EARTH SURFACE PROCESSES AND LANDFORMS, VOL.

18, 541-556 (1993)

RIVER RESPONSE TO CHANNEL REGULATION: CASE STUDY

OF THE RABA RIVER, CARPATHIANS, POLAND

BARTŁOMIEJ WYŻGA
Polish Academy of Sciences, Nature and Natural Resources Protection Research Center, ul. Lubicz 46, 31-512 Kraków,
Poland

Received 14 September 1991

Revised 15 March 1992

ABSTRACT
Processes induced by the channelization of the Raba River in the present century are examined to illustrate the response
of a gravel-bed stream following narrowing and straightening of its channel. Up to 3 m of incision has occurred. The
change from a slow and relatively steady degradation in the lower reaches to separate degradation events in the higher
reaches is attributed to the differential rate of headcut retreat and to the control exerted by mid-channel bars upon the
rate of river-energy dissipation. Progressive outwashing of finer grains from bed material has followed the diminishing
sediment yield of the basin and the increase in stream power. The ensuing growth in mean grain size and changes in
sediment fabric have increased boundary resistance to flow and reduced particle susceptibility to entrainment.
Downstream magnification of peak discharges has become increasingly pronounced with the advancing incision. The
decrease in flood-plain storage and self-acceleration of flows passing the relatively deep and straight channel has caused
flood waves to become progressively more flashy in nature.
An increase in channel depth and reduction in gradient caused by downward and backward erosion, as well as bed
material coarsening has promoted the re-establishment of an equilibrium. Conversely, flow velocity increases due to flow
concentration in the deepened channel. Reduction in grain mobility allows the river to attain a new equilibrium at flow-
velocity and stream-power levels higher than before the channelization.
Numerous disadvantages of the applied regulation scheme and its failure to reduce flood hazard raise the question of
its maintenance. To be successful, any regulation design must take into account changes in sediment supply and flood
hydrographs resulting from the simultaneous alterations in basin management.

KEY WORDS Channelization River response Channel incision Bed-material changes Flood-flow modifications Flood hazard

INTRODUCTION
Regulation of river channels causes a direct modification of fluvial processes operating within and
downstream of the channelized reaches. It induces adjustment in channel and flood-plain morphology as well
as flow characteristics (Gregory, 1987; Simon, 1989). However, where the channel width and sinuosity are
fixed by channel-control structures, the processes of river adjustment within the channelized reach do not
follow strictly the rules outlined by Schumm (1968).
The valleys of Carpathian tributaries to the Vistula River were repeatedly subjected to severe flooding
during the 19th century (Mikulski, 1954). The great intensity of meteorological and hydrologic phenomena
during the final part of the Little Ice Age (Kotarba, 1989), as well as an acceleration of water run-off due to
the deforestation of mountain basins and introduction of potato cultivation (Klimek, 1987), gave rise to flash
floods during which Carpathian rivers transported large amounts of bedload and deposited them in the
lower, foreland reaches (Wyżga, in press). This resulted in channel aggradation (Wyżga, in press) and fast
lateral migration of shallow, wide channels (Klimek, 1974).

0197-9337/93/060541 -16$ 13.00

© 1993 by John Wiley & Sons, Ltd.
542 B. WYŻGA

River-control works were started in the middle and lower courses of the Carpathian tributaries to the
Vistula at the beginning of the present century (Kędzior, 1928) to minimize flood hazard and stabilize
channel position within the valley floors. They consisted of channel straightening through meander cut-offs,
channel narrowing by groins, and the lining of concave banks by gabions and rip-rap. At the same time
construction of check-dams in the headwaters was initiated (Kędzior, 1928).
Some alterations also occurred in the management of Carpathian river basins during the 20th century.
Contour ploughing and the terracing of slopes was increased and the grazing of forests decreased. In the
eastern part of the Polish Carpathians much arable land was turned into pasture and meadows. Channel
downcutting has occurred with increase in stream power of the Carpathian tributaries to the Vistula which
has been caused by river-control works, combined with a reduction in bed-material load due to the changes
in basin management. The Carpathian tributaries to the Vistula have incised by 1·5−3·3 m until the present,
as shown by the lowering of their minimum annual stages in water-gauge channel cross-sections (Wyżga,
1991, Figure 1). This downcutting has resulted in a number of detrimental effects (Klimek, 1987; Wyżga,
1991).
This paper is a case study of the changes in the Raba, a mountain gravel-bed river, which have followed its
channel regulation during the present century, especially in the last 40 years. The paper has two main aims:
first, to show the complex nature of the mechanisms involved in approaching a new equilibrium; and second,
to discuss disadvantages of the applied regulation scheme being revealed on the catchment and regional
scales, and to utilize them as a geomorphological lesson for river engineering.

FIELD SETTING AND STUDY METHODS

The Raba River drains the northern slopes of the Western Carpathians (Figure 1A). This river is
characterized by great variability of water stage and discharge (Punzet, 1969). Such a hydrological regime
results from the low retention potential of the flysch bedrock, high relief (extreme elevations 1310 and 180 m
a.s.l.), the low forest cover of the basin, and the predominant northwestern aspect of the slopes, favouring a
high precipitation from oceanic air masses arriving from this direction. Floods usually occur during the
summer, and snow melt results in prolonged freshets in March and April.
Most of the data on the hydrologic, geomorphic and depositional history of the river are available for its
middle course (within the Carpathian Foothills), and so the reach between Dobczyce and Wieniec (Figures
1A, B) is analysed in detail where the average river slope is 0·00197. The river impoundment at Dobczyce
(Figure 1A) took place in 1987, and so the changes discussed in the paper were not affected by the dam.

