ramping rate presents one additional factor that can cause fish drift and impact the ecological conditions. In terms of water uses, flow fluctuations caused by hydropeaking can be intense and disruptive for existing irrigation schemes (Bieri et al, 2016) and negatively impact recreational activities such as fishing, kayaking, and swimming (Charmasson et al., 2011). Finally, hydropeaking can worsen the drinking water quality by stirring up sediments and other pollutants. Therefore, in this study, we will focus on how RRRs can contribute to flow management, including ramping rates, minimum flow, and peak flow which are critical criteria for hydropeaking mitigation (Moreira et al., 2019 & Richter et al., 2010).
Model-based Design Development
A full restoration of a river regime to its natural state requires a large RRR volume which might not always be feasible due to economic or land availability constraints. However, shaving the peak flow, increasing the minimum flow, and limiting ramp rates can substantially restore the river regime to its natural state. Such objectives could be resolved with RRRs which necessitate flows to be retained and adequately released into the waterway (Fig.1). Inadequate (i.e., too slow) water release might result in small volume availability in the RRR to accommodate water from peak flows and up-ramping events. Thus, to effectively manage RRRs, a model is required to determine the timing and amount of water that needs to be stored or released. Once this is achieved, the model can calculate the required RRR volume. With this mind, the theoretical foundation of the model developed in this study was based on two main objectives. The primary objective of the RRR is to reduce the hourly peak flow (Qmax­­­­) and increase the minimum hourly flow (Qmin­­­­) induced by hydropeaking. Secondarily, the RRR aims to reduce the up- and down- ramping rates and increase the timespan during flow change occurs. The ramping flow rate (∆Q(t)) [m3s-1min-1] given in equation (Eq. 1), represents the increase (i.e., up ramping, positive values) or decrease (down ramping, negative values) in the flow over a given time step, where, Q(t) is the discharge at time t (m3.s-1), Q(t - ∆t) is the discharge at time t- ∆t (m3.s-1) and ∆t is the time step (min).
\(Q\left(t\right)=\frac{Q\left(t\right)-Q(t-t)}{t}\) (Eq. 1)
A flow pattern resembling a regulated river regime exposed to frequent hydropeaking was needed to design the model. For this purpose, Kemijoki River, one of the most regulated rivers in Finland, with a mean annual discharge of 515 m3s-1 (Ashraf et al., 2016) was selected and hourly discharge data for the lower part of the main river channel of the Taivalkoski HPP from 2015 to 2017 was obtained from national datasets maintained by the Finnish Environment Institute (Hertta-database, for more details, see Ashraf et al., 2016). To generate the required flow pattern (hereinafter called scaled flow), characterized by an average discharge of 1 m3.s-1, the Taivalkoski HPP discharge data was scaled down by dividing the hourly discharge by the average hourly discharge per day (Eq. 2).
\(Q_{\text{scaled}}\ (t)=\frac{Q\left(t\right)}{Q_{\text{avg}}(d)}\)(Eq. 2)
Where Qscaled (t) is the scaled discharge at time t (m3.s-1), Q (t) is the actual discharge at time t (m3.s-1), and Qavg (d) is the average hourly discharge per day (m3.s-1).
