Stream channel changes as a result of sudden sediment release due to check dam lowering (Polish Carpathians)

To show changes in channel morphology after the restoration works, the channel was studied in 2013, before dam lowering but after erection of artificial riffles, and in 2014, after dam lowering and a flood event. Both geodetic measurements and geomorphologic mapping were performed.

Geodetic measurements, which included cross-sectional analysis, were performed using a GPS Topcon Hiper II receiver set. A total of 20 cross sections were examined downstream of the dam and 6 upstream of it (Fig. 1b). Cross sections were situated on crests of artificial riffles and downstream from them. It proved impossible to establish reliable cross sections above riffle no. 2, because the section between the check dam and riffle no. 2 was quickly extracted after flood to dig out a buried water level gauge. Cross sections were also not made below riffle no. 12, because the lower section of the stream experienced changes in morphology in 2014 due to a flood on the Raba River. Raba then altered its course and forged forward into the valley of Krzczonówka stream increasing its water levels.

An attempt was made to estimate the volume of debris accumulated in reaches between riffles. Sediment volume was computed by dividing average net change in channel area measured from cross sections of adjacent riffles by the channel length between the given cross sections (Madej and Ozaki 1996). Net change in streambed cross-sectional area was defined as the difference between fill and scour in the streambed between surveys in 2013 and 2014. This calculation of debris volume must be treated as an estimate and may be used to show certain tendencies that are associated with riffle effects on deposition.

Geomorphologic surveys of the studied channel were performed using a method developed specifically for Carpathian rivers (Kamykowska et al. 1999) and tested multiple times in the Carpathians as well as in other mountain areas (Kaszowski and Krzemień 1977; Chełmicki and Krzemień 1999; Korpak 2007). The channel was divided into three reaches based on the general pattern of the watercourse and distribution of channel forms (Fig. 1b):

The reaches were described using a special table on the basis of many different quantitative and qualitative characteristics. Six of them were selected for further analysis of channel reaches morphology:

The length, width, and height of fluvial forms were measured in the field with a tape.

General changes in channel course and land cover patterns were analyzed using a topographic map from 1879 (Austrian map, 1:75,000) as well as aerial photographs taken in 1963 (1:19,000), 1977 (1:16,000), and 1993 (1:30,000), and orthophotomaps from 2006 to 2016. Prior to comparative analysis, all materials were brought to a uniform scale of 1:10,000 and a coordinate system known as PUWG1992. Boundaries of the active stream channel were then digitized from georeferenced and ortho-rectified sources using OCAD software. An active stream channel was defined as a wetted low-flow channel and gravel bars without vegetation or with patches of young vegetation (Hajdukiewicz and Wyżga 2019). The area of active channel was measured for the same three reaches determined during geomorphologic surveys. Then for each reach, the channel width was calculated by dividing the area of active channel reach by its length. Moreover, the minimum and maximum widths of channel reaches were measured perpendicular to the active channel axis. Measurement error resulting from the difficulty in determining where precisely the water’s edge is located due to streambanks concealed by trees or shadows cast by trees was estimated to be 0–5 m.

Archived materials on river regulation projects obtained from the Kraków Regional Water Management Authority, as well as project of renovation of the Krzczonówka stream obtained from Ab Ovo Association were used to identify engineering works performed on the studied stream in the past.

In 1879, the studied reach was alluvial with a wandering channel pattern despite the presence of water mills, which involved local channel deepening and the installation of structures to manage the flow of water (Fig. 2).

In the years 1935–1951, a 7 m high concrete and stone check dam at 2.25 km was built. Its main purpose was to prevent the erosion of the channel bottom, which caused the process of sliding on high, steep slopes present in most cases along the right stream bank. Once completed, the dam interrupted sediment transfer downstream resulting in sediment deposition upstream and erosion downstream with associated channel morphology (Korpak 2007).

In the 1960s and 1970s, the studied check dam reservoir still trapped large quantities of sediment, and the channel upstream from the dam was alluvial, with width exceeding 100 m at some locations (Figs. 3b, 4). The channel downstream from the dam was narrowed and deepened (Figs. 3e, 4). Erosion threatened the stability of roads situated along the banks. To protect roads, an array of longitudinal hydraulic structures such as groins, longitudinal dams, and embankments were built in the late 1950s and early 1960s.

In the 1980s, the installation of concrete drop structures upstream of the studied channel section was begun. Additional drop structures were added along the river up to the 4.6 km from the mouth (Fig. 1a). This project substantially impacted further development of the channel. Material carried by the stream became trapped between drop structures, which then starved the stream channel downstream of material.

In 1993, channel regulation with drop structures continued and was expected to include the last reach with a length of about 3.4 km. Hence, the channel upstream of the dam was still experiencing debris supply. It was much more narrow than it had been in 1977 (Fig. 4), but still tended to follow a braided pattern in some areas (Fig. 3c).

The renovation of the studied dam occurred in the late 1990s. The reservoir above the dam was already completely filled with sediment. The dam was in very poor technical condition. There were concerns that the self-destruction of the dam would cause strong erosion in the channel reach upstream from it, which in turn would threaten the stability of recently performed stream regulation using drop structures. Renovation works helped lower the principal chute spillway by 2 m. The dam was re-engineered into a cascade composed of three concrete steps (Fig. 5a).

