u.r hydrologie-hydraulique, lyon m. lang

6
The Court of Miracles of Hydrology, 18-20 June, Engref Paris 2008 1 Do monsters resist to rating curves scrutiny ? U.R Hydrologie-Hydraulique, Lyon M. Lang Example of flowing conditions during floods (Ardèche & Volane, 20 Oct. 2001) (picutres by M. Lang, Cemagref Lyon) The Court of Miracles of Hydrology, 18-20 June, Engref Paris 2008 2 Degree of rating curve extrapolation in France Class 0 1 2 3 4 1 8 31 7 53 Size (%) T > 100 yrs 10 < T < 100 yrs 2< T < 10 yrs 1 < T < 2 years T < 1 yr Criteria . . . .. . Class • 60% of the stations have not been gauged beyond the 2 year flood peak • Less than 10% of the stations have not been gauged beyond the 10 year flood • Large basins with smooth floods have been better gauged (measurements of discharge are easier) Return period of the maximum gauged flow few rating curves on extreme discharges

Upload: others

Post on 20-Jun-2022

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: U.R Hydrologie-Hydraulique, Lyon M. Lang

The Court of Miracles of Hydrology, 18-20 June, Engref Paris 20081

Do monsters resist to rating curves scrutiny ?

U.R Hydrologie-Hydraulique, Lyon M. Lang

Example of flowing conditions during floods (Ardèche & Volane, 20 Oct. 2001)

(pic

utr

es b

y

M.

Lan

g, C

ema

gre

f L

yon

)

The Court of Miracles of Hydrology, 18-20 June, Engref Paris 20082

Degree of rating curve extrapolation in France

Class

0

1

2

3

4

1831753Size (%)

T > 100

yrs

10 < T <

100 yrs

2< T < 10

yrs

1 < T < 2

years

T < 1 yrCriteria

......Class

• 60% of the stations have not been gauged beyond the 2 year flood peak

• Less than 10% of the stations have not been gauged beyond the 10 year flood

• Large basins with smooth floods have been better gauged

(measurements of discharge are easier)

Return period of the maximum gauged flow

�few rating curves on extreme discharges

Page 2: U.R Hydrologie-Hydraulique, Lyon M. Lang

The Court of Miracles of Hydrology, 18-20 June, Engref Paris 20083

• Active policy of gauging measurement

main source of information

• Data collection on flood marks

several marks to assess the water surface profile, for model calibration

• Hydraulic extrapolation of the rating curves

no empirical calculation, specific behaviour for overflowing discharge

• Feedback on gauging equipment and gauging methodologies

knowledge of the various limitations

• Exchange groups on new technologies

ADCP, video information

How to improve the discharge assessment on large floods ?

The Court of Miracles of Hydrology, 18-20 June, Engref Paris 20084

Review of rating curves by hydraulic modelling

• Morphological data (main channel, flood plain, bridges, thresholds)

• Hydrometrical data (discharge data series, rating curves, gauging values, flood marks)

Data collection

• Sensitive analysis on the downstream condition

• Determination of the roughness coefficients

� main channel : calibration with gauged values

� flood plain : a priori values from field visit & Ven Te Chow tables

Calibration of the hydraulic model

0

5

10

15

20

25

30

35

0 10 20 30 40 50 60 70

Discharge(m3/s)

K=

1/n q

q-10%

q+10%

Overflow of the main channel

Domains : 1 2 3 4

�Representative Manning

coefficient from domain 2

Page 3: U.R Hydrologie-Hydraulique, Lyon M. Lang

The Court of Miracles of Hydrology, 18-20 June, Engref Paris 20085

Review of rating curves by hydraulic modelling

• Sensitive analysis on the gauging values (H, Q)

(H-5 cm, Q+10%) and (H+5 cm, Q-10%)

Extrapolation of the hydraulic model

0

20

40

60

80

0 0.5 1 1.5 2 2.5

Water depth (m)

