Research Article Open Access
*
1 Introduction
Central European rivers have been the subject of intensive
river regulation works in the past, which caused a dramatic
decrease in the physical and biotic complexity of
riverine landscapes [1, 2]. Lowland meandering rivers were
the most affected as they were systematically engineered
from the nineteenth century in order to acquire new arable
land, maintain flood protection, and enhance river navigation.
The traditional empirical approach towards the understanding
of meandering rivers focused primarily on determination
of the variables which control planform geometry
and the evolution of their channels [3–6]. Later, theoretical
approaches dealing with the numerical modelling
of the physical basis of meandering [7–10], and physical
laboratory experiments [11–13] complemented empirical
studies.
A substantial body of knowledge on the dynamics of
meandering rivers is now available. Nonetheless, as indicated
by [14], many gaps in research still exist, and thus
new research themes have arisen as well in meandering
river research. In Europe and North America, considerable
effort has been put into establishing linkages between
the knowledge gained on the dynamics of meandering
rivers and river management and restoration practices.
Just as empirical research into meandering river dynamics
focused, in the past, primarily on natural rivers, it is now
evident that attention should also be paid to meandering
systems affected directly or indirectly by human interven-
tion.
Understanding the dynamics of quasi-natural (artificially
influenced) meandering rivers is one of the key
prerequisites for the development of appropriate strategies
targeting the restoration of artificially controlled river
reaches to attain an ecologically favourable state in accordance
with the European Water Framework Directive
(2000/60/EC)¹. Therefore, traditional research of meander-
1 European Parliament and Council, Directive 2000/60/EC of the European
Parliament and of the Council of 23 October 2000 establishing
a Framework for Community action in the Field of Water Policy, 2000.
ing rivers is receiving new impulse from river managers
who demand models of possible development trajectories
of newly formed (revitalised) meandering channels
where only straightened channelized streams existed for
many decades. In this context, with regard to meandering
river research, the key questions that need to be addressed
are: (1) how do bank erosion and channel migration affect
production of aquatic and semi-terrestrial habitats in
floodplains (e.g. through cut offs and oxbow lake formation)
[15–17]?; (2) how should channel migration zones be
delimited to mitigate the conflicts between the dynamic
changes of meandering channels and economic activities,
typically arable land erosion and disruption of transportation
infrastructure [18–20]?; (3) what controls channel migration
rates of meandering rivers [21–23]?
In the Czech Republic, as a consequence of riverstraightening
practices, meandering reaches of large lowland
rivers have almost completely disappeared. Examples
of the last river reaches with a spontaneous development
of meandering can be found on the middle and lower
course of the Morava River in the Litovelské Pomoraví
(40 km in length) and the Strážnické Pomoraví (13 km in
length) regions. The latter region represents an interesting
example of the quasi-natural development of meandering
river reaches after significant channel and floodplain engineering
adjustments were undergone in the 1930s. The aim
of this paper - with special focus on Strážnické Pomoraví
- is to: (1) quantify the spatio-temporal evolution of channel
migration rates and related changes in channel geometry;
(2) test the relationships between the pattern of channel
migration and environmental and geometric variables;
and (3) characterise the influence of channel engineering
adjustments on the channel migration rates. Our contribution
brings further information about the behaviour of a
medium-sized meandering river flowing through a highly
transformed settlement-agricultural landscape with various
lithologies, land use categories and a complicated history
of river-engineering practices.
2 Geographical setting
The Morava River drainage basin is the largest fluvial system
in the eastern part of the Czech Republic. Above the
confluence with the Danube, the Morava has a length of
353 km with a catchment area of 26,578 km2
. The Morava
rises in the Králický Sněžník Mountains close to the CzechPolish
border at an altitude of 1,371 m and leaves the
territory of the Czech Republic in the lowlands of south
Moravia after 269 km of its course and has catchment area
of 25,273 km2
at an altitude of 148 m. The lower reach of the
river forms a natural border between the Czech Republic
and Slovakia, and further downstream between Slovakia
and Austria. Among the most important tributaries are the
Bečva River (1,626 km2
), draining the eastern part of the
catchment, and the Dyje (Thaya) River (13,419 km2
), draining
half of the whole catchment.
The study area is a small segment of the floodplain
stretched along the 5.5 km long reach of the River Morava
between the Strážnice, Bzenec-Přívoz and Rohatec settlements
in south-eastern Czech Republic. It is located in
the Dolnomoravský úval Basin (northern tip of the Vienna
Basin, altitude between 150-200 m), where the Morava
River floodplain is on average 3.5 km wide (Figure 1). Sedimentary
infill of the Dolnomoravskýúval Basin consists
of claystones, siltstones, and sandstones of Miocene age.
Overlying Quaternary sediments are represented by Upper
Pleistocene sands of the so-called Moravian Sahara
on the western side side of the floodplain. These deposits
are originally of lacustrine origin, however they have been
extensively reworked by wind action and moulded into
the form of sand dunes [24]. These sands are eroded by
lateral erosion of the Morava and form unstable concave
banks reaching a height of up to 15 m (Figure 2). Alluvial
sediments are formed by a layer up to 10 m thick of
sandy gravels of Upper Pleistocene to Holocene age that
are covered by Holocene flood loams. Uplands adjacent
to the alluvial valley are composed of flysch sediments of
the Magura Nappe unit of the Western Carpathians, whose
thrust fronts are oriented from SE to NW [25].
The Strážnice stream flow-gauging station (river 133.5
km, drainage area 9,145.8 km2, mean annual discharge
59.6 m3
s−1
), operated by the Czech Hydrometeorological
Institute since 1940s, is located at the upstream end of
the study reach. The history of water level and discharge
measurements extends back to 1886. Discharges with Nyears
recurrence intervals are summarised in Table 1. Since
the 1920s, discharges have reached Q100 once, Q50 twice,
Q50 once, Q10 six times, Q5 eighteen times, and Q2 twentyeight
times (Figure 3). Among the highest frequencies of
floods are decades 1961-1970, with eleven events, and 2001-
2010, with seven events [26]. So far the highest peak discharge
at Strážnice, a flow of 901 m3
s−1
, was recorded on
10 July 1997, during the ‘flood of the 20th century’ which
exceeded the value of Q100.
Today, the Morava River in the Strážnické Pomoraví region
has a meandering pattern with a sinuosity of 1.53, the
width of the meander belt reaches 700 m, the channel bed
has a very gentle slope of 0.36 , and mean channel width
is 69 m. The river bed is formed mainly by sand and gravel,
whereas the river banks are mostly formed by cohesive maBrought
to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
Figure 1: Study reach of the Morava River within the Czech Republic (A), the whole drainage (B), and the Strážnické Pomoraví region (C).
Figure 2: High river bank formed by aeolian sands of the Moravian Sahara in the locality of Osypané břehy, Strážnické Pomoraví region.
Note the contact of flood loams and aeolian sands on the left-hand side of the image. (Photo by J. Ondruch, 2011).
Brought to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
terial of flood loams with a thickness of up to 6 m. In some
bank profiles unconsolidated sand material prevails [27].
Until the 1920s, an anabranching pattern with straight
as well as meandering river channels was characteristic
of the study river reach (Figure 4). As a consequence of
river regulation works starting in the 1920s, the functionality
of smaller anabranching channels had weakened and
the flow was progressively concentrated into a single dominant
channel. Tributaries and side channels were progressively
channelized and a network of drainage ditches
was built in order to increase the area of arable land. The
length of the dominant channel was shortened by the artificial
cut off of two sequences of meanders. A significant
reduction in the connectivity between channels and
floodplain emerged with the construction of flood-defence
dykes, which completely obstructed the overbank flows of
the Baťa channel (former Morávka side channel) and the
Radějovka River and partly of the Morava River itself, and
the inundation area dropped to 20% of its original extent
[28]. River regulations in 1920s-30s triggered the incision
of a dominant channel. Currently, the river bed is approximately
as low as 6 m beneath the floodplain. As a consequence,
the frequency of overbank flows has significantly
decreased and the floodplain has gradually become hydrologically
disconnected; where, at the end of the nineteenth
century, the overbank flows occurred even several times
a year, today the bankfull stage is reached only with five
years recurrence interval [29].