Figure 1. (A) Location of the study area. The Raba River reach analysed in (B) is shown by the thick line. (B) Changes in the course of
the Raba River channel between Dobczyce and Wieniec in the period 1952-1974 as shown on the maps at 1:50000 scale. This reach was
channelized in the 1950s to 1970s. The range of the sampling of channel sediments for textural analysis is indicated by the dotted line,
and the location of the reach shown on Figure 2 by the window
 RIVER CHANNEL REGULATION 543

Table I. Decade-average values of the highest and mean annual discharge, and of the duration of flows higher than the mean
as recorded at the Gdów and Proszówki gauging stations

In the period 1951-1980 in the middle of the reach at the Gdów water-gauge station (Figure 1), mean
annual discharge and mean annual flood amounted to 12·14 and 318 m3s-1, respectively. During this period,
however, changes occurred in both parameters due to climatic fluctuations, as revealed by their decade-
average values for the Gdów station (Table I). Following the 30 per cent rise in mean annual discharge, the
mean annual flood was reduced by more than half, and some growth in the duration of flows higher than the
mean took place. Relatively infrequent and high flood waves gave way to flattened and more prolonged
waves or to frequent but low ones.
A comparison of maps and aerial photos has allowed evaluation of the changes in channel pattern, width
and sinuosity. The at-a-station changes in depth and shape of the channel have been reconstructed on the
basis of multi-year variations of the lowest annual water stages as well as repeated measurements of water-
gauge channel cross-sections made by the Hydrologic Survey. These have been supplemented by measure-
ments of low-water table elevation made by the Water Authority and subsequently followed by the author's
surveys to show changes in vertical channel location along the river. Facies patterns of sediments of known
age have been studied in river cutbanks and walls of gravel pits, followed by an analysis of the fabric and
grain-size distribution of gravel-bar sediments to make inferences on depositional conditions during the
passage of flood waves. Samples have been collected to represent the observed differentiation of sediments
from a given period, and sampling has been confined to a short valley reach (Figure IB) to avoid distortion
of temporal sediment-change patterns by downstream trends. Only subpavement material has been sampled
in the present-day channel. Finally, conclusions about the channelization-induced modifications in flood
flows have been drawn from at-a-station stage-discharge relationships and downstream changes in
discharge-frequency relations.

REGULATION OF THE RABA RIVER AND CHANGES IN PLANFORM OF THE

CHANNEL
The river-control works carried out during the first 30 years of the 20th century caused a 15 per cent
shortening of the Raba River channel in the lower course, but the shortening was insignificant in the middle
course. The channel was narrowed by training structures in the middle course, and during the early half of the
century the braided channel was transformed into a single channel. This single-thread channel showed
a conspicuous tendency to meander (Figures IB, 2A). The cessation of river-control works during and
immediately after the Second World War allowed the river to increase its sinuosity; this increase amounted to
16 per cent in the Dobczyce-Gdów reach.
The channelization works were resumed late in the 1950s. The resulting changes in the river planform may
be detected using aerial photos taken in 1955, 1974, and 1987 at similar low-flow conditions (Figure 2). For
each river course thalweg length, river sinuosity and mean width of low-flow channel and flood channel
(equal to the zone of active sediment transport) have been determined (Table II), the latter being based on
measurements at 100 m intervals.
544 B. WYŻGA

Figure 2. Channel changes in the middle course of the Raba River between 1955 and 1987 as exemplified by aerial photos of the reach
near Winiary taken in (A) 1955, (B) 1974 and (C) 1987
 RIVER CHANNEL REGULATION 545

Table II. Changes in length, sinuosity and width of the Raba River channel between Dobczyce and Wieniec in the period
1955-1987. (LFC, low flow channel; FC, flood channel)

The pre-regulation channel from 1955 was sinuous and wide (Figures 2A, 5B; Table II). The zone of active
bed-material transport was three times wider then the low-flow channel, and favourable conditions existed
for efficient sediment storage in the channel, as shown by the great number of broad point bars as well as a
few mid-channel bars (cf. Church and Jones, 1982). There was some correlation between both width values
(r = 0·27, significant at the 99·99 per cent level). This is attributed to the relatively free channel development
at high flows, and indicates that under conditions of relatively shallow flows the pattern of constrictions
and expansions of the flood channel played a predominant role in determining width variations at lower
discharges.
River shortening was an important part of the repeated regulation scheme. By the mid-1970s the river-
shortening works were nearly finished, resulting in a 14 per cent reduction of river sinuosity in the
Dobczyce-Wieniec reach (Figure 2B, Table II) and by approximately 10 per cent in the 20 km long
downstream reach. However, the decrease in width continued beyond this time (Figures 2B, C; Table II). Up
to 1987 the low-flow channel width diminished by about 40 per cent, but the decrease in width of the bed-
material transport zone was much greater and amounted to about 60 per cent. There was, however, no
significant correlation between the width values for the 1974 and 1987 channels, this being at least partly
related to the growing three-dimensionality of the flows. Progressive downcutting caused the low-flow
channel to be narrowest at bends where, in fact, the channel is widest at high flows, and this has tended to
obscure the previous relationship.
Summarizing, as a result of channelization works carried out in the 1960s and 1970s, the Raba River
channel in the middle course is now relatively straight and narrow (Figures 1B, 2C, 5B). There are only a few
mid-channel bars in the channel which serve as efficient sediment-storage features, and these are connected
by long transport reaches where side bars are narrow and non-active depositionally (Figures 1B, 2C, 5).