Once the scaled flow is attained, it is possible to establish a hierarchy of operational objectives with a range of distinct thresholds that dictate the timing and amount of water that needs to be stored or released by the RRR. Thus, a re-regulation algorithm that operates the RRR based on the following list of hierarchal objectives and their associated thresholds was developed;
As the ideal flow conditions for the various ecosystem services may be different, a range of thresholds was utilized in the algorithm to determine the required RRR volume for several hydropeaking mitigation scenarios. Thus, the threshold range for flow magnitude was selected to include all the possible mitigation scenarios, by using 10% flow adjustment increments. Whereas, the ramping rate thresholds were carefully chosen to include a range of scenarios, by incrementally adjusting the lower and upper limits of the range. The threshold range started from a threshold below the average ramping rate and was extended to reach up to 50% of the maximum ramping rate, using increments of 0.5 m3sec-1min-1. The unscaled average up- and down- ramping rates downstream of Taivalkoski HPP during 2015 to 2017 were 1.5085 and -1.36 m3sec-1min-1,respectively. As such, the lower limit for ramping rate threshold was set to 1 m3sec-1min-1. Whereas the maximum unscaled ramping rate reached up to 8 m3sec-1min-1, as such the upper limit for the ramping rate threshold range was set to 4 m3sec-1min-1. To further expand the scope of possible mitigation scenarios, thirty-five permutations were created and tested by matching the peak and minimum flow thresholds (i.e., priority 2) with ramp rate thresholds (i.e., priority 3). Hereinafter, the permutations will be referred to as P (X%, Y), with X% being the percentage adjusted from Qmax and Qmin i.e., (100-X) %×Qmax and (100+X) %×Qmin, while Y is the ramp rate threshold i.e., (Y) for up-ramping and (-Y) for down-ramping. One example from the permutations is P (10%, 2.5) which matches the 10% flow adjustment (i.e., 90%*Qmax and 110%*Qmin threshold) with a ramp rate threshold of 2.5 m3s-1min-1. Additionally, we demonstrate how the RRR would re-regulate the flow downstream of Taivalkoski (Kemijoki) for permutation P (40%, 2) by using the re-regulation algorithm.
This model was used to determine the required volume of RRRs downstream of HPPs operating at the Kemijoki River. It has the potential to be utilized in other rivers with similar flow patterns to achieve the above listed priorities. However, the range of thresholds employed by the re-regulation algorithm must be modified to best suit the flow pattern of the investigated river. It is important to note the model assumes the location of the RRR is immediately downstream of the HPP, thus not accounting for flow velocity or the time required for water to reach the RRR. The model also assumes ideal RRR conditions with no consideration of any water losses that might occur due to evaporation and seepage.
Results
Table 1. presents our calculations showing clear theoretical possibilities for regulating hydropeaking with RRRs. For example, assuming that the future flow is consistent with the scaled flow of the study period, the required RRR volumes (Table 1) are sufficient to ensure the hydropeaking thresholds are respected throughout the year. The results indicate that, for most of the tested permutations, the required volume of the RRRs increased as the thresholds for peak and minimum hourly flows and the ramping rates became more stringent. Nonetheless, for some permutations, this trend was not observed. One example is the required RRR volume for permutations P (10%, 3.5, volume: 0.256 million cubic meters (MCM)) and P (10%, 4, volume: 0.368 MCM) are larger than the reservoir volume needed for permutations P (10%, 2.5, volume: 0.147) and P (10%, 3, volume: 0.143) which have more stringent ramp rate threshold. Furthermore, for permutation P (10%, 4), the required RRR volume (0.368 MCM) decreases slightly compared to P (20%, 4, volume: 0.262 MCM), which has more stringent peak and minimum flow thresholds. Our theoretical approach demonstrates the relationship between the required RRR volume, daily peak discharge, and ramp rate thresholds (Figure 2).
Furthermore, the required RRR volume was determined for each month separately, illustrated in Figure 3. The results indicate that for most of the investigated permutations except permutations with 50% flow adjustment, July and August require the largest reservoir volume to achieve the objectives and priorities stated in section 2.3. Whereas, for permutations with a 50% flow adjustment, January is the month that requires the largest reservoir volume. However, it is important to note that January and February are the months with largest volume requirement when the ramp rate thresholds exceed 2.5 m3.s-1.min-1 for permutations with a 10% flow adjustment. Additionally, the RRR operation for P (40%, 2) for the scaled flow downstream of Taivalkoski (Kemijoki) is demonstrated in Figure 4. For all the permutations, the RRR limits peak and minimum hourly flows and ramp rates according to thresholds defined by the algorithm. Nevertheless, Figure 4.a. demonstrates that the RRR increases the minimum flow beyond the defined threshold in numerous time steps without violating other priorities. However, as illustrated in Figure 4.b., there are time steps where priority 2 takes precedence over priority 3, causing the ramp rates to surpass the defined threshold.