Dis

cha

rge

(m3/s

)

Maximal curveMedian curveMinimal curve

Q+10%

QQ-10%

H

H-5 cm H+5 cm

135

140

145

150

155

160

67000 67500 68000 68500 69000 69500 70000 70500

Pk (m)

Lev

el (

m N

GF

)

Q=3420 m3/s (Kmin)

Q=2470 m3/s (Kmed)Q=1730m3/s (Kmax)marks of the 1992 flood

Bed level (m NGF)

New

bridge

Railroad

bridge

Old

bridge

• Validation with past flood levels

The Court of Miracles of Hydrology, 18-20 June, Engref Paris 20086

Review of rating curves by hydraulic modelling

• Small to medium error on gauge discharge : 3-5% to 10%

• Large error on the extrapolated domain of the rating curve

� with good hydraulic conditions : about 30% for the 10 year flood

� without gauged values or with complex hydraulic conditions : 60-100 % !

Main results on 20 stations

• Specific difficulties

� downstream influence, hydraulic jump, submerged bridge

� flowing on large flood plain

� debris flow

• Flood marks

� useful for extreme discharge assessment

� not only at the flood scale

Page 4: U.R Hydrologie-Hydraulique, Lyon M. Lang

The Court of Miracles of Hydrology, 18-20 June, Engref Paris 20087

Data set of 195 long series

1. Choice of hydrological stations

Criteria : long series (> 40 years), good quality

(no influence, no evolution of the bed river, good

rating curves)

Preliminary visit to the data managers

2. Local analysis of changes

Use of various samping variables (from the daily discharge data series)

Checking of the results for each station: discussion with the data managers

Example of erroneous diagnosis for trend detection

National study on flood and droughts (Renard, 2006)

The Court of Miracles of Hydrology, 18-20 June, Engref Paris 20088

First results of a set of 195 long data series(test of deviance, GEV distribution)

�Many significant changes (10% error level) : 1/4 for floods, 1/3 for low flows

� About the same number of increasings and decreasings

� Poor spatial coherency

Impact of climate change ?

NFlood (Maxan) Low Flow (7 day Discharge)

Example of erroneous diagnosis for trend detection

Page 5: U.R Hydrologie-Hydraulique, Lyon M. Lang

The Court of Miracles of Hydrology, 18-20 June, Engref Paris 20089

Review of the results

On the whole set

of 195 stations….

Flood Low Flow

New data set124 stations for rainfall-runoff floods

25 stations for snow melting floods

90 stations for low flows

Example of erroneous diagnosis for trend detection

Suspected explanation

of change

Metrological issue No explanation

Influence No information

Detected change

No change

The Court of Miracles of Hydrology, 18-20 June, Engref Paris 200810

New results on rainfall-runoff floods (reduced data set of 124 stations)

0 2 3N=

A few detected changes (10% error level)

About the same number of increasings and decreasings

Poor spatial coherency

Page 6: U.R Hydrologie-Hydraulique, Lyon M. Lang

The Court of Miracles of Hydrology, 18-20 June, Engref Paris 200811

The number of detected changes is not significant

227Mean discharge3686Mean

discharge

722Date of MINAN

1222MINAN

3658Low flows

2833Date of QCX10

833Threshold

discharge QCX104115

Snow melting

floods

318Date of MAXAN

1219MAXAN

3683Floods

Observed

Percentage

Critical

percentage N0.05

VariableNb of

common

years

Nb of

stations

Regional significance of local tests Local and national risk: 5%

The Court of Miracles of Hydrology, 18-20 June, Engref Paris 200812

How to deal with discharge uncertainty ?

1. Reduce uncertainty� Promote investment on data measurement (gaugings, flood marks, good rating curves)

� Exchange information for the setup of reviewed data sets

2. Consider uncertainty� Include discharge uncertainty within statistical analysis, model calibration …

3. Use rainfall information� Extreme flood is mainly due to extreme rainfall …