RESPONSE OF THE RIVER TO CHANNEL REGULATION

Channel incision
Streambed degradation is perhaps the most apparent response of a river to channel regulation. At Gdów
degradation started in 1908, four years after the initiation of channelization works in the middle course of the
Raba, and has resulted in approximately 3-0 m of channel incision (Figure 3). Between 1908 and 1958 four
degradation-aggradation cycles occurred, resulting in an overall downcutting of the channel by 1 m
(20 mm year-1 on average). The channel bed was, however, relatively stable in the final part of the period
when the marked increase in river sinuosity took place. During the years 1958-1964 the river bed was rapidly
degraded by 0·6 m (100 mm year-1), followed by the vertical stability of the channel until 1971. Channel
incision amounted to 0·9 m over the period 1971-1976 (176 mm year-1), and the last years revealed a
somewhat lower rate (45 mm year-1).
At a newly installed gauging station at Dobczyce, channel downcutting of about 1·25 m was recorded
during the period 1971-1984 (average rate 96 mm year -1), and there seems to have been no significant
 546 B. WYŻGA

Figure 3. (A) Changes in channel cross-section of the Raba River at the Gdów water-gauge station between 1928 and 1987. The
elevation in metres a.s.l. of mean annual stage is also marked. (B) Changes in the lowest annual stage of the Raba River since the end of
the 19th century at the Gdów station

incision prior to this time. Most of the incision was achieved in two events of rapid river-bed degradation
(by 0-4 and 0-9 m, respectively), separated by a period of relative channel stability.
Downstream of the reach under consideration, 2-3 m of river incision occurred during the present century,
with the rates of downcutting declining rapidly since the mid-1970s.
Temporal changes in mean flow velocity at the Gdów station have been used to determine the possible
influence of the channelization works on the vertical river-bed instability. Velocity values are shown for the
range of flows that occurred at the gauge cross-section before (1952) and after (1987) the channelization
(Figure 4), indicating a considerable increase in mean velocity of a given flow following the repeated channel
regulation of the 1960s and 1970s.
Two phenomena are important when analysing the timing of the incision. The first is a marked delay in the
onset of incision when moving upstream. The second involves changes in the rates and time distribution of
downcutting along the river. The slow and relatively steady degradation of the lower reaches turned in the
higher reaches to a set of separate degradation events whose rates and magnitudes have been growing in the
upstream direction (Wyżga, 1991). Time-shifting of the main phase of channelization works, especially of
river-shortening works, for different river reaches seems to be only a partial explanation for the first
phenomenon. It is evidenced by the amount of incision that occurred during the years 1976-1987 in the reach
under consideration (Figure 5A) where the works had been carried out more or less synchronously in the
1960s and 1970s. At the downstream end of the reach nearly 3 m of river downcutting had taken place by
1976. With the total amount of incision being similar (3 m) up to 8 km of the reach, the period in question
 RIVER CHANNEL REGULATION 547

Figure 4. Plot of mean flow velocity (v) versus discharge (Q) for the range of flows that occurred at the Gdów water-gauge cross-section
before (1952) and after (1987) the channelization

yielded approximately 0·5 m of streambed degradation at Gdow, and nearly 2·5 m at 8—8·5 km. Until now,
1-1·5 m of incision has occurred upstream of 8 km, most of it having been achieved in the youngest period.
As a result of streambed degradation following channelization of a given reach, the river slope is
moderated but steepened immediately upstream. A headcut formed in this way retreats to dissipate excess
power of the river by an upstream-migrating 'wave of channel-bed degradation' (cf. Daniels, 1960). As the
river was channelized over a long distance, the incision-triggering result of works in a particular subreach
adds to the degradational effect of the downstream works. The consequence is a series of backward-moving
steps when looking at a reach (Figure 5A), or of degradation events when analysing the history of a station
(Figure 3B).
It seems evident that the growth in rapidity of degradation events and in headcut height observed in
the upstream direction (Figure 5A) must reflect a differential rate of backward headcut movement,
the downstream headcuts being able to retreat faster and thus to overtake and join those located upstream.
The reason for this differentiation is, however, unclear. It is possible that with the grain size of bed material
increasing upstream, the river is able to attain the power needed to exceed the threshold of channel stability
[Bull's (1979) threshold of critical power] over progressively longer time intervals. However, channel
armouring during the passage of a 'degradation wave' leaves behind bed material as coarse as that in the
hitherto undissected reach upstream. Instead, the explanation may depend upon the mechanism of positive
feedback occasioned by an initial downcutting, which will be discussed later. High flows are concentrated
within the deepened part of the channel, thus yielding high velocity and stream-power values in the channel
itself. On the contrary, a large proportion of flood flows is forced to escape the channel in its undissected part,
thus lowering velocity and stream power.
Observations testify to the important role played by mid-channel bars in the temporal arrest of upstream
movement of a 'channel-degradation wave'. Reaches with mid-channel bars are typified by relatively steep
 548 B. WYŻGA