3 Material and methods
A GIS analysis of the lateral migration of the Morava channel
in the period 1938-2012 was conducted for the 5.5
km long river reach (upstream most part of the 13 km
long (quasi)natural course of the Morava River). Available
aerial images were provided by the Military Geographical
and Hydrometeorological Institute (VGHMÚř) and the
Czech Office for Surveying, Mapping and Cadastre (ČÚZK).
The first aerial images were taken in this area shortly after
significant river regulation works of the Morava were made
in the 1930s.
First, georeferencing of aerial images was done in ArcMap
10.0 software. All images were georeferenced according
to the orthophotomap produced by ČÚZK from
the aerial images taken on 20 October 2012 on a scale
of 1:1,000. Polynomial transformation of first order was
employed with the use of at least 18 reference points
and a polynomial transformation of second order with
25 points. As reference points, corners of buildings, engineering
constructions on drainage ditches, crossings
of roads or drainage ditches, margins of distinctive land
cover patches, and solitary trees, were used. If the mean
square errors were lower than 5 m, georeferencing was accepted
and an aerial image was rectified with a grid resolution
of 0.1 m.
For a quantification of spatio-temporal variability of
the channel migration rates, seven periods were used
(1938-1953, 1953-1963, 1963-1973, 1973-1982, 1982-1993, 1993-
2003, 2003-2012). These intervals result from the availability
of aerial images enabling document channel changes
with a relatively high temporal resolution. For every respective
year, bank lines were digitalised from georeferenced
images on a uniform scale of 1:1,000. In non-forested
segments, where the floodplain was covered by grassland
or arable land, the bank line on the concave side of meander
bends was clearly visible in the incised channel. On
the convex sides of meander bends the floodplain was distinguished
from point bars by vegetation cover. In forested
segments, bank lines were delineated by connecting visible
fragments of bank lines with the central points of treetops
that were the most adjacent to the water surface. This
approach assumes a circular shape of treetops and vertical
position of trees growing directly on the bank line.
However, this is not always the case in the study area and
therefore an error could arise. Nonetheless, the river channel
is deeply incised into the floodplain and furthermore
many aerial images were taken in the winter season when
leaves were missing. Even in forested segments, there were
many stretches with clearly visible bank lines, substantially
decreasing the possibility of this type of error occurring.
Due to the incision of the Morava into the floodplain,
delineation of both bank lines was not affected by
the differences in the water level in the years when aerial
photographs were taken. Despite discharge data not being
available for the situations caught on aerial images, bankfull
discharge was not reached in either year, thus even the
inner banks were clearly visible and possible to delineate.
Based on the methodology of [30], lateral migration
polygons (polygons delineated by the position of the bank
line in times t and t +1) were constructed for every respective
period for both bank lines (Figure 5). In previous studies,
authors used either bank lines or centrelines for describing
the pattern of lateral channel migration. Hooke
and Redmond [31] found that maximum values of migration
are higher for polygons constructed from bank lines
than for those from centrelines. Nonetheless, they are both
of the same order. The advantage of the approach based on
bank lines is that it provides the possibility to also study
spatio-temporal changes in channel width and thus to calBrought
to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
Table 1: The list of the N-years discharges of the Morava River at the Strážnice gauging station (133.5 river kilometre, drainage area 9145.84
km2) for 1940-2014. (Source: [26])
Q1 Q2 Q5 Q10 Q20 Q50 Q100
Q (m3
/s) 375 440 525 588 649 730 790
Figure 3: Magnitude of floods with a recurrence interval equal to or higher than Q2, which were documented at the gauging station
Strážnice, the Morava River, between 1938 and 2012. If more than one flood with a recurrence interval equal to or higher than Q2 occurred,all
events are differentiated by colour (Qmax represents the largest, Q2max the second largest, and Q3max the third largest flood
for particular year). N-years discharges are highlighted by horizontal solid lines. Years when aerial images were taken are highlighted by
vertical dashed lines.
Figure 4: Drainage system and flood defence of the Morava River in Strážnické Pomoraví region in 1882 and 2012. Note that roads and railways
are delineated only in Strážnické Pomoraví region for better clarity of the figure.
Brought to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
Figure 5: (A) Methods for quantification of lateral channel shifts utilising
the lateral migration polygons approach. An average migration
for the whole polygon is calculated as a polygon area divided
by its perimeter.
culate the differences between the rates of erosion and accretion
on opposite channel banks.
Employing the technique of lateral migration polygons,
we quantified the rates of erosion (Er), accumulation
(Ar), and channel width change (∆W). Mean channel
shift for the period analysed is then calculated as a polygon
area divided by half of its perimeter [30]. For a detailed
description of spatio-temporal changes of Er, Ar and ∆W,
lateral migration polygons were further divided into 25 m
long segments, similar to the studies of [21, 32, 33], representing
in the case of the Morava approximately half of the
channel width. For every segment, a mean channel shift
was calculated as its area divided by a half of the bank line
length in times t and t + 1.
Patterns of channel migration rates were visualised by
cumulative curves of Er, Ar and ∆W plotted against a distance
from the beginning of the study reach. The construction
of such curves for different time periods enables the
determination of laterally active and stable river reaches
and their comparison in time.
On three spatial scales (reach, segments, and individual
meanders), relationships between measures of lateral
channel migration (Er and Ar) and anticipated explanatory
variables, bank material, floodplain land cover, flood
frequency and magnitude, and channel geometry (radius
of curvature, sinuosity, channel width), were tested employing
non-linear correlation and Kruskal-Wallis nonparametrical
ANOVA.
Eroded floodplain material was identified from aerial
images and verified by the field validation exercise. Information
on flood frequency and magnitude in the studied
period was supplied by the Czech Hydrometeorological
Institute (the Strážnice gauging station). Floodplain land
cover for the respective years was delineated from aerial
images on a scale of 1:1,000. Land cover patches were delineated
only if their area exceeded 100 m2
. Only one, prevailing,
land cover category was further assigned to migration
polygons.
4 Results
4.1 Changes in the overall channel
planimetry
The channel of the Morava River in the Strážnické Pomoraví
region has experienced significant changes to its
planform in the past 75 years (Figure 6). As a consequence
of erosion, 60.43 ha of river floodplain were removed between
1938-2012. Accumulation on the convex bank was
lagging behind the erosion with 44.12 ha of new floodplain
developed in the same period. The most dynamic
evolution of the channel was recorded in the first time
interval, 1938-53 (floodplain erosion EF = 1.37 ha/year;
floodplain accretion AF = 0.69 ha/year). In the following
period, 1953-63, the rate of erosion decreased (EF = 1.07
ha/year), whereas the rate of accumulation rate increased
(AF = 0.90ha/year). Both processes were, nonetheless, still
reaching relatively high values in comparison with the remaining
periods. For the period 1963-2012, the stabilisation
of rates of lateral channel migration is characteristic
of the fluctuation between 0.43-0.79 ha/year for erosion
and 0.31-0.62 ha/year for accumulation. Of key importance
regarding the planform changes of the channel
documented for the period 1938-53 is the fact that 34.1%
(20.62 ha) of the total eroded area of the floodplain falls
within the opening time period. Similarly, in this relatively
short time, 23.5% (10.38 ha) of total newly formed floodplain
area was formed here (Table 2). Differences between
erosion rates in concave banks and accumulation rates on
convex ones caused an increase of the active channel area
by 16.31 ha, while the most rapid increase was also documented
in 1938-53 (10.28 ha, i.e. 63% of total area).
The reach average rate of bank line retreat (river bank
erosion) for a whole studied period was 1.04 m/year; the
average accretion rate was 0.75 m/year. The average river
bank erosion rate (Er) reached maxima of 2.19 m/year in
the opening time period (standard deviation (SD) = 2.1
m/yr). Maximum erosion was documented in the area of
meander M7, where local Er reached 9.8 m/year in the first
Brought to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
Figure 6: Changes in the planform pattern of the channel in the 5.5 km long study reach of the Morava River in the Strážnické Pomoraví region
for selected periods between 1938 and 2012. M1-M10 addresses the codes for meanders which were selected for the detailed study of
the relationships between spatio-temporal variability of the migration rates (Er and Ar) and selected controlling variables. Note the artificial
cut off of 6 meanders, which took place in 1930s just before the aerial image in 1938 was taken.
