Hydrological Models for Flood Forecast · P. Givone, Cemagref - Direction Scientifique PREVISION DES CRUES 10 TENTATIVE S/T LOCKS AND CRITICAL COMPONENTS IDENTIFICATION For the headwater

1P. Givone, Cemagref - Direction Scientifique PREVISION DES CRUES

Hydrological Models for Flood Forecast


OverviewThis presentation deals with "flood forecasting" and

"hydrological modelling" with a specific angle

Hydrological modelling for flood forecasting : state of art, locks and challenges, regarding the whole forecasting system. Which gain for the whole system is expected from (hydrological) modelling ? Who is in charge of the global picture ?


THE FLOOD FORECASTING SYSTEM

• Météorological forecasting• Rainfall forecasting, nowcasting• Quantification and spatialization of rainfalls• Discharges/water stages forecasting• Flood plains forecasting• Message and delivery service/system• Evaluation of the system


FLOOD FORECASTING/WARNING SYSTEMS ARE NOT GIVEN RESEARCH OBJECTS

It probably makes no sense to study any hydrological or meteorological components (for real time purposes) out of the framework of the global organization of the forecasting/warning system itself,

How to go beyond the limits of hydrological and meteorological products recycling ?