Figure 5. (A) Changes in low-water table elevation between 1976 and 1987 in the Dobczyce-Wieniec reach. Location of mid-channel
bars in the channel is indicated by the letter B, and the extent of repeated regulation works that destroyed the pool-riffle pattern of
the river by arrows. (B) Changes in width of the zone of active sediment transport in the Dobczyce-Wieniec reach between 1955
and 1987

gradients (cf. Hooke, 1986), particularly in their distal parts (Figure 5A; note especially the bar at 7·75 km of
the 1987 channel). Monitoring shows the vertical stability of the channel under these conditions. However,
the attachment of a mid-channel bar to one of the banks or restitution of the single-thread flow conditions by
man, accelerates sediment and water transfer in the channel and allows further headcut recession. Clearly,
a significant portion of the shear stresses generated by flows must be spent by the river to overcome high
channel-form resistance (Bathurst, 1982) in reaches with mid-channel bars. As bed material is being deposited
 RIVER CHANNEL REGULATION 549

on the bars, channel sediments may be stored in the reaches, and therefore are not available for transport
downstream. Indeed, the minimally degraded upper part of the reach under consideration is the one where
comparatively high potential for sediment storage by the river has been preserved (Figure 5B).
Bed-material changes
The channel bars of the 19th century Raba River were formed by structureless, very poorly sorted
(Figure 6), bimodal gravels, distinctly coarser than the older sediments. The gravels were recognized as the
deposits of a shallow, aggrading low-sinuosity river (Wyżga, in press). With a relatively narrow channel and
concentrated flow, normally loose gravels (with a gravel mode dominating over sand, like the present-day
sample D, see Figure 7D) were deposited by this river. Such gravels occupy the lower part of the field of 19th
century sediments on the plot of mean size versus standard deviation (Figure 6). Dilated framework gravels
(Figure 8A) were formed in wider channel reaches, the flow being diverged among mid-channel bars. These
deposits consist of nearly equal proportions of gravel and sand (Figure 7A), and are grouped in the upper
part of the 19th century field on the Mz-σ1 plot (Figure 6). Such gravels must have originated by the
simultaneous deposition of gravel and sand particles (Plumley, 1948). Characteristically, pavement horizons
have not been found within the 19th century Raba deposits.
The depositional form of the 19th century Raba River was interpreted (Wyżga, in press) as resulting from
the passage of flood waves with great peak discharges and short time-bases. Waves of these flash floods
originated in the mountainous part of the basin and carried huge volumes of bed-material load. After the
peak of such a flood wave, the sediment was deposited rapidly, and subjected to little or no reworking. The
marked homogeneity of these sediments resulted from the minimal difference in bed shear stresses generated
within the wide, shallow channel.
From the beginning of the present century massive, normally loose gravels predominate, and the dilated
framework and openwork gravels are of secondary importance within the bar deposits of the braided Raba
channel. All the sediments are very poorly sorted but vary greatly in mean grain size (Figure 6).

Figure 6. Scatter diagram of mean size (Mz) versus standard deviation (σ1) for the bar sediments of a particular age: ●,19th century;
□, beginning of 20th century; *, 1950s; ▼, present-day low-sinuosity reaches; ▲, present-day high-sinuosity reaches. Letters denote
sediments shown in Figures 7 and 8
550 B. WYŻGA

Figure 7. Histograms of the grain-size distribution (in weight frequency per cent) of the selected samples of the Raba River bar gravels.
Sample dating is indicated on Figures 6 and 8. Hatched areas represent percentage of undivided fines

Figure 8. Photographs showing changes in texture and structure of the Raba River bar gravels between the second half of the 19th
century and the present day. (A) Massive, dilated framework gravels (second half of 19th century). Scale is given by the box of matches;
(B) filled underloose gravels, visible gravel imbrication (1950s); (C) normally graded, openwork gravels, clasts dip downflow (1950s);
(E) normally graded, planar cross-stratified, openwork gravels originating on the foresets of the present-day transverse bar in a high-
sinuosity reach. 25 cm scale on levelling rod

The point-bar deposits of the sinuous river from the 1950s have a highly variable texture. They cover a
range of sediment-fabric types from dilated framework gravels to normally loose and filled underloose
(Figure 8B) to openwork gravels (Figure 8C). The improvement in sorting and increase in mean grain size
observed when shifting from dilated framework to openwork gravels on the Mz-σ1 plot (Figure 6), reflect a
decreasing percentage of sand in the sediments (Figure 7B, C).
 RIVER CHANNEL REGULATION 551

