1. (with ) by clicking on it and then
Recalculation of historical flood flows recorded on hydrometric stations and impact assessment
Arnaud Belleville , Amélie Tacnet, Damien Sevrez, Alexandre Hauet, Thibault Mathevet
EDF-DTG, Grenoble, France
PICO Navigation
The orange ‘Home’
button takes you to
the section menu
The left/right arrows
take you to the
previous/next page
Enlarge some pictures
(with ) by clicking on it and then
close the window
[close]

2. Calculation of flow series Extrapolation and uncertainties
2 MINUTE- MADNESS - OVERVIEW
Context
Hydrometric data are essential for many hydrological issues such as flood forecasting and warning, engineering design and policy decisions related to water resource management as well as calibration of hydrological models.
Calculation of flow series
Hydrometric stations do not measure directly river discharge. Streamflow time series are most often computed through use of a stage-discharge rating curve (RC) that relates measured river stage to streamflow discharge.
Q (m3/s)
H (m)
Extrapolation and uncertainties
Gauging flood discharge can be very tricky.
rating curves must be extrapolated for high flows, which brings a lot of uncertainties.
The flood discharge series can be unreliable, impacting significantly flood forecasting, or the sizing of structure like dam spillway.
Aim of the study
Use topographic survey and hydraulic modelling to extrapolate RC
Recalculate peak flow series on a sample of hydrometric stations, so as to propose homogeneous and less biased flood data
Assess the deviation between modeled and historical peak flows.
Water level of historical flood ?
Gaugings
Operational RC
Modeled RC

3. Extrapolation of rating curves Recalculation of historical peak flows
Hydrometric data are essential for many hydrological issues such as flood forecasting and warning, engineering design and policy decisions related to water resource management as well as calibration of hydrological models.
1/5
Hydraulic modelling
Recalculation of historical peak flows
Results
HS – EGU – General Assembly, Vienna – 11 april 2018

4. Extrapolation of rating curves Recalculation of historical peak flows
Hydrometric stations do not measure directly river discharge. Streamflow time series are most often computed through use of a stage-discharge rating curve that relates measured river stage to streamflow discharge. The rating curve is a hydraulic model, which is usually developed using discrete measurements of stage and discharge (so called gaugings) as calibration data.
2/5
Hydraulic modelling
Recalculation of historical peak flows
Results
HS – EGU – General Assembly, Vienna – 11 april 2018

5. Extrapolation of rating curves Recalculation of historical peak flows
The river bed can change over time and lead to non-stationary stage-discharge relationships depending on the geological nature of the catchement , on seasonal vegetation growth, or an hydro-meteorological events.
3/5
Hydraulic modelling
Recalculation of historical peak flows
Results
HS – EGU – General Assembly, Vienna – 11 april 2018

6. Extrapolation of rating curves Recalculation of historical peak flows
Gauging flood discharge can be very tricky, especially for mountainous river. As a consequence, the rating curves must be extrapolated for high flows, which
brings a lot of uncertainties.
The flood discharge series can be unreliable, impacting significantly flood forecasting, or the sizing of structure like dam spillway.
4/5
Hydraulic modelling
Recalculation of historical peak flows
Results
HS – EGU – General Assembly, Vienna – 11 april 2018

7. Extrapolation of rating curves Recalculation of historical peak flows
After a description of extrapolation practices of the rating curves, the contribution of topographic data and hydraulic modelling for the extrapolation of RC is detailed. We present a methodology for recalculating peak flow series, so as to propose homogeneous and less biased flood data. This methodology is applied to a sample of hydrometric data, followed by an impact assessment on flood data.
5/5
Hydraulic modelling
Recalculation of historical peak flows
Results
HS – EGU – General Assembly, Vienna – 11 april 2018