The Krzczonówka stream channel, both downstream and upstream from the dam, was single threaded and narrow in 2006 (Fig. 4), and in many places, it was cut down to the bedrock. By this time, the period of regulation via drop structures had ended. Hence, the influx of material to the studied reach was very limited.

The width of the Krzczonówka stream channel decreased and became uniform along the entire studied section during the period 1963–2006 due to the interventions performed (Fig. 4). The channel upstream from the dam changed much more than the channel downstream from it. The greatest change occurred during the period 1977–1993 and was driven primarily by the filling of the dam reservoir with sediment and a significant decrease in the supply of material from the channel segment regulated using drop structures. Only in reach no. 1 did the main change in channel width occured earlier—in the period 1963–1977. For unknown reasons, a significant part of the area of the active channel became transformed into a terrace at the time. Drop structures in the upstream reach had not been constructed yet and there does not exist any information on any other regulation work at the time. The most likely reason was channel deepening due to the extraction of debris, but there is no proof of this.

Major changes in the width and morphology of the channel are visible on an orthophoto from 2016 (Figs. 3d, g, 4). These changes are mainly the result of restoration work.

In the years 2013–2014, a restoration project was implemented along the Krzczonówka channel segment from the check dam to the mouth. The main goals of the project were to prevent further downcutting downstream from the dam and to restore the continuity of the stream and the opportunity for fish to migrate upstream to their spawning grounds. Twelve artificial riffles about 120 m apart from each other, with height rangeing from 0.6 to 1.2 m, were created in 2013 to decrease the channel gradient, unit stream power as well as the intensity of erosion. The riffles were made of piled-up broken rocks of a size exceeding the natural grain size. A reconstruction of the check dam was performed in 2014. The highest step of the dam was lowered by 1.8 m, the middle by 0.96 m, and the lowest by 0.37 m. Additionally a trapezoidal notch was made in the central part of every step to provide for the free movement of fish (Fig. 5b).

The results of field observations will be presented in relation to the three channel reaches identified in the course of geomorphologic mapping (Fig. 1b).

In reach no. 1, the number and area of bars nearly did not change and remained small in the period 2013–2014 (Fig. 6). The number and area of cut banks did increase substantially. This reach was also characterized by the largest number of rock steps with a height of about 0.3 m.

The most readily observable change in reach no. 2 is a large increase in the number and area of cut banks (Fig. 6). The number and area of bars decreased. A large part of sediment, especially those found in the reservoir, was eroded and transported downstream. The stream channel bottom here was lowered by as much as 2 m (Fig. 7a).

Finally, in reach no. 3, the number and area of cut banks increased to some extent, but in comparison with the two other reaches, the effects of bank erosion here were the least important. The most readily observable change in this reach was a large increase in the area of bars and a relatively small increase in their number (Fig. 6). About 38% of the bars surveyed in 2014 were mid-channel bars, as seen in the increased braiding ratio The large deposition was the effect of an abrupt delivery of debris from the reservoir and a decrease in the channel gradient along this section due to the construction of artificial riffles. Almost all the artificial riffles in 2014 were covered with sediment and not visible. The thickness of the layer of deposited material reached values ranging from 0.21 to 0.91 m and, in general, was largest nearby the dam and decreased in the direction of the stream mouth (Lenar-Matyas et al. 2015, Fig. 7b, c). This trend is also visible in the diagram showing the volume of material and area of bars deposited between riffles 2 and 12 (Fig. 8). Field surveys conducted the day after the studied flood indicated the largest deposition just below the dam before riffle no. 2. However, the cross sections were measured later when this reach had already been dug out. In 2013, small bars occurred usually between riffles 6 and 12 (Fig. 8). After a flood, a large number of bars were noted close to the dam, especially between riffles 5 and 6 due to the small gradient of this river section (Fig. 8). Deposition substantially decreased further downstream—it did not occur near or atop riffle no. 10. The large deposition of material helped widen the channel in relation to its width in 2006 (Fig. 4), especially between the dam and riffle no. 10 where the mean width of the channel increased from 16.4 to 21.7 m The width of the channel below riffle No. 10 virtually did not change, which serves as proof of low deposition in this river section. It may be inferred that the debris wave propagating from the dam reached riffle no. 10 covering a distance of about 1.2 km. Channel material found further downstream came mostly from bank erosion.

The morphology and dynamics of the stream channel along the studied reaches completely changed in the years 2013–2014. Such substantial channel changes had occurred in the past due to extensive regulation works such as the construction of a dam and drop structures. The channel downstream from the dam was an erosional channel since the construction of a dam in 1951. As a result of restoration works, performed in the years 2013–2014, the channel became transformed into a depositional channel. An opposite tendency applies to the channel section between the studied dam and the footbridge—it changed from a depositional channel into an erosional one. The channel segment between the footbridge and the first drop structure changed after the erection of drop structures. Sediment trapping between drop structures has been essential to its functioning as an erosional channel. Lateral erosion plays a dominant role in the shaping of the channel upstream from the dam. The threat of the loss of streambank stability, which was the original reason for the construction of the dam in the late 1990s, became relevant again. It was just within 1 year’s time that lateral erosion on the left bank caused the widening of the channel by about 10 m along three channel sections of an approximate length of 80 m each. In one of such places, the river approached homes, which created a dangerous situation in the case of a flood. On the steep right bank, a landslide occurred due to a flood, which narrowed the channel and decreased its bedload capacity.