The clustering of the present-day deposits in two distinct fields on the Mz-σ1 plot (Figure 6) reflects the
differentiation of the river into low- and high-sinuosity reaches. The former are typified by normally loose
and filled underloose gravels, showing a significant sand content (Figure 7D) and very poor sorting
(Figure 6).
The present-day point-bar deposits are openwork gravels (Figure 8E). They contain little or no sand
(Figure 7E) and show poor to fair sorting (Figure 6). The gravels are planar cross-stratified due to deposition
on the slip-face of a transverse bar, and exhibit perpendicular normal grading when viewed in section
(Figure 8E).
Pavement horizons and imbricated gravels occur occasionally in deposits from the beginning of the
present century, but are more frequent in the point-bar deposits from the 1950s. Now the channel bed is
paved along its whole width and shows numerous pebble clusters, the mean size of pavement typically
exceeding that of subpavement material, shown on Figure 6, by 1-2 Φ units.
The two possible reasons for the 20th century sedimentation changes are difficult to separate from each
other. First of all, a reduction in fine-sediment load of the river must have occurred during the period (Wyżga,
in press), brought about by the alterations in basin management as well as by the great fall in peak discharges.
The reduction is indicated by the steepening angle of openwork-gravel foresets [known to reflect the lowered
percentage of sand in the supplied sediment (Carling and Glaister, 1987)], by the increasing contribution of
openwork gravels in the bar structure, pointing directly to a smaller amount of sand carried in the bedload,
and by the progressive enhancement of pavement formation [conditioned by considerable flow turbulence,
its intensity being inversely related to the suspended-sediment concentration of a river (Beckinsale, 1972)].
The increase in flow velocity and stream power owing to river channelization has resulted in progressive
outwashing of finer grains from bed material, the prolongation of flood flows adding to the efficiency of the
process. The importance of the channelization works is shown by the marked reorganization of the bar-
deposits facies pattern after the 1950s and especially by the great difference between the sediments originating
in low- and high-sinuosity reaches of the present-day river (Figures 6-8). Again, very efficient sorting
mechanisms operating on the point bars, as opposed to those active in low-sinuosity reaches (Wyżga, in
press), show the great changes in pattern of bed shear stresses between the reaches, caused by various rates of
stream energy dissipation due to the resistance of channel form to flow.
The 20th century changes of channel sediments were therefore an important mechanism of river self-
adjustment under fixed-sinuosity conditions. The river tended to re-establish equilibrium conditions between
its increased sediment-transporting ability and the abating sediment supply through the reduction in particle
susceptibility to entrainment. Selective outwashing of fine grains from bed material brought about a
progressive increase in the mean size of channel sediments (Figure 6). Overloose gravels, the packing of which
allows the bulk of particles to be easily set in motion from the flow boundary (Church, 1978), ceased to
originate in the channel. Instead, deposition of underloose gravels grew in significance, accompanied by the
formation of pebble clusters and channel pavement. Tight packing of bed-material particles and their
aggregation in cluster bedforms considerably reduces grain mobility (Laronne and Carson, 1976; Reid and
Frostick, 1984) and causes the channel bed to react as if consisting of its coarsest elements (Reid and Frostick,
1984).
Flood-flow modifications
The stream-channel and flood-plain system developed for a dominant flow is known to have a persistent
effect on flood waves moving down a river (Burkham, 1976). A low-flow system, developed by and for low
flood flows and characterized by a low-gradient, sinuous, narrow, deep channel, attenuates flood waves
passing through the reach. On the other hand, the shape of flood waves passing the steep, relatively straight,
wide and shallow channel of a high-flow system is not altered, and their peak discharges are sustained.
Little attention has been given to the influence exerted upon flood flows by an 'unnatural', channelization-
formed system being characterized by a steep, straight and deep channel. Under conditions of an over-
steepened gradient, the bed scouring was not compensated by deposition on the banks (Figure 3A). As a
result, increasing discharges could have been accommodated by the channel as incision progressed (Table
III). Flood flows, which previously would have been divided into in-bank and overbank flow, are now
552 B. WYŻGA

Table III. Changes in discharge and recurrence interval of flood flows of a given stage which followed the channel-bed
degradation at the Gdów and Proszówki gauging stations

transferred entirely within the channel. This incision-induced within-channel flow concentration must have
caused an increase in flow velocity and stream power, thus promoting further streambed degradation.
To examine the way the within-channel flow concentration affects flood waves in their downstream
movement, temporal changes in the ratio of annual peak discharge in the downstream gauging station
(Proszówki, Figure 1A) to that recorded in the upstream station (Gdów) have been analysed (Figure 9). First,
a slow but significant fall in the ratio occurred from 1·67 in the 1920s to 1·06 during the 1940s, followed by
a fast increase up to 2·41 in the 1970s. The pattern of changes is more apparent when the value of 1·45 is
subtracted from each annual ratio value; this value represents a multi-year average ratio of mean annual