8. EXTRAPOLATION OF RATING CURVES
Strong assumption for the extrapolation: existence of a timestable stage-discharge relationship beyond a certain discharge value
By examining historical rating curves of some hydrometric stations, heterogeneity is often observed in the extrapolations for high flow. This heterogeneity contradicts the assumption of the convergence of the stage-discharge relationship. ?
Hmax varies with time
For a same H, Q may vary from one period to another
Heterogeneity can be explained by the input of new informations over time, that have not necesserily been valorized over the past period: flood gaugings, topographic survey, hydraulic modelling,
With the aim of producing homogeneous and less biased streamflow series, a work of reanalysis has been done, by using hydraulic modelling
HS – EGU – General Assembly, Vienna – 11 april 2018

9. Calibration of the model Extrapolation with the model
HYDRAULIC MODELLING
For 170 of its hydrometric stations, EDF DTG has realized topographic surveys and 1D hydraulic models. The model MASCARET is used for high flow conditions, to understand the hydraulics of flood. 1 2
Topographic survey
Calibration of the model
Location of the transects
Bathymetry of the river and topography of the flood plain
(POH2) 3
Extrapolation with the model
Hydrometric station of Trebons-Bas (France)
HS – EGU – General Assembly, Vienna – 11 april 2018

10. RECALCULATION OF PEAK FLOWS
Method
1/ Recalculation of peak flows is made from the water level series available in the database. The homogeneity of the water level series is analyzed in order to detect breaks in the mode of production of the series: staff gauge moving, drastic change of the hydraulic control…
A homogeneous period is then identified to carry out the recalculation.
2/ The floods are sampled from the water level series, at the rate of 2 floods per year on average. The selection is made from the water level series, and not from the discharge series, because of the heterogeneity of the flow series induced by the heterogeneity of the extrapolation of the rating curves.
3/ Peak flows are finally recalculated using two rating curves and then compared to the historical peak flows stored in the database:
The rating curve used currently by the field hydrologist. This rating curve valorizes the maximum of available hydrometric information and the most complete expertise from the field hydrologist. The recalculated peak flows are called homogenized peak flows.
The rating curve resulting from the hydraulic modelling. The recalculated peak flows are called modeled peak flows.
HS – EGU – General Assembly, Vienna – 11 april 2018

11. RESULTS (1/2) Illustration on the hydrometric station of Trebons-Bas
Water level series available since 1983
No major changes in the hydraulic control or in the datum of the staff gauge
70 floods selected
Relation between modeled and historical peak flows
Distribution of peak flow deviations between modeled and historical discharges
4 major floods
HS – EGU – General Assembly, Vienna – 11 april 2018

12. Results on a sample of hydrometric stations
Sample: 85 hydrometric stations (with a validated hydraulic modelling study)
HS – EGU – General Assembly, Vienna – 11 april 2018

13. [close] Mean lifetime of the rating curve
The map represents the location of hydrometric stations operated by EDF – Electricité de France. The stations are mainly located on the French mountain ranges (Alps, Pyrenees and Massif Central), at high altitude, on small rivers with important slopes. The variability of the rating curves, which reflects the stability of the hydraulic controls, differs from one massif to another.
Thus, in the Alps, where the erosion and sedimentation processes are more significant, the frequency of updating the curves is higher than in other massifs.

14. [close] Water level (m) Gaugings used for the calibration
H simulated by the model
Discharge (m3/s)

15. [close]

16. [close]

17. [close]
The global bias is very limited, with a mean value of -2.2% (hydraulic modelling leads to a slight decrease of peak flows)
The median deviation is less than 10% in 70% of the cases, but the deviation and the variance may be considerable for some stations, reaching 60%

18. [close]
Histogram computed on 4074 floods sampled on the 85 hydrometric stations
Overestimations from hydraulic modelling are mainly due to overflows on floodplain, increasing the wetted areas
Underestimations fom hydraulic modelling are mainly due to the activation of downstream hydraulic controls for high flows, like bridges or dams

19. Spatial distribution of median deviations between modeled and historical peak flows
[close]
No regional trend can be particularly highlighted. Each analysis depends on local conditions, related to the specificity of the hydraulic control, and the management of the rating curve.