Extrait de : REGULATED RIVERS: RESEARCH &
MANAGEMENT Regul. Rivers: Res. Mgmt. 14:
13-23
CHANNELIZATION AND CONSEQUENCES ON FLOODPLAIN
SYSTEM FUNCTIONING ON THE GARONNE RIVER, SW FRANCE
J. STEIGER & F. GAZELLE
ABSTRACT
This study focuses on changes in flow dynamics and morphology
of the river bed in altering floodplain system functioning. Even
though direct training works on the Garonne River channel on the
upstream reach are moderate, human effects on the river channel
have irreversible effects. A few aspects of the complex
relationship
between
the
contemporary
hydrological
and
geomorphological river channel changes and the adjacent
floodplain system are examined to provide further understanding
of the whole fluvial hydrosystem, its habitats and its
ecological functioning.
The decrease in bedload material resulting from dam construction
and industrial gravel extraction after 1960 caused accelerated
channel incision in the Garonne River. Despite an increase in
channel capacity, bank-full discharge related to a recurrence
interval of 1.58 years, as well as low flow discharges, have a
tendency to diminish since the beginning of the century. The
increase in the channel cross-section reduces overbank flows,
especially those generated by high frequency floods. Thus, the
regularly flooded areas diminished, causing a decrease in the
interaction time between the floodwave and the floodplain.
Furthermore, channelization processes have altered floodplain
construction processes. Riparian wood die-back during the last
10-15 years is one of the consequences of these hydrological
and geomorphological changes of the fluvial ecosystem. © 1998
John Wiley & Sons, Ltd.
KEY WORDS:
Garonne River; human effects; channel incision; bank-full discharge; runoff
fluctuations; flood dynamics; flood-plain; riparian wood die-back
INTRODUCTION
The fluvial dynamics of river systems, i.e. the interrelationship
between flow regime, sediment load transport and channel response,
have been recognized for a long time (Fargue, 1868, Leopold and
Maddock, 1953; Schumm, 1969). In particular, the role of bank-full or
dominant discharge, as well as of single flood events of high magnitude
and low frequency, in channel adjustment is well known (Tricart, 1960;
Wolman and Miller, 1960; Beven and Carling, 1989; Gurnell' and Petts,
1995) and has to be considered when studying channelization processes
and their consequence on floodplain system functioning.
Most river systems in Western Europe have endured direct or indirect
human effects (Petts, 1984). Therefore, the strong interference of
human activities plays an important role in channel changes in these
fluvial systems, and case studies show the physical channel adjustments
of river channel control variables to human-induced changes (e.g.
Babinski, 1992; Darby and Thorne, 1992; Thorns and Walker, 1992; Sear,
1995; Ibanez et al., 1996).
Nevertheless, habitat characteristics and habitat quality in fluvial
hydrosystems do not depend only on channel morphology. The floodplain
also controls the establishment and maintenance of aquatic and
intermediate habitats (Schiemer et al., 1995). Furthermore, far fewer
studies have been devoted to the biotic functioning of the aquaticterrestrial transition zones (e.g. Pinay et al., 1995; Ward and
Stanford, 1995). Therefore, the aim of this paper is to focus on recent
hydrological and geomorphological effects, as well as on closely
related ecological effects resulting from human activity, affecting a
river reach of the upper part of a 525 km long river, the Garonne.
CCC 0886-9375/98/010013-11S17.50
© 1998 John Wiley & Sons, Ltd.
14
J. ET A
CHARACTERISTICS OF THE UPPER GARONNE RIVER
The Upper Garonne River (SW France) covers a drainage basin of-32350- km2
upstream of the confluence with the Tarn River (Figure 1). The studied
river reach, located between Toulouse and the Tarn tributary, is
characterized by a mean channel width of 150 m and a mean coefficient of
sinuosity of 1.3.
