2.1. Data
This research used sediment transport data from 33 gauging stations of the Russian Federal Service for Hydrometeorology and Environmental Monitoring (Roshydromet) located in the Terek River basin in the Northern Caucasus (cf. Table 1 and Fig. 1 for location). This dataset covers a period from 1925 to 2018 (seeTable 1 for details). It consists of average annual suspended sediment discharges (SSD , kg·s−1) and annual suspended sediment load (SSL , t·yr−1). In this study we sometimes opt to an SSD as a proxy of SSL .
We have split our database into three groups according to the gauging station altitude, i.e., low mountains (< 500 m), middle mountains (500-1000 m), and high mountains (> 1000 m). It should be noticed that bedload data is not available for rivers of the Terek River basin.
### TABLE 1 ###
### FIGURE 1 ###
According to Handbook for hydrological measurements at these gauging stations (РД 52.08.104-2002), suspended sediment data is collected using depth-integrated sediment samplers. The samples should be mainly collected once (at 8:00 AM) or twice a day (at 8:00 AM/PM), depending on the river and its flow regime. Then water samples are filtered using paper filters with a pore diameter of 2-3 µm. This is crucial, as, in sediment transport studies, membrane filters with a pore diameter of 0.45 µm are more common (Kennedy et al. , 1974). Therefore, uncertainties related to this issue are discussed in an appropriate section.
Suspended sediment discharge in the upper Terek basin is characterized by a seasonal pattern typical for an Alpine environment (Fig. 2 ). During winter (November–March), sediment sources are limited because a significant fraction of the catchments is covered by snow, and precipitation occurs in solid form. Streamflow is mainly determined by baseflow (Rets et al. , 2018; Kireeva et al. , 2020), andSSD reaches the lowest values. In spring, SSD increases when snowmelt-driven runoff mobilizes sediments along hillslopes and in channels. Simultaneously, snow cover decreases, and rainfall events over gradually increasing snow-free surfaces erode and transport sediment downstream, resulting in SSD peaks in June-July.
### FIGURE 2 ###
Several Caucasian rivers are regulated with dams and diversions constructed and operated for hydropower production (seeSupplementary 1 ). Most of them are small run-of-river hydropower plants (<10 megawatts [MW]), but there are also run-of-river medium-sized ones (10-100 MW). Such diversion projects aim to channel a portion of the river through a canal or penstock to produce electrical energy. However, the role and importance of flow diversion for small hydroelectric power plants on suspended sediment transport are currently unclear. Indeed, this type of hydropower plant in steep mountain rivers has the lowest impact on sediment flow than other hydropower plant types (Kondolf, 1997). A recent study by Csiki and Rhoads (2010) shows that run-of-river dams have almost no impact on fine suspended sediment flux if the dam is fully submerged. Moreover, the transport of fine and coarse sediments should be unimpeded during high flow periods (Angelaki and Harbor, 1995; Csiki and Rhoads, 2010). We checked literature, photos, and satellite images for every of the 18th hydropower plants. We assume that only 8 of them have altered the suspended sediment flux (see Supplementary 1 ). These are mostly run-of-river dams with in-channel reservoirs, e.g., Zaramagskaia, Kashkhatau, Gizeldonskaia, etc. The others either have a low-height dam (< 2 m), which leads to minimal upstream sediment deposition (e.g., Kokadoiskaia, Fasnalskaia, Mukholskaia, etc.), or they are located on irrigation channels (e.g., Psykhurey, Akbashskaya) or lakes (e.g., Bekanskaya).