Geography Reference
In-Depth Information
in significant casualties and economic impacts. The flood
on September 8-9, 2002 in the Gard River, for example,
killed 24 people and caused damages estimated at 1.2 bil-
lion
Nationale du Rhone (CNR) gauging station. The reach
at Sauze-St Martin has a catchment area of 2240 km
2
and a mean annual discharge of 63m
3
/s. The instan-
taneous discharges (Q
T
) for return period T years are:
Q
2
=
(Delrieu et al., 2005). The understanding of flood
generation and propagation processes requires reliable
discharge estimates throughout the river network, in real
time. The standard procedure is to use current meter and
acoustic velocimeters to obtain point measurements of
velocity across the cross-section and calculate discharge
using the velocity-area method. Assuming that the chan-
nel bathymetry is known, point estimates of discharge
are calculated by multiplying the point velocities by the
area of the channel cross-section for which they are rep-
resentative. The sum of the point estimates of discharge
provides an estimate of the total discharge for the chan-
nel. The water level or stage is recorded and discharge is
estimated at various stages to construct a stage-discharge
rating curve.
A number of problems make this standard procedure
difficult or impossible to apply to large or flashy floods:
i) high flow velocities and floating debris endanger the
operators and the equipment, ii) the river level and
discharge can vary during the measurement period, com-
promising the quality of the discharge estimate, iii) the
standard procedure is time consuming, labour intensive,
and contributes to the high cost of river monitoring,
iv) the accuracy of results at high discharges is reduced
due to poor quality or missing data and may neces-
sitate the extrapolation of rating curves beyond their
verified ranges, and v) a developed rating curve may
change in time due to geomorphologic changes or to
modification of the hydraulic control downstream. More
continuous and real-time monitoring of river discharge
is desirable.
¤
3390m
3
/s.
The stage-discharge rating curve used by CNR at Sauze-St
Martin is well documented, with 39 direct measurements
of discharge between December 2003 and February
2009 at flow rates up to 2700m
3
/s with a stable stream
morphology. Natural tracers such as bubbles, surface
ripples and vegetal debris were sufficient for the LSPIV
measurements. Tests were conducted on November
22-23 2007 at discharges that reached a maximum of
760m
3
/s. To verify the method, discharge was calculated
from simultaneous Acoustic Doppler Current Profiler
(ADCP) measurements for two discharges at approx-
imately 330m
3
/s. ADCP measurements were acquired
with a Teledyne RDI RioGrande 600 kHz instrument.
In this study, a video camera was set on a mobile
lightweight telescopic mast for which the height could be
varied from 2 to 10m. A relatively cheap commercially-
available digital video camera (Canon MV750i) with
sufficient resolution and frame rates was selected for this
application. Video was recorded in mini-DV format at
a rate of 25 frames per second with an image resolution
of 720x576 pixels. Estimates of mean velocity were made
using the LSPIV technique from a 2 minute series of
images with a time interval (
1830m
3
/s, Q
5
=
2770m
3
/s, and Q
10
=
t) of 0.2 s for a total of 600
images. Mapping coefficients were calculated using image
ortho-rectification (Equation 16.1), with 10GroundCon-
trol Points (GCP) measured along both sides of the river
using a total station and white square targets that were
visible in the image.
Δ
16.3.3 Imageprocessing
16.3.2 Fieldsiteandapparatus
The LSPIV algorithms for estimating velocities are the
same as those used in conventional PIV (Adrian, 1991,
Fujita et al., 1998, Bradley et al., 2002). They are based
on the same concept as human vision in which pattern-
recognition is used to follow the displacements of water
surface tracers in consecutive images. An index of similar-
ity is calculated between a small InterrogationArea (IA) in
the first image and multiple IAs of the same size within a
larger SearchArea (SA) in the second image. Avector from
the IA in the first image to the IA in the second image with
the highest similarity index is assumed to be the displace-
ment vector of the water surface over the time interval
(
Δ
t) between the capture of the two images. Velocity is
calculated by dividing the displacement distance by
This section reports on tests that were conducted with a
mobile Large-Scale Particle Image Velocimetry (LSPIV)
system. The mobile LSPIV system has been developed for
preliminary tests and flood-triggered measurements at
gauging stations that are not equipped with a fixed LSPIV
system (Kim et al., 2008). The objectives of these experi-
ments were to improve the monitoring of Mediterranean
flash-floods throughout the Ardeche River catchment,
to assess the accuracy and reliability of LSPIV discharge
estimation, and to improve the extrapolation of rating
curves to high ungauged discharge values.
The field tests were conducted at the outlet of the
Ardeche River at
Δ
the Sauze-St Martin Compagnie
t.
Search WWH ::
Custom Search