Figure 9. Flood-flow changes in the middle and lower Raba River channel between 1921 and 1981; ◊, cumulative values of the ratio of
maximum annual discharge at the downstream gauging station (Proszówki) to that recorded at the upstream station (Gdów) (decade
average values of the ratio are shown at top of the figure); ▼, cumulative values of the 'corrected' ratio of the maximum annual
discharges; ○, changes in vertical channel location in the Gdów-Proszówki reach since 1920. The data are weighted means, alterations
in the extreme stations (Gdów and Proszówki) accounting for one-quarter of the total change each, and those in the middle of the reach
(Książnice station) for one-half
 RIVER CHANNEL REGULATION 553

discharge at both stations, which may approximate to the increase in volume of the flood waves due to an
addition in the lower part of the basin, and thus describe the contribution of the volume increase to the
downstream peak-discharge changes. The value is perhaps somewhat conservative as the maximum unit run-
off is greater in the mountainous part of the basin than in the foothills (Punzet, 1969); however, this effect may
be at least partly counterbalanced by the years when the highest discharge is due to a snow-melt freshet,
melting being faster in the deforested, lower part of the basin. Therefore, negative or positive values of the
'corrected' ratio represent the lower and higher downstream growth in peak discharges than could be
accounted for by a downstream increase in flood wave volume, with falls and rises of the cumulative
'corrected'-ratio curve reflecting changing trends of the phenomenon.
When the trends in discharge change are compared with the curve showing alteration in vertical Raba
channel location in the Gdów-Proszówki reach since 1920 (Figure 9), a striking coincidence is evident
between the channel downcutting, which commenced at the end of the 1950s, and the peak-discharge
changes. This shows the importance of channel incision in flood-flow modification under conditions of a
fixed river sinuosity. On the other hand, the trend towards attenuation of peak discharges in the period of
relatively stable vertical channel position, especially pronounced during the 1940s, may reasonably be related
to the growing river sinuosity of that time.
Since the growth in peak discharges with downstream flood-wave movement was much greater than that
accounted for by the addition of water with increasing basin area, a change in shape of flood waves must have
occurred, the increasingly peaked nature of flood waves being compensated for by their shortened time-bases.
It is suggested that this is due to the combined effect of a decrease in flood-plain water storage due
to channelization (cf. Brookes, 1987) and self-acceleration of flows concentrated in the deepened channel. The
former factor seems to be of limited significance; any downstream growth in discharge due to within-channel
flow concentration in a given reach should increase flood-plain inundation, thus resulting in flow at-
tenuation. In the Raba channel a decrease in potential within-channel flow concentration may even be seen
with the lowering channel incision in the downstream direction (Table III). In a deep and straight, 'unnatural'
channel the increase in relative smoothness (ratio of water depth to the height of protrusion of bed-material
particles to the flow) with growing discharge, proceeds faster than in a wide, shallow channel of a high-flow
system, but the ensuing relative decrease in resistance to flow is not compensated for by flow retardation on a
flood-plain typical of the channel of a natural low-flow system. It results in the faster movement of the later,
more voluminous parts of a flood wave that override its earlier parts (cf. Leopold and Miller, 1956) to cause
downstream magnification of peak discharges.

A MODEL OF RIVER RESPONSE TO CHANNEL REGULATION

The versatile and relatively well-documented analysis of channelization-induced fluvial processes presented
above, allows the formulation of a conceptual model of river response to channel regulation (Figure 10).
The steepening of the channel slope due to river shortening, and the growth in mean depth or hydraulic
radius following narrowing, increase mean flow velocity, mean bed shear stress and stream power. To
compensate for these increases, flows entrain and transport greater amounts of bed material, which causes
the channel incision. Channel enlargement due to an increase in depth as downcutting progresses, tends to
diminish the above parameters, and thus to re-establish equilibrium conditions.
Backward erosion is induced with slope steepening immediately above a degraded reach. It operates
through headcut (or knickpoint) retreat, and can result in propagation of channel degradation, both in a stem
river and in its tributaries, high above a regulated reach (Daniels, 1960; Emerson, 1971; Simon, 1989). Both
the increase in depth and reduction in slope of a stream channel, occasioned by backward erosion, contribute
to the formation of an equilibrium.
Apart from a reduction in mean velocity for a given in-bank flow with growing channel cross-sectional
area, the result of channel enlargement, due to depth increase, is also a concentration of progressively higher
flood flows in the narrow channel zone. The advancing within-channel flow concentration brings about an
increase within the channel of flow velocities and stream power associated with a given discharge. A positive-
feedback loop originates, resulting in continued bed scouring after the initial downcutting (headcut or
knickpoint passage, cf. Daniels, 1960). Furthermore, a fast rise in peak discharges takes place due to the
 554 B. WYŻGA

Figure 10. Conceptual model of changes in flow velocity following channel regulation

decrease in flood-plain water storage as well as flow self-acceleration as flood waves move through a
deepened channel section. Channel enlargement occurs downstream of channelization works to allow
accommodation of the increased flows (Brookes, 1987). When increased flows are conveyed by a regulated
channel, bed scouring offers the river the only possible way to enlarge its channel. Thus, another positive-
feedback mechanism is brought into operation by within-channel flow concentration—the downstream bed
degradation lowers a local base level, and this can, after some time, result in channel incision upstream
through backward erosion.
The progressive outwashing of finer grains from the bed material follows channelization. As a result, the
mean grain size of channel sediments increases, and a pavement forms on the streambed surface. The ensuing
growth in grain roughness of the bed diminishes flow velocity owing to the greater boundary resistance to
flow. Furthermore, the coarsening of bed material and changes in sediment packing as well as development of
pavement and cluster bedforms on the bed surface tend to reduce particle susceptibility to entrainment. This
is important since any tendency of a river to degrade can only be realized through setting bed-material
particles in motion. A new equilibrium is thus likely to be established at flow-velocity and stream-power
levels higher than before the channelization. The secondary increase in flow velocity, caused by flow
concentration in a deepened channel, may have no immediate influence upon channel incision, owing to the
reduced grain mobility. However, any retrogressive changes of channel sediments (for example pavement
disruption, exploitation of the largest clasts from a channel bed) may activate the positive-feedback loop.