Table 2: Area of eroded (EF) and accreted floodplain (AF), difference in rate of floodplain erosion and accumulation (∆EAF = EF − AF), erosion
rates Er, accumulation rates Ar, and rates of the channel width changes (∆w/∆t) along a 5.5 km stretch of the Morava River in the Strážnické
Pomoraví region for the periods between 1938 and 2012. An average and median (in brackets) is given in the table for Er, Ar, ∆w/∆t.
Period EF (ha) ∆EF (ha/year) AF (ha) ∆AF (ha/year) ∆EAF (ha) Er (m/yr) Ar (m/yr) ∆w/∆t (m/yr)
1938-1953 20.62 1.37 10.38 0.69 10.24 2.19 (1.5) 1.43 (0.7) 0.75 (0.6)
1953-1963 10.67 1.07 9.02 0.9 1.64 1.47 (0.8) 1.35 (0.9) 0.04 (-0.1)
1963-1973 6.72 0.67 6.2 0.62 0.52 0.87 (0.6) 0.89 (0.5) -0.16 (0.0)
1973-1982 5.06 0.56 4.81 0.53 0.25 0.69 (0.6) 0.77 (0.5) -0.08 (0.1)
1982-1993 4.7 0.42 5.56 0.51 -0.86 0.53 (0.5) 0.70 (0.5) -0.26 (-0.1)
1993-2003 7.86 0.79 3.13 0.31 4.73 0.83 (0.6) 0.39 (0.0) 0.50 (0.3)
2003-2012 4.81 0.53 5.03 0.56 -0.22 0.71(0.4) 0.91 (0.7) -0.09 (-0.2)
total 60.43 0.77 44.12 0.59 16.31 1.04 0.92 0.1
Brought to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
15 year and the concave bank moved laterally by 147 m.
The period 1938-53 also experienced the maximum accretion
rates (Ar), although the rate was significantly lower
(1.43 m/year, SD = 2.12 m/year). High values of SD result
from alternating channel segments with low and high lateral
mobility, causing considerable scatter of Er and Ar values
for the entire river reach. An absolute local maximum
of Ar was reached 100 m downstream from the local erosional
maximum with a value of 11 m/year that caused a
shift of the inner bank by 165 m. Minimum Er falls within
1982-93 (Er = 0.53 m/year, SD = 0.50m/year), whereas the
minimum accretion rate (Ar = 0.39 m/year, SD = 0.61) occurred
one period later (1993-2003).
The channel length of the study reach was 4,978 m in
1938. During the first fifteen years, an increase in length
reached 830 m (on average increasing by 56.7 m/year). After
this, the increase was relatively stable until 2003, with
an average of 26.1 m/year, the length of the channel reaching
7,111 m in 2003. During the final period, 2003-12, local
shortening took place as a consequence of a natural cut
off (meander M6) in 2006. In 2012, the study reach was
6,459 m in length. Changes in channel sinuosity are related
to length development. Average reach sinuosity with the
value of 1.31 in 1938 increased, in the opening period, to
1.47. In the following 50 years, sinuosity slowly increased
to 1.80 in 2003. A neck cut off caused a decrease in sinuosity
that subsequently reached 1.63 in 2012.
Average channel width was 57.1 m (SD = 15.8) in 1938.
During the opening period the channel widened, on average,
by 19% to 67.9 m (SD = 14.4). In the remaining periods,
only slight variations in width took place (66–68 m).
In the study reach, spatial variability in erosion and
accretion is evident. Figure 7 depicts cumulative values of
Er, Ar and ∆W for the whole reach. Based on the shapes
of the depicted curves in the charts, the reach was subdivided
into six segments. Laterally stable segments are located
between 0-500 m, 1250-1900 m, and 3400-4800 m.
Between these segments with minimal channel planform
changes, segments with a relatively dynamic development
are located, which are related to meander bends consisting
of 2 to 5 meanders. On average, we characterised laterally
dynamic segments by three to fourfold higher rates of
lateral shifts (Er, Ar). Figure 8 summarises the descriptive
statistics of Er and Ar for stable and dynamic segments.
it Segment 1 (stable), 0-500 m. The first segment represents
a relatively stable bend situated immediately downstream
from the bridge that divides the freely meandering
reach from the channelized one. Values of Er varied between
0.15-0.80 m/year for all periods with a maximum
in the opening period 1938-1953, and then a minimum in
1982-93. In all periods, values of Er were lower than reach
average. Sediment accretion rates varied between 0.10-1.07
m/year; with the maximum in 1953-63 and the minimum in
1993-2003. Extreme values of Ar are, then, delayed by one
decade behind the Er extremes. Similar to the case of Er,
in all periods Ar reached below-average values. Channel
narrowing (Ar>Er) was documented in periods 1953-63 and
1982-93 with rates of 0.52 and 0.42 m/year, respectively. Remaining
periods experienced channel widening (with the
highest rate of 0.34 m/year in the opening period 1938-53).
Segment 2 (dynamic), 500-1250 m. Stable segment #1
is followed by the sequence of three bends (M1-M3). Values
of Er ranged from 0.50-2.11 m/year; with the maximum in
1938-53 and the minimum in 1953-63. Above-average values
in comparison with the whole reach in the segment
were documented in all periods from 1973. Significant reduction
of erosional activity occurred in 1953-63, when
Er of the segment reached as low as 34% of reach average.
Sediment accretion rates (Ar) varied between 0.49-
1.54 m/year; with the maximum in the opening period.
Minimum Ar was reached in the two periods; 1963-73 and
1993-03. The minimum documented in 1993-2003 was still
higher than reach average, but the other minimum of Ar
had a value that represented only 56% of the reach average.
Except for the latter period, all time intervals were
characterised by above-average values of Ar. As a consequence
of secondary maximum and minimum in 1993-
2003, significant channel widening occurred, at a rate of
1.32 m/year. Considerable widening also occurred between
1938-53, at a rate of 0.56 m/year. Periods 1953-63 and 1982-
93 documented channel narrowing by 1.09 m/year and
0.37 m/year, respectively.
Segment 3 (stable), 1250-1900. This segment represents
the most stable part of the study reach. Values of
Er varied in the range of 0.16-0.61 m/year, with two maxima
in 1938-53 and 1953-63. In all periods, the values of Er
moved within the below-average interval. Sediment accretion
rates ranged between 0.04 and 1.39 m/year. The most
important period for new floodplain establishment was
1953-63, when a side bar, located by the downstream end
of the segment, was attached to the floodplain. This period
also represents the only period when Ar exceeded, slightly,
the average reach value. As a consequence of meander
(M3) downstream translation, the closing period (2003-
12) experienced relatively higher Ar (0.59 m/year). Channel
widening occurred in 1938-53 (0.45 m/year) and 1963-73
(0.33 m/year). Significant channel narrowing (0.75 m/year)
was documented in 1953-63.
Segment 4 (dynamic), 1900-3400 m. This highly dynamic
segment is represented by four mature bends (M4M7).
Values of Er varied in the range of 0.78-4.27 m/year.
Maximum was reached in the opening period and even
Brought to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
Figure 7: Cumulative erosion rates (A), accretion rates (B) and rates of the channel width change (C) in the 5.5 km long study reach of the
Morava River in the Strážnické Pomoraví region for selected periods between 1938-2012. Dashed lines divide the reach into individual dynamic
and stable segments.
Brought to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
Figure 8: Box plots of erosion (Er) and accretion (Ar) rates for individual segments in the 5.5 km long study reach of the Morava River in
the Strážnické Pomoraví region for selected periods between 1938 and 2012. Median value is highlighted by the line, the upper and lower
margin of the box represent 1st and 3rdquartile and whiskers explains the 1.5 times of IQR. Empty circles depict extreme values.
the following period preserved very high values, with Er
>3 m/year. The minimum was documented in 1982-93 (the
only period with Er < 1 m/year). All periods exceeded the
reach’s average Er values. Three intervals delimited by the
years 1938 and 1973 were characterised by Er as high as
twice the reach average. Values of sediment accretion rates
ranged between 0.61 and 3.17 m/years. In contrast with the
first dynamic segment (500-1250 m), the maximum was
documented in the opening period (1938-53). The maximum
was followed by a gradual decrease to 0.61 m/year
in 1993-2003, which represented the only period with Ar <
1 m/year. Similarly, in comparison with Er, values of Ar exceeded
reach average; the first two periods reached double
the average reach values. Channel widening occurred in
1938-53 at a rate 0.73 m/year and 1953-63 (0.55 m/year). The
following three periods experienced channel narrowing at
the rate of 0.75, 0.19, and 0.55 m/year, respectively, which
balanced the previous channel widening. In 1993-2003 and
2003-12, subsequent widening occurred at the rate of 0.77
and 0.34 m/year, respectively.