In the Upper Garonne drainage basin, the Pyrenees mountains and the
intensive cultivated piedmont can be distinguished as the main sediment
production areas. Downstream of Toulouse, the longitudinal gradient is
lower than 0.001 and the Holocene floodplain widens up to 2-4 km. However,
the lowest and most frequently inundated floodplain, where riparian
vegetation still persists, does not exceed a width of 350 m.
The mean abundance at Toulouse between 1910 and 1993 was 194 m3 s ~ '
(19.4 1 s ~ ' km~2). Floods with highest peak discharges, as well as more
than 30% of all floods that attain at least 2 m at Toulouse, the first
alert level, occurred during the months of May and June, when
precipitation is high.
CHANGES IN RUNOFF; BEDLOAD DISCHARGE AND CHANNEL MORPHOLOGY
Runoff fluctuations
During the 19th century more natural catastrophes, such as
avalanches and flood events, were generated in the Pyrenees, than
during this century (Metailie, 1991; Steiger, 1990, 1991). However,
the main causes of the appearance of these high magnitude events during
the last century are still the subject of controversy (Antoine et al.,
1990).
Either
higher
precipitation
or
deforestation
and
overexploitation of
A
20
40km
studied river reach
Atlantic Ocean
Figure 1. The Garonne River, South-west France and the study area
1998 John Wiley & Sons, Ltd.
Regul. Rivers: Res. Mgmt. 14: 13-23 (1998)
CHANNELIZATION ON THE GARONNE RIVER
15
year
Figure 2. Low water flow (230 = the 30th lowest discharge of each year) is determined by
atmospheric conditions, but influences of increased water intake for irrigation purposes
are incontestable. The moving averages of 5 years show a cyclic trend. The overall mean
is 61 m3 s ~' (a). The hydrological deficit shows as well as the low flow, a cyclic
trend (moving average of 5 years). Even though the deficit was not the highest in 1989
(b), lowest discharges were observed on the Garonne River (a), confirming the
modification of the hydrologial cycle by human interference
the Pyrenean landscape at this period are blamed. Probably, a
combination of these two factors has to be considered.
During the second half of the 20th century, human effects on the flow
regime
have
increased,
with
the
construction
of
dams
and
intensification of irrigation practices. Irrigation of water-indigent
cultures such as corn increased significantly after the 1960s. In
Figure 2a the 30th lowest discharge (Q30) for each year is represented
for the period 1913-1993. The general cyclic trend of higher and lower
discharges has to be considered as natural (Probst, 1989). However, low
flow discharges show a tendency to decrease during this century.
Low discharges on the Garonne River are generally related to low
precipitation and therefore to a decrease in water storage in
aquifers, and a decrease in snow and ice storage in the high
Pyrennes mountains (Lambert et al., 1989). Atmospheric conditions have
caused several low flow periods in the Upper Garonne Basin (Figure 2a).
However, despite a more severe period of climatic dryness during the
years 1942-1949 than for 1983-1990 (Figure 2b), the lowest discharges
were observed to occur in 1986, especially in 1989. These low
discharges can be explained by the increase in water intake in the
Garonne River and its aquifer, especially during the irrigation season,
which coincides with natural low water flow. Indeed, water intake and
water deviation through canals for irrigation in the Upper Garonne
River basin has increased considerably compared with the 1940s and is
much more important than that for domestic needs or industry.
Furthermore, Galibert (1956) observed that the glacial activity
decreased rapidly in the Pyrenees mountains. Thus, the part of the
melted ice that could sustain low flow discharges during the summer
months also decreased.
Bank-full discharge
Bank-full discharge, which can be associated with the dominant
discharge concept, plays a decisive role in fluvial morphology. It is
considered as an important parameter controlling channel and
floodplain morphology (e.g. Wolman and Miller, 1960). Dury (1981) and
other authors found a mean return period
1998 John Wiley & Sons, Ltd.
16
of 1.58 years for this specific discharge. Even though it is well
accepted now that the river bed is a product of a range of discharges
rather than of only one discharge (Biedenharn and Thorne, 1994), and
despite some disagreement about this number (e.g. Williams, 1978), it
is still used as an indicative value.