CONCLUDING REMARKS
The Raba River was channelized in its middle and lower course during the present century, and a number of
processes have been brought into operation owing to the subsequent increase in flow velocity and stream
power. Up to 3 m of river incision has occurred so far. Growth in mean grain size and sorting improvement of
channel sediments as well as changes in sediment fabric have followed the progressive outwashing of finer
grains from bed material. This has increased boundary resistance to flow and reduced particle susceptibility
to entrainment. In time, downstream magnification of peak discharges has become more pronounced and the
decrease in flood-plain storage and self-acceleration of flows passing the relatively deep and straight channel
has caused flood waves to become progressively more flashy in nature.
 RIVER CHANNEL REGULATION 555

Both the channel incision and bed-material changes must have been strengthened by the reduction in
sediment supply linked to changes in basin management. During the 20th century, the natural evolutionary
tendency of the river under conditions of abating sediment supply would comprise sinuosity increase and
slope reduction (Schumm, 1968), and therefore would be opposite to the changes caused by regulation. This
tendency may have been realized during a pause in the river-control works in the 1940s and early 1950s, and
indeed channel degradation then stopped in the middle river course. On the contrary, the shortening and
narrowing of the river, due to the repeated channelization, increased the stream power and brought the river
into disequilibrium.
A number of unintentional effects of the applied regulation scheme, unfavourable for the natural
environment and the economy, have been revealed in the channel and on the valley floor. The lowering of
the ground-water level on the valley floor, the increase in river-bank susceptibility to erosion, the decrease
or cessation of overbank-sediment accretion, and the impoverishment of plant and animal communities of
riverside biotopes are the principal negative effects of channel downcutting (Wyżga, 1991). Undermining
bridge abutments and regulation structures may supplement the list. But of primary significance seems to be
that channelization has failed to reduce flood hazard. In fact, the danger has been merely shifted downstream,
and at the same time has magnified owing to the increasingly peaked nature of flood waves passing the
channelized reach and to the channel aggradation that takes place downstream of degraded reaches. Indeed,
during the present century within the Upper Vistula River drainage basin, especially in its western part, a
marked increase in the occurrence of extremely high stages, and thus in flood frequency has been observed
(Punzet, 1981).
With the detrimental effects of the applied regulation scheme and its failure to reduce flood hazard, the
question arises as to whether the scheme should be maintained in the future. If the scheme were abandoned
channel stabilization within the valley floor would be lost. However, some mitigation of the disadvantages of
channelization would be gained, as well as the arrest of the tendency to increase flood-peak discharges. Also,
by allowing the river to develop its channel freely when meandering on the flood-plain, the formation of a
narrow, deep and sinuous channel might result, as well as the re-establishment of the conditions of flood-
plain storage and peak-discharge attenuation. The middle river course is suggested as the location of scheme
abandonment. Here there are broad zones devoid of fine-grained overbank sediments and with a very sparse
cover of riparian vegetation developed in the 20th century as a result of the river migration under conditions
of simultaneous channel degradation (Wyżga, 1991).
Two practical conclusions for river engineering may be formulated on the basis of this case study. Firstly,
channel stability depends upon the choice of an appropriately designed regulation scheme for conditions of
water and sediment transfer in the river (river regime). It is, however, a common occurrence that
environmental changes resulting from alterations in basin management or climatic fluctuations lead to
channel instability (Lewin et al., 1988). As river adjustment to environmental change is a process of long
duration (Burkham, 1972; Lewin et al., 1988), it is especially important to evaluate trends in the changes in
flood hydrographs and sediment supply under the present and forecasted style of basin management before
starting with any regulation scheme.
The second point is a corollary of the complementary nature of the streambed degradation and bed-
material changes in the process of river adjustment to channelization. Therefore, any human activity
disturbing the desirable changes of channel sediments (for example exploitation of cobbles forming channel
pavement and the destruction of sediment structures like gravel imbrication or cluster bedforms as a result
of crossing the channel by vehicles, channel dredging or remodelling by bulldozers) will inevitably result in
strengthening the tendency to channel incision.