Segment 5 (stable), 3400-4800 m. This segment is
characterised as being very stable and with both its Er
and Ar values being gradually increased by meander M8,
which was formed during the opening period at the most
upstream part of the segment and which immediately follows
meander M7. Development of the segment was significantly
influenced by river regulation works that took
place in the 1930s (immediately before the time when the
aerial photo was taken in 1938), which cut off the meander
bend compound of six meanders. Values of Er ranged
from 0.31-0.95 m/year (lower than the reach average for all
periods) with the maximum in 1938-53 and the minimum
in 2003-2012. Sediment accretion rates (Ar) varied between
0.30 and 0.68 m/year, with the maximum in 1973-82 and
the minimum in 1953-63. All time intervals until 1973, as
well as the closing one, documented values of Ar that was
significantly lower than reach average. Between 1973 and
2003, values were comparable with the reach average. The
first two periods experienced channel widening, at a rate
of 0.86 and 0.57 m /years respectively, which was followed
by a stabilised channel width for remaining periods.
Segment 6 (dynamic), 4800-5500 m. The study reach
of the Morava River is closed by two meander bends (M9
and M10), which are situated downstream from the confluence
with the Velička River and the artificially cut off meander
bend. Erosion rate (Er) varied in the range 0.26-2.87
m/year with the maximum in the opening period (1938-53)
and the minimum in the last period (2003-12). Since 1973-
82, Er stabilised at low values (0.26-0.55 m/year). Aboveaverage
Er was reached only in the opening period. Sediment
accretion rates fell into the range of 0.31-1.50 m/year
with the maximum in 1938-53 and the minimum in 1982Brought
to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
Figure 9: The evolution of the erosion (Er) and accretion (Ar) rates
for individual meanders in the 5.5 km long study reach of the
Morava River in the Strážnické Pomoraví region for selected periods
between 1938 and 2012.
93. In 1938-53 and 1963-73, above-average values of Ar
were documented, and in 1973-82 sediment accretion rates
equalled the reach average. Significant channel widening
took place in 1938-53 and 1953-63 at a rate of 1.69 and 0.46
m/year, respectively. Channel narrowing was documented
in the segment in 1963-73 (0.35 m/year) and 2003-12 (0.53
m/year).
4.2 Temporal changes of selected meanders
Special focus was paid to ten meanders (M1-M10) that are
situated in the study reach. The bends were used to test the
effects of channel geometry, flood plain land cover and hydrology
(flood frequency and magnitude) on lateral migration
rates (both Er and Ar). The average length of the bends
varied between 277 and 739 m with the average width being
between 55 and 70 m. All meanders, excluding meander
M7, are classified (sensu [34]) as simple symmetric (M3-M6,
M8-M10) and simple asymmetric (M1, M2). In the time period
1963-73, the straight segment in the apex of meander
M7 was formed and the meander transformed from simple
symmetric to compound symmetric. For the study meanders,
an evolution of erosion rate (Er) and accretion rate
(Ar) is depicted in Figure 9.
During the opening period (1938-53), apparent extension
and downstream migration took place on all meanders
except for M7, which experienced extension complemented
by rotation in the downstream direction. In
the period that followed, substantial translation was documented
on meanders M4-M6, which, on meander M6,
was supplemented by extension. Meander M7 was characterised
by an intensive extension. The other meanders remained
more or less stable, and experienced slight extension
and downstream translation, respectively. The same
pattern of meander development was preserved in 1963-73.
On meander M7, significantly higher dynamics in terms of
planform development was apparent on the downstream
bend, which caused a formation of the straight segment
and compound meander. This process also conditioned
the lateral translation of meander M8. The time intervals
1973-82 and 1982-93 documented a decrease to a cessation
of meander translation (except for meander M6) and
a slight extension on meanders M1-M8. Between 1993 and
2003, more intensive extension occurred on meanders M1M8.
In the closing period, a cut off of meander M6 took
place, which conditioned the development of meanders
M6 and M7 (upstream bend). In this period, meanders
M1-M3 experienced a relatively significant extension and
downstream translation. Meander M8 also experienced extension,
but without translation.
4.3 Determination of controlling factors
The control of geometrical variables on changes in channel
planimetry (Er, Ar, and ∆W) of the whole reach, dynamic
segments, and selected meanders, was tested employing
non-parametric correlation. Geometrical variables
were represented by channel width (W), sinuosity (SI) and
radius of curvature divided by channel width (Rc). Results
are summarised in Table 3.
On the reach scale, moderate correlation was ascertained
between Er and sinuosity (r = -0.69; r2
= 0.48), Ar
and sinuosity (r = -0.79; r2
= 0.62) and no significant correlation
was found for rate of channel width change. For
dynamic segments (i.e. segments 2, 4, and 6), correlation
was ascertained between Er and channel width (r = 0.37; r2
= 0.14) and Ar and channel width (r = 0.46; r2
= 0.21). Rate
of width change did not correlate with geometric variables.
On the scale of individual meanders, positive correlations
were found between Er (Ar) and channel width and sinuosity,
and negative correlation was found between Er (Ar)
and radius of curvature divided by channel width.
As hydrological variables, we determined the number
of floods with a recurrence interval equal to or higher than
Q2(n ≥ Q2) and magnitude of the largest flood (Qmax). On
the whole reach scale, equally tight correlation was ascertained
between Er and Ar, respectively, and n ≥ Q2 (r = 0.57;
r2
= 0.32). Er also correlated with Qmax (r = 0.32; r2
= 0.10).
Rate of channel width (∆W) change correlated negatively
with Qmax (r = -0.60; r2
= 0.36). If only dynamic segments
were taken into account, moderate correlation was found
only between Ar and n≥Q2 (r = 0.41; r2
= 0.17), and a negative
correlation between ∆W and Qmax (r = -0.51; r2
= 0.26).
On the scale of individual meanders, a positive correlation
was ascertained between Ar and n≥Q2 (r = 0.30; r2
= 0.09)
and ∆W and Qmax (r = 0.42; r2
= 0.18).
Brought to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
Employing the non-parametric Kruskall-Wallis test as
an alternative to ANOVA for the dataset possessing otherthan-normal
distribution, the control of floodplain land
cover on the lateral dynamics of the channel was tested.
The land cover was categorised into three categories; (1)
arable land (and the area of open-pit sand mining); (2) cultivated
meadows and clearings; and (3) forest. A statistically
significant difference was proven (α=0.05) between
the first and second category for Er (p = 0.00602), and not
proven for Ar (p = 0.0893). Further, the difference between
arable land and forest was proven for both Er and Ar (p
= 0.00014 resp. p = 0.0066). For ∆W, a statistically significant
difference between categories was not proven for any
variables (p = 0.065).
In the Strážnické Pomoraví region, two lithological facies
are eroded by the Morava River; flood loams and aeolian
sands. The control of river bank material on the planform
evolution of the channel is not statistically significant
for the whole reach (p = 0.4625 for Er, p = 0.6041
for Ar, and p = 0.6599 for ∆W). Nonetheless, local influence
of the bank material is apparent from a visual comparison
of aerial images. The control of material on Er
could be proven, to a certain level, when only meanders
M1, M3, M5, and M7 entered the analysis. These meanders
partly eroded along their course both flood loams and
aeolian sands, in the periods since 1973. Meanders were
divided into 25m long segments and information about
river bank material was obtained for all. With the use of
the Mann-Whitney test, differences between Er for flood
loams and aeolian sands were tested. A statistically significant
difference for Er was proven for the periods 1982-93
(p = 0.00001647) and 1993-03 (p = 0.02233). In the closing
period (2003-12), erosion rates are comparable for flood
loams and aeolian sands without proven difference (p =
0.9733).