In this study, the theoretical discharge with a recurrence interval of
1.58 years was calculated for three time periods of 30 years (Figure
3). The discharge associated with a recurrence interval of 1.58 years
decreased from the beginning of the century until the present day. The
lowest value, of 1200 m3 s~', was obtained for the period 1964-1993.
This means that the calculated bank-full discharge decreased by about
30%, from 1500 m3 s-1 at the beginning of the century to 1200 m3 s ~ '
nowadays. The recurrence interval of the bank-full discharge was
calculated from an annual maximum series. Therefore, the decrease in
the calculated bank-full discharge translates to a decrease in flood
magnitudes during this century. This decrease cannot be related to
the construction of river dams nor water intake, which do not
influence floodwave generation significantly.
Bedload decrease and channel adjustment
According to several reports, the channel bed of the Upper Garonne
River has never been completely covered by a coherent layer of gravels
and cobbles during the 20th century. Harle (1895), Serret (1900) and
Denizot (1953) observed, respectively, a few years before the
beginning of industrial extraction of gravels in the river bed, that
the bedrock appeared in the river channel.
The first locally introduced artificial bedload retention artifices at
altitudes from about 1500 m to 2600 m were constructed in 1870 and in
the 1920s and 1930s on natural tarns formed in Pleistocene cirques. To
our knowledge no studies have been published to estimate the
alteration of the bedload transport of the Upper Garonne River by
these first constructions on the river network in the Pyrennes
mountains. The major hydroelectric power plants on the Garonne River
between the Pyrenees and Toulouse have only been constructed since
1960. This cascade of dams stopped bedload supply to downstream
reaches.
Intensive industrial gravel extraction in the main channel
downstream of Toulouse started in the mid-1960s. Thus, on the one
hand bedload was extracted artificially and on the other hand
bedload transfer from the upstream to the downstream sections was
interrupted by the construction of dams. Today, the only bedload
source may consist of bank failure. Indeed, field observations after
several floods
time periods
Figure 3. Discharges with a recurrence interval of 1.58 years (theoretical bank-full
discharge) of the Garonne River at Toulouse were calculated for
three time-periods of 30 years during this century
1998 John Wiley & Sons, Ltd.
CHANNELIZATION ON THE GARONNE RIVER
17
Figure 4. Relative incision rates of the Garonne River in the studied river reach
(after Beaudelin, 1989; Steiger and Gazelle, 1994)
in the first half of the 1990s revealed local bank erosion of nonprotected river banks. However, even though the Garonne River has
never been canalized in this section, at least 30% of the river
banks between Toulouse and the Tarn River, especially concave banks,
were stabilized after the high magnitude flood of February 1952 and
cannot be eroded (SMEPAG, 1989). As a consequence, net channel depth
increase during the last 30-35 years has been observed. Channel
incision rates of 68 mm y r ~ ' (1959-1980) and 52 mm yr"1 (1970-1984)
were estimated by Beaudelin (1989). Steiger and Gazelle (1994)
observed the same tendency of the river to channel incision by
comparison of flood gauging heights for floods with the same magnitude
occurring in the 1960s and 1990s (Figure 4).
Observed channel adjustments of the Garonne River to discharge and
bedload decrease do not correspond exactly to the adjustments
proposed by the conceptual model of channel adjustments (e.g.
Knighton, 1984). Especially since 1958, the stabilization of concave
river banks has stopped the natural meandering, and, consequently,
sinuosity and meander wavelength have not changed significantly. As
already described above, the interactive processes of recent
channelization of the river reach and bedload decrease as a
consequence of human effects during the last decades caused an
increase in stream bed erosion and local channel widening occurred at
non-protected and over-steepened river banks.