ACKNOWLEDGEMENTS

This study was supported by the grant 6 6242 91 02 of the Committee of Scientific Research of Poland. I am
grateful to Dr. Anton Imeson, Amsterdam University, for his helpful comments and suggestions that
improved an earlier version of the paper. Thanks are also due to an anonymous referee for critical comments
on the manuscript. The rendering of measurement data by the Hydrologic Survey and the Water Authority
as well as aerial photos by the Cartographic Survey accessible to the author is kindly acknowledged.
556 B. WYŻGA

REFERENCES

Bathurst, J. C. 1982. Theoretical aspects of flow resistance', in Hey, R. D., Bathurst, J. C. and Thorne, C. R. (Eds), Gravel-bed Rivers,
Wiley, Chichester, 83-108.
Beckinsale, R. P. 1972. The effect upon river channels of sudden changes in sediment load', Acta Geographica Debrecina, 10, 181-186.
Brookes, A. 1987. 'River channel adjustment downstream from channelization works in England and Wales', Earth Surface Processes
and Landforms, 12, 337-351.
Bull, W. B. 1979. Threshold of critical power in streams', Geological Society of America Bulletin, 90, 453-464.
Burkham, D. E. 1972. 'Channel changes of the Gila River in Safford Valley, Arizona 1846-1970', U.S. Geological Survey Professional
Paper, 655-G, 1-24.
Burkham, D. E. 1976. 'Effects of changes in an alluvial channel on the timing, magnitude and transformation of flood waves,
Southeastern Arizona', U.S. Geological Survey Professional Paper, 655-K, 1-25.
Carling, P. A. and Glaister, M. S. 1987. 'Rapid deposition of sand and gravel mixtures downstream of a negative step: the role of matrix-
infilling and particle overpassing in the process of bar-front accretion', Quarterly Journal of the Geological Society of London, 144,
543-551.
Church, M. 1978. 'Palaeohydrological reconstructions from a Holocene valley fill', in Miall, A. D. (Ed.), Fluvial Sedimentology, Memoir
of Canadian Society of Petroleum Geologists, 5, 743-772.
Church, M. and Jones, D. 1982. 'Channel bars in gravel-bed rivers', in Hey, R. D., Bathurst, J. C. and Thorne, C. R. (Eds), Gravel-bed
Rivers, Wiley, Chichester, 291-338.
Daniels, R. B. 1960. 'Entrenchment of the Willow Drainage Ditch, Harrison County, Iowa', American Journal of Science, 258, 161-176.
Emerson, J. W. 1971. 'Channelization: A case study', Science, 173, 325-326.
Gregory, K. J. 1987. 'River channels', in Gregory, K. J. and Walling, D. E. (Eds), Human Impact and Environmental Processes, Wiley,
Chichester, 207-253.
Hooke, J. M. 1986. 'The significance of mid-channel bars in an active meandering river', Sedimentology, 33, 839-850.
Kędzior, A. 1928. 'Roboty wodne i melioracyjne w Południowej Małopolsce wykonane z inicjatywy Sejmu i Wydziału Krajowego', Lwów,
(in Polish).
Klimek, K. 1974. 'The retreat of alluvial river banks in the Wisłoka Valley (South Poland)', Geographia Polonica, 28, 59-75.
Klimek, K. 1987. 'Man's impact on fluvial processes in the Polish Western Carpathians', Geografiska Annaler, 69A, 221-226.
Kotarba, A. 1989. 'On the age of debris flows in the Tatra Mountains', Studia Geomorphologica Carpatho-Balcanica, 23, 139-152.
Laronne, J. B. and Carson, M. A. 1976. 'Interrelationships between bed morphology and bed-material transport for a small, gravel-bed
channel', Sedimentology, 23, 67-85.
Leopold, L. B. and Miller, J. P. 1956. 'Ephemeral streams—hydraulic factors and their relation to the drainage net', U.S. Geological
Survey Professional Paper, 282-A, 1-37.
Lewin, J., Macklin, M. G. and Newson, M. D. 1988. 'Regime theory and environmental change—irreconcilable concepts?', in White,
W. R. (Ed.), International Conference on River Regime, 18-20 May 1988, Wiley, Chichester, 431-445.
Mikulski, Z. 1954. 'Katastrofalne powodzie w Polsce' ('Inondations désastreuses en Pologne',—French summary), Czasopismo
Geograficzne, 25, 380-396.
Plumley, W. J. 1948. 'Black Hills terrace gravels: a study in sediment transport', Journal of Geology, 56, 526-577.
Punzet, J. 1969. 'Hydrological characteristics of the River Raba' (English summary), Acta Hydrobiologica, 11, 423-477.
Punzet, J. 1981. 'Changes in the course of water stages in the Upper Vistula basin over a century 1871-1970' (English summary), Folia
Geographica, Ser. Geographica Physica, 14, 5-28.
Reid, I. and Frostick, L. E. 1984. 'Particle interaction and its effect on the threshold of initial and final bedload motion in coarse alluvial
channels', in Koster, E. H. and Steel, R. J. (Eds), Sedimentology of Gravels and Conglomerates, Memoir of Canadian Society of Petroleum
Geologists, 10, 61-68.
Schumm, S. A. 1968. 'River adjustment to altered hydrologic regimen—Murrumbidgee River and paleochannels, Australia',
U.S. Geological Survey Professional Paper, 598, 1-65.
Simon, A. 1989. 'A model of channel response in disturbed alluvial channels', Earth Surface Processes and Landforms, 14, 11-26.
Wyżga, B. 1991. 'Present-day downcutting of the Raba River channel (Western Carpathians, Poland) and its environmental effects',
Catena, 18, 551-566.
Wyżga, B. in press. 'Present-day changes in the hydrologic regime of the Raba River (Carpathians, Poland) as inferred from facies
pattern and channel geometry', in Marzo, M. and Puigdefàbregas, C. (Eds), Alluvial Sedimentation, Special Publication of International
Association of Sedimentologists, 17.