5 Discussion
5.1 Planform changes
The study reach of the Morava River is characterised by an
alternating pattern of segments with a high and low dynamics
of planform channel development. High rates of
lateral shifts are focused into three segments with developed
meanders (sinuosity 1.17-2.37), whereas intervening
segments with very low rates are straight or formed into
slightly curved bends (sinuosity 1.05-1.18). Such a spatial
pattern remained unchanged for all studied periods. Before
the river regulation works in the first half of the twentieth
century, the Morava River in Strážnické Pomoraví region
had been characterised as an anastomosing river. The
studyreach is situated on the former dominant anastomosing
arm, which has preserved, according to evidence from
old maps, its meandering pattern since at least the second
half of the eighteenth century. The current spatial pattern
of dynamic and stable segments distribution had already
been documented by [26] who employed visual analysis of
old maps from the second half of the nineteenth century.
While the channel migration rates varied significantly in
that time and were conditioned primarily by river engineering,
the spatial pattern was rather persistent. A similarpattern
of spatio-temporal relationshipsin the development
of a meandering channel was described on the river
Dane by [35].
Rates of the planform evolution of the Morava River in
Strážnické Pomoraví in the nineteenth century were studied
by [36] and [37] with the use of GIS analyses of maps
from the second and third Austrian Military Survey on a
scale of 1:28,800 and 1:25,000, respectively. The results of
[36] suggest that between 1841 and 1938 an average meander
migration rate varied between 0.35 and 1.09 m/year,
while values for meanders eroding the floodplain with
arable land were twice as high as meanders in the forest
or permanent grassland. Our results show that during the
first 25 years, after river regulation works, lateral migration
rates (both Er and Ar) reached 1.5 to 2.5 times higher
values in comparison with the nineteenth and beginning
of the twentieth centuries. In the following period, lateral
migration rates significantly decreased and equalled values
from the nineteenth century. We estimate the response
time to accommodate the disturbance caused by river regulation
was approximately 25 years.
Methods for the quantitative description of erosion
(accretion) rates differ in individual studies, therefore it
is relatively difficult to compare the erosion rates of the
Morava River with other rivers. Nonetheless, it is evident
that in the period immediately following the river regulation
works (1938-53), the whole reach was developing with
significantly higher erosion rates compared to rivers of the
same size order; e.g. Big Fork River: Er = 0.72 m/year [38],
Beaver River: 1.41 m/year [39], Swan River: 1.52 m/year [39].
The most dynamic segment of the Morava River (segment
4) reached lateral migration rates of the same order of magnitude
as substantially larger rivers (e.g. the Sacramento:
5.5 m/year [30], Prophet River: 2.34 m/year [39], the Fort
Nelson: 4.44 m/year [39], the Brazos River: 3.28 m/year
[40]). Similar behaviour was documented by [35] on the
Dane River, which reached local maxima of Er of around
3 m/year and likewise exceeded the average rates of rivers
of the same order.
Brought to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
Table 3: Correlations between geometrical (W, SI, and Rc) and hydrological
variables (n≥Q2, Qmax) and variables describing lateral
dynamics (Er, Ar, and ∆W) for whole reach dynamic segments and
selected meanders.
Reach
W SI Rc n≥Q2 Qmax
Er 0.07 -0.69 - 0.57 0.32
Ar 0.21 -0.79 - 0.57 0.06
∆W -0.21 -0.29 - -0.19 -0.60
Dynamic segments
W SI Rc n≥Q2 Qmax
Er 0.37 0.14 - 0.2 0.24
Ar 0.46 0.14 - 0.41 -0.05
∆W -0.04 -0.07 - -0.08 -0.51
Meanders (M1-M10)
W SI Rc n≥Q2 Qmax
Er 0.5 0.4 -0.47 0.26 0.18
Ar 0.54 0.43 -0.43 0.3 -0.08
∆W -0.09 -0.07 0.01 -0.11 0.42
5.2 Variables controlling planform changes
We discuss here several controlling variables, which were
determined to play a key role in studies of other authors.
Correlations of tested variables with erosion rates are, in
general, rather weak, which might be due to limited range
of values (studied reach is 5.5 km in length) as well as the
fact that several variables appears to play an important
role, which, in turn, decrease the strength of correlation
for individual variables.
Channel width was described as an important factor
for comparison of lateral migration rates between particular
reaches as well as rivers [41, 42]. Our results prove a spatial
pattern with wider channels in meander bends as well
as dynamic segments with higher Er and Ar, and narrower
channels in sites with lower values of Er and Ar. This pattern
could be explained by the larger difference between
Er and Ar in the more dynamic meander bends, where accretion
lags more behind erosion. Such spatial distribution
was also referred to by other studies [e.g. [41, 43, 44]. In
the case of study reach of the Morava River, we conclude
that channel width reflects differences between Er and Ar
rather than acts as controlling variable determining rates
of erosion and/or accretion.
Before artificial shortening in the 1930s, reach sinuosity
was as high as 1.59 and such value was reached again
in the period 1973-82, in comparison with 1.31 after cut
off. In the opening studied period, rate in sinuosity increase
doubled in comparison with other periods before
cut off (with 1.47 in 1953). The neck cut off event in 2006
led to the decrease in sinuosity from 1.80 to approximately
1.63 (real sinuosity right after cut off was slightly lower
as meanders underwent development between 2006 and
2012). Assuming that river development is controlled by
self-organisation processes (e.g. [45, 46]) we can estimate
that the Morava River within the study reach attains contemporary
(quasi)equilibrium state around a sinuosity of
1.60.
Sinuosity belongs to features commonly used as a descriptor
of channel planform. A general theory for sinuosity
was recently proposed by [47]. On the reach scale,
sinuosity correlates negatively with both Er and Ar. This is
mainly due to low sinuosity values in the opening two periods
(1938-63). We argue that acceleration of erosion occurs
when sinuosity suddenly decreases, thus sinuosity could
partly describe the lateral dynamics of the Morava River in
the study reach. Similar conclusions could be drawn from
the works of [45, 46, 48].
The correlation could also explain the difference in the
extent of the accelerated erosion as a consequence of artificial
and natural cut offs. Whereas the decrease in sinuosity
related to artificial cut off (which was followed by
extended acceleration of erosion) reached 0.27, a natural
cut off (with rather local acceleration of erosion as a consequence)
caused a decrease in sinuosity only by 0.17. On
the other hand, on the scale of individual meanders, sinuosity
correlates positively with both Er and Ar. This fact can
be understood as sinuosity explains both spatial and temporal
variability (as opposed to only temporalvariability in
the case of the reach scale) and adds information about the
differences in the curvature of individual meanders. Correlation
between sinuosity and radius of curvature recalculated
to channel width is -0.55 (r2
= 0.30).
Radius of curvature divided by channel width ratio
(Rc) is generally understood as an important control variable
determining lateral migration rates [49–51]. Along the
study reach of the Morava River, a negative correlation between
the radius of curvatureof selected meanders and the
measures of channel migration rate (Er and Ar) was found
(r = -0.47; r2
= 0.22 for Er and r = -0.43; r2
= 0.18 for Ar).
Thus we can conclude that Rc is among important factors
in the Morava River. Meander M6 serves as an example.
Substantially accelerated lateral erosion occurred as the
Rc dropped from 2.20 to 1.54. The natural neck cut off of
meander M6 that took place in 2006 shortened the study
reach by 827 m. Local sinuosity was preserved at relatively
high values (decreasing from 2.9 to 2.1). Hooke [52] tested
Brought to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
a possible explanation for cut offs on the Bollin River. The
case of a natural cut off on the Morava River supports the
statement that cut off occurred as a consequence of high
flow (during the flood with Q50 discharge). As mentioned
above, our findings suggest that the behaviour of the study
river reach resembles a self-organising system and thus the
cut off could also occur as a consequence of increasing sinuosity
that reached critical value. Nonetheless, more cut
offs need to be analysed to convincingly prove our pre-
sumption.
Furthermore, maximum rates of Er were documented
on meanders with Rc of 1.5-2.2 and 2.9-3.2 that correspond
with the values presented in studies on English, Canadian
[53], and American [54] rivers.
We assume that another important geometric variable
controlling erosion rates of meanders is river bed slope. Although
longitudinal profiles for the selected years are not
available, our conclusion comes from indirect evidence
from the artificial channel straightening that was immediately
followed by maximal values of erosion rates. Artificial
channel shortening, apparent on the aerial photo from
1938, caused cut off of six meanders (see Figure 6). The
channel length was reduced by 1.1 km and local sinuosity
(in segment 5) from 3.0 to 1.0. In the Strážnické Pomoraví
region, the upstream propagation of knick points after artificial
cut offs appears to be a primary controlling factor
of spatio-temporal variability in channel migration rates.
This conclusion does not fully correspond to previous findings;
for example [55] addressed, in a sand-bed channel
with cohesive river banks, a change in river bed slope as
a factor subordinate to alterations of hydrological conditions.
A model of channel incision and subsequent widening
after river engineering was advocated by [56]. Rapid
channel widening in the period following artificial cut off
may be most probably attributed to such a channel incision
resulting in bank destabilisation.
The temporal variability of channel migration rates
since 1938 also partly corresponds with the magnitude
(Qmax) and the frequency of floods (n≥Q2). The
highest frequencies were documented in the period
1953-63, when frequency was 1.00 flood/year. Relatively
high frequencies were also documented in 1938-53 (0.87
flood/year), 1963-1973 (0.90 flood/year), and 1973-82 (0.89
flood/year). Minimum frequency occurred between 1982-
93 (0.27 flood/year). The final two periods also documented
low frequencies (0.40 and 0.67respectively). On
the reach scale, n≥Q2 appears to be an important controlling
factor in determining both Er and Ar. That is consistent
with results found e.g. by [21, 35], or [43]. In a detailed
scale of (dynamic) segments and individual meanders,
n≥Q2 does not seem to influence Er and Ar. Anthropogenic
activity seems to play more important role. This
suggests that controlling factors differ with scale. Such
findings were also documented in the works of [6, 46].
The contribution of flow magnitude to the explanation
of the temporal variability of lateral erosion rates could be
found on the reach scale. On the segment and meander
scale, other variables appear to play a more important role.
Our finding stands in contradiction with results from other
rivers, where flow magnitude was a factor of key importance
[e.g. [40, 57–60]. Further, we found that flood magnitude
could belong to the controlling variables influencing
rate of channel width change, as there was a positive
correlation on the meander scale between Qmax and ∆W.
This suggests that, in general, during large floods, in meander
bends erosion surpasses accretion more than during
floods of moderate magnitude. As this finding could not be
proved on other scales, we argue that this controlling variable
is of rather secondary importance.
We also argue that specific conditions occurring in the
study reach before and during floods can play an important
role in the determination of the erosion capability of
the flood. This could be seen in last two studied periods.
During 1993-2003, one flood with a magnitude exceeding
Q100 occurred in the study reach. Three small floods with
a magnitude of Q2 occurred later in the period. The erosion
rate was, nonetheless, slightly higher than during the following
period that experienced six floods, two of them exceeding
Q50 and one with a magnitude slightly below Q50.
Our finding stays in agreement with the works of many authors
[e.g. [61–64] that emphasized the importance of local
and/or temporal specifics on pattern of river bank erosion.
An assessment of the control of the magnitude and
frequency of floods over channel migration rates is, obviously,
influenced by the fact that during the twentieth
century, the value of bankfull discharge substantially increased.
Whereas, even in the beginning of the twentieth
century, overbank flow was reached, on average, once a
year, nowadays, bankfull discharge occurs, on average,
once in five years [26, 29]. The cause of the change was
a progressive increase in the channel capacity - degradation
and widening [36, 37]. Today, bankfull discharges exert
higher shear stress on the wetted perimeter than did the
considerably lower bankfull flows at the beginning of the
twentieth century. When considering twofold widening of
the channel in comparison to pre-engineered conditions,
the mean shear stress induced by full channel discharge
is estimated to rise approximately from 4 to 13.5 N/m2
. Assuming
a higher erosional potential of smaller discharges
during the early periods (1938-63) of channel adjustment
to river regulation, then the magnitude of floods is no
doubt a more important controlling variable of channel
Brought to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
change than was ascribed to by our statistical analysis.
This may be supported by comparing stream power between
1938 and 2012. Although the bankfull discharge
stage multiplied at least 1.2 times, and values of stream
power per unit length of the channel rose from approximately
1380 to 1646 W/m), specific stream power stayed
virtually the same (24.2 W/m2
in 1938 and 24.6 W/m2
in
2012) because of considerable channel widening.
Lithology was determined as an important factor in
controlling erosion rates in many studies [e.g. [8, 65,
66].These authors ascribe key control to river bank properties
as an explanation of the lateral erosion rates. The results
from non-parametric analysis of variance in the study
reach found a role of the material only at a local scale
(recent acceleration of Er and Ar in meandersM1 and M3,
which started to erode aeolian sand facies). Nonetheless,
lithology were not capable of satisfactory explanation of
the spatio-temporal variability of lateral erosion for the
whole reach nor for individual meanders that eroded sand
facies in the past 75 years. A similar conclusion was mentioned
in the study of [21] on the Sacramento River.
Land cover belongs to the group of controlling factors
that influence erosion rates in the study reach of
the Morava River. Statistical testing proved higher erosion
rates in meanders eroding arable land (Er = 2.67 m/year,
SD = 1.45 m/year) in comparison with ones with forest
cover (Er = 1.05 m/year, SD = 0.58 m/year), and, respectively,
meadows and clearings (Er = 1.24 m/year, SD = 0.80
m/year). In the Strážnické Pomoraví region, [36] studied
the influence of land cover on channel migration rates,
and documented, for the time period 1841-2007, erosion
rates that were approximately twice as high for meanders
eroding arable land than those of deciduous forest
and grassland. In this study we proved a statistical difference
between arable land and forest and arable land and
meadows and clearings. No statistical difference was ascertained
between forest and meadows and clearings. Similar
findings concerned in controls of land cover on erosion
rates were described in another studies, for example
[22, 67, 68]. On the other hand, the findings on the Sacramento
River [21] did not ascribe a significant influence on
erosion and accretion rates to the land cover.
Land cover influences the erosion rate mainly by increasing
the shear strength of river banks, which is dependent
on root density. Although this might be the case in
flood loams, trees covering dunes of aeolian sand could
accelerate the erosion by the gravity exertion of unconsolidated
sedimentary bodies. Among the indirect influences
of land cover observed in the study reach is the import of
woody debris into the channel and the formation of wood
accumulations. In turn, woody debris protects bank material
from entrainment, resulting in a deceleration of river
bank erosion. However, wood accumulations can also divert
the flow towards the bank and thus increase the rate
of erosion. Both effects of woody debris were observed in
the study reach. The effect of woody debris only started to
occur in recent years as large wood ceased to be removed
from the channel by river managers. Although the effect
of trees and woody debris is apparent from aerial photos,
it belongs to the secondary control variables. Even a fully
mature poplar forest could not substantially mitigate the
accelerated erosion that occurred as a result of the meander
cut off in spring 2006. In the area of the breached
meander neck, the accelerated erosion took place despite
eroding the floodplain with mature forest, which caused
a substantial input of large woody debris and subsequent
wood accumulation development.
Analyses of the influence of river regulation works suggest
that the most important human-induced factors are
engineering activity affecting (1) flow conditions and (2)
changes in geometric parameters such as river bed slope
and radius of curvature. This is apparent from the patterns
of planform shifts of ten analysed meanders that are situated
at different distances from the segments that were
affected by direct engineering alterations (channelization
of segment 5). Meanders M1-M3, situated at the furthest
distance from the artificial cut off, experienced, immediately
after the alterations, substantially lower rates of erosion
and accretion in comparison with meanders M4-M10.
Channel destabilisation induced by local shortening did
not, apparently, reach such remote distances. Despite this
fact, the rate of erosion of meanders M1-M3 in the years
1938-1963 was substantially higher than in the following
periods. During the first 25 years, the effect of engineering
adjustments and relatively high flood frequencies jointly
influenced erosion rates. Besides the artificial cut off, acceleration
of the erosion was conditioned by the channelizing
of side arms and tributaries and channel clearing of
sediment and large wood [26], which supported the concentration
of run-off in the main channel of the Morava
River and its bed degradation and channel widening. In
the middle part of the reach (meanders M4-M8), the influence
of channel shortening (slope) and other abovementioned
adjustments was the most apparent, thus resulted
in the largest, dynamic lateral movements.
6 Conclusions
Detailed analysis of spatial and temporal patterns of lateral
movement rates (both erosion and accretion) was perBrought
to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
formed on the 5.5 km long reach of the Morava River in the
Strážnické Pomoraví region for a period delineated by the
years 1938 and 2012. Our results demonstrate that bank
erosion and accretion rates differed substantially among
river segments and time periods even within such a short
river reach. Rates of erosion and accretion were controlled
by the simultaneous effect of several factors. Among the
controlling variables, which were identified as key ones,
are: (1) engineering works causing changes in channel geometry
(mainly river bed slope) and channel hydraulics;
(2) frequency and magnitude of floods; and (3) radius of
curvature. Local influence is also ascribed to river bank
lithology and vegetation. The relative importance of the
above-mentioned factors varied with time and scale.
As a key factor, engineering works were identified (especially
the artificial cut off and straightening of segment
5). This action induced a dramatic increase in bank erosion
rates. Approximately 25 years went by before the erosion
rates returned to the values documented before the river
engineering adjustments. This dynamic period was characterised
by bank erosion rates reaching 2.19 (1938-53) and
1.47 (1953-63) m/year. It was possible to detect the reaction
of the channel to the shortening as far as 2,000 m upstream
from the cut off and at least 500 m downstream from the
cut off. The influence of the artificial channel adjustment
was supplemented by the relatively high flood frequency.
Another substantial acceleration of river bank erosion was
induced by the natural neck cut off of one mature meander
during the spring flood in 2006, with a magnitude of Q50.
In the first five years after the event, local values of erosion
rates varied between 5.7 and 7.9 m/year. The spatial
extent of the influence of the natural cut off was more localised.
The influence of other factors (flood magnitude,
land cover, bank material properties, radius of curvature)
was identified as being of rather secondary importance.
Another characteristic feature documented in the
study reach was the different erosion and accretion rates.
In periods with higher dynamics of planform changes
(1938-53, 1953-63, 2003-12), erosion rates substantially exceeded
accretion rates, resulting in channel widening (in
the first 15 years, at a rate as high as 0.75 m/year). Channel
width in the remaining periods remained more or less
stable. In general, a substantial increase in channel capacity
took place in the study reach. The same trend was
also documented by [26, 29], who showed a decrease in
bankfull flow frequency from occurring every year to occurring
once every five years. Besides the shortening of
the channel, other artificial activities such as side arms
and tributaries channelization (transformation from anastomosing
to single-thread meandering pattern), sediment
dredging, and removal of large woody debris contributed
to the increase in river bed and river banks erosion (incision,
widening).
References
[1] Kiss T., Fiala K., Sipos G., Alterations of channel parameters
in response to river regulation works since 1840 on the Lower
Tisza River (Hungary), Geomorphology, 2008, 98, 96-110.
[2] Gregory K.J., The human role in changing river channels, Geomorphology,
2006, 79, 172-191.
[3] Allen J.R.L., A review of the origin and characteristics of recent
alluvial sediments. Sedimentology, 1965, 5, 89-191.
[4] Hansen E., On the formation of meanders as a stability problem.
Reports on Progress in Physics 13, Coastal Engineering Laboratory,
Technical University of Denmark, Lyngby, 1967.
[5] Langbien W.B., Leopold L.B., River Meander - Theory of Minimum
Variance. USGS Professional Paper, 1966, 422-H, H1-H15.
[6] Leopold L.B., Wolman M.G., River meanders. Geol. Soc. Am.
Bull., 1960, 71, 769-794.
[7] Camporeale C., Perona P., Porporato A., Ridolfi L., Hierarchy of
models for meandering rivers and related morphodynamic processes.
Rev. Geophys., 2007, 45, 1, RG1001.
[8] Ikeda S., Parker G., Sawai K., Bend theory of river meanders.
Part 1. Linear development, J. Fluid Mech., 1981, 112, 363-377.
[9] Johannesson H., Parker G., Linear theory of river meanders, In:
Ikeda S., Parker G. (Eds.), River Meandering, Water Res. M.,
1989, 12, 181-214.
[10] Zolezzi G., Seminara G., Downstream and upstream influence
in river meandering. Part 1. General theory and application to
overdeepening, J. Fluid Mech., 2001, 438, 183-211.
[11] Rüther N., Olsen N.R.B., Modelling free-forming meander evolution
in a laboratory channel using three-dimensional computational
fluid dynamics. Geomorphology, 2007, 89, 308-319.
[12] Tal M., Paola C., Effects of vegetation on channel morphodynamics:
results and insights from laboratory experiments. Earth
Surf. Proc. Land., 2010, 35(9), 1014-1028.
[13] Nardi L., Rinaldi M., Solari L., An experimental investigation
on mass failures occurring in a river bank composed of sandy
gravel. Geomorphology, 2012, 163-164, 56-69.
[14] Güneralp I., Abad J.D., Zolezzi G., Hooke J., Advances and
challenges in meandering channels research. Geomorphology,
2012, 163-164, 1-9.
[15] Florsheim J.L., Mount J.F., Chin A., Bank erosion as a desirable
attribute of rivers. BioScience, 2008, 58(6), 519-529.
[16] Camporeale C., Perucca E., Ridolfi L., Gurnell A.M., Modeling the
interactions between river morphodynamics and riparian vegetation,
Rev. Geophys., 2013, 51(3), 379-414.
Brought to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
[17] Greco S.E., Patch change and the lifting mosaic of an endangered
bird’s habitat on a large meandering river. River Res.
Appl., 2013, 29(6), 707-717.
[18] Rapp C.F., Abbe T.B., A framework for delineating channel
migration zones. Ecology Final Draft Publication #03-06-027,
Washington State Department of Ecology, 2003, 52 p.
[19] Piégay H., Darby S.A., Mosselmann E., Surian N., The erodible
corridor concept: applicability and limitations for river management.
River Res. App., 2005, 21, 773-789.
[20] Larsen E.W., Girvetz E.H., Fremier A.K., Landscape level planning
in alluvial riparian floodplain ecosystems: using geomorphic
modelling to avoid conflicts between human infrastructure
and habitat conservation. Landscape Urban Plan., 2007, 79,
338-346.
[21] Michalková M., Piégay H., Kondolf G.M., Greco S.E. Lateral erosion
of the Sacramento River, California (1942-1999), and responses
of channel and floodplain lake to human influences.
Earth Surf. Proc. Land, 2011, 36, 257-272.
[22] Micheli E., Kirchner J., Larsen E., Quantifying the effect of riparian
forest versus agricultural vegetation on river meander migration
rates, Central Sacramento River, California, USA. River
Res. App., 2004, 20, 537-548.
[23] Wallick J.R., Lancaster S.T., Bolte J.P., Determination of bank
erodibility for natural and anthropogenic bank materials using
a model of lateral migration and observed erosion along the
Willamette River, Oregon, USA. River Res. App., 2006, 22, 631-
649.
[24] Kadlec J., Kocurek G., Mohrig D., Shinde D.P., Murari M. K.,
Varma V., et al., Response of fluvial, aeolian and lacustrine systems
to Late Pleistocene to Holocene climate change, Lower
Moravian Basin, Czech Republic. Geomorphology, 2015, 232,
193-208.
[25] Havlíček P., Kučera Z., Vachek M., Přírodní park Strážnické
Pomoraví-Osypané břehy: zkrácení toku Moravy. Zprávy o geologických
výzkumech v roce 2007, 2008, 91-92.
[26] Brázdil R., Máčka Z., Řezníčková L., Soukalová E., Dobrovolný P.,
Grygar T.M., Floods and floodplain changes of the River Morava,
the Strážnické Pomoraví region (Czech Republic) over the past
130 years, Hydrolog. Sci. J., 2011a, 56(7), 1166-1185.
[27] Kadlec J., Grygar T., Světlík I., Ettler V., Mihaljevic M., Diehl J.F. et
al., Morava River floodplain development during the last millennium,
Strážnické Pomoraví, Czech Republic. Holocene, 2009,
19, 499-509.
[28] Grygar T., Sětlík I., Lisá L., Optíková L., Bajer A., Wray D.S. et al.,
Geochemical tools for the stratigraphic correlation of floodplain
deposits of the Morava River in Strážnické Pomoraví, Czech Republic
from the last millennium. Catena, 2010, 80(2), 106-121.
[29] Brázdil R., Řezníčková L., Valášek H., Havlíček M., Dobrovolný
P., Soukalová E., Řehánek T., Skokanová H., Fluctuations of
floods of the River Morava (Czech Republic) in the 1691-2009
period: interactions of natural and anthropogenic factors. Hydrolog.
Sci. J., 2011b, 56, 468-485.
[30] Micheli E.R., Larsen E.W. River channel cut off dynamics, Sacramento
river, California, USA, River Res. App., 2010, 27(3), 328-
344.
[31] Hooke J.M., Redmond C.E., Use of Cartographic Sources for Analyzing
River Channel Change with Examples from Britain. In:
Petts G.E. (Ed.), Historical Change of Large Alluvial Rivers: Western
Europe. John Wiley & Sons, Ltd., Chichester, 1989, 79-93.
[32] Richard G., Quantification and prediction of lateral channel adjustments
downstream from Cochiti Dam, Rio Grande, NM. Dissertation
thesis. Colorado State University. Fort Collins, Colorado,
2001, 244 p.
[33] Shields F.D. Jr., Simon A., Steffen L.J., Reservoir effects on downstream
river channel migration. Environ. Conserv., 2000, 27, 54-
66.
[34] Brice J.C., Air Photo Interpretation of the Form and Behavior of
Alluvial Rivers, Final Report to the U.S. Army Research Office–
Durham, Washington University, St. Louis, 1975, 10 p.
[35] Hooke J., Spatial variability, mechanics and propagation of
chase in an active meandering river. Geomorphology, 2007, 84,
277-296.
[36] Smetana M., DynamikakorytaMoravyvevztahu k příbřežní vegetaci
na základě studia historických map a současných měření:
případová studie ze Strážnického Pomoraví. Geologické
výzkumy na Moravě a ve Slezsku, 2011, 18(2), 58-63.
[37] Stehlík F., Analýza polohových změn koryt řeky Moravy ve
Strážnickém Pomoraví: vliv regulace na dynamiku toku. In:
Roszková A., Vlačiky M., Ivanov M. (Eds.): Sborník příspěvků
ze semináře 14. kvartér 2008.Ústav geologických věd PřF MU.
Brno., 2007, 24-25.
[38] MacDonald T.E., Parker G., Luethe D.P., Inventory and analysis
of stream meander problems in Minnesota. Master Thesis, University
of Minnesota, 1991.
[39] Nanson G.C., Hickin E.J., A statistical analysis of bank erosion
and channel migration in western Canada, Geol. Soc. Am. Bull.,
1986, v97, 497-504.
[40] Giardino J.R., Lee A.A., Rates of Channel Migration on the Brazos
River, Final Report No. 0904830898, Texas Water Development
Board, Texas, 2011.
[41] Lagasse P.F., Spitz W.J., Zevenbergen L.W., Zachmann D.W.,
Handbook for Predicting Stream Meander Migration. Transportation
Research Board, Washington D.C., 2004, 107 p.
[42] Richard G.A., Julien P.Y., Baird D.C., Statistical analysis of lateral
migration of the Rio Grande, New Mexico, Geomorphology,
2005, 71, 139-155.
[43] Brice J.C., Stream Channel Stability Assessment, Federal Highway
Administration Report, 1982, 42 p.
[44] Luchi R., Hooke J.M., Zolezzi G., Bertoldi W., Width variations
and mid-channel bar inception in meanders: River Bollin (UK),
Geomorphology, 2010, 119, 1-8.
[45] Stolum H.H., River meandering as a self-organization process.
Science, 1996, 271, 5256, 1710-1713.
[46] Camporeale C., Perucca E., Porporato A., Ridolfi L., On the longterm
behavior of meandering rivers, Water Resour. Res., 2005,
41, 1-13.
[47] Lazarus E.D., Constantine J.A., Generic theory for channel sinuosity,
PNAS, 2013, 110(21), 8447-8452.
[48] Constantine J.A., Dunne T., Meander cut off and the controls on
the production of oxbow lakes, Geology, 2008, 36(1), 23-26.
[49] Heo J., Duc T.A., Cho H.S., Choi S.U., Characterisation and prediction
of meandering channel migration in the GIS environment:
A case study of the Sabine River in the USA. Environ.
Monti. Assess., 2009, 152, 155-165.
[50] Hickin E., Mean flow structure in meanders of the Squamish
River, British Columbia, Can. J. Earth Sci., 1978, 15, 1833-1849.
[51] Hooke J., River meander behaviour and instability: a framework
for analysis. T. I. Brit. Geogr. 2003, 28(2), 238-253.
Brought to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM
[52] Hooke J., Cutoffs galore!: occurrence and causes of multiple cutoffs
on a meandering river, Geomorphology, 2004, 61, 225-238.
[53] Hooke J.M., Interaction of sediment dynamics and channel morphology.
In: Wang S., Langendoen E.J., Shields F.D. (Eds). Management
of Landscapes Disturbed by Channel Incision. Mississippi,
University of Mississippi, Oxford, 1997.
[54] Hudson P.F., Kesel R.H., Channel migration and meander-bend
curvature in the lower Mississippi River prior to major human
modification, Geology, 2000, 28(6), 531-534.
[55] Darby S.E., Thorne C.R., Modelling the sensitivity of channel
adjustment in destabilized sand-bed rivers. Earth Surf. Proc.
Land., 1996, 21, 1109-1125.
[56] Schumm S.A., Harvey M.D., Watson C.C., Incised channels: morphology,
dynamics, and control. Water Resources Publications,
Littleton, Colorado, 1984, 200 p.
[57] Buer K., Eaves N., Scott R., McMillan J., Enlarged Shasta Dam,
Downstream Geomorphic Changes - Preliminary Report. DWRND
Memorandum Report, California Department of Water Resources,
Sacramento, 1983, 50 p.
[58] Hooke J.M., An analysis of the processes of river bank erosion.
J. Hydrol., 1979, 42, 39-62.
[59] Miller J.R., Barr R., Grow D., Lechler P., Richardson D., Waltman
K. et al., Effects of the 1997 flood on the transport and storage
of sediment and mercury within the Carson River valley, westcentral
Nevada. J. Geol., 1999, 107(3), 313-327.
[60] Carroll R.W.H., Warwick J.J., James A.I., Miller J.R., Modeling erosion
and overbank deposition during extréme flood conditions
on the Carson River, Nevada, J. Hydrol., 2004, 297, 1-21.
[61] Beschta R.L., Conceptual models of sediment transport in streams.
In: Sediment transport in gravel-bedrivers, Thome C.R.,
Bathurst J.C., Hey R.D. (Eds.), John Wiley & Sons, Inc., Chichester,
1987, 387-420.
[62] Luppi L., Rinaldi M., Teruggi L.B., Darby S.E., Nardi L., Monitoring
and numerical modelling of river bank erosion processes:
a case study along the Cecina River (central Italy), Earth Surf.
Proc. Land., 2008, 4, 530-546.
[63] Marren P.M., Smith H.G., Erosion and Channel Changes on Corryong
Creek, NE Victoria Following Drought-Ending Floods: Implications
for Channel and Riparian Vegetation Management.
In: 6th Australian Stream Management Conference, Canberra,
2012, 31-38.
[64] Newson M.D., The geomorphological effectiveness of floods A
contribution stimulated by two recent events in mid-Wales.
Earth Surf. Proc. Land., 1980, 5, 1-16.
[65] Constantine C., Dunne T., Hanson G.J., Examining the physical
meaning of the bank erosion coefficient used in meander migration
modeling. Geomorphology, 2009, 106, 242-252.
[66] Mossa J., Coley D., Planform change rates in rivers with and
without instream and floodplain sand and gravel mining: assessing
instability in the Pascagoula River and tributaries, Mississippi,
Report No. 04HQGR0178, USGS, 2004.
[67] Smith D.G., Effect of vegetation on lateral migration of anastomosed
channel of a glacier melt-water river. Geol. Soc. Am. Bull,
1976, 87, 587-860.
[68] Gyssels G., Poesen J., Bochet E., Li Y., Impact of plant roots on
the resistence of soils to erosion by water: a review. Prog. Phys.
Geog., 2005, 29, 189-217.
Brought to you by | Masarykova Univerzita v Brne
Authenticated
Download Date | 2/23/18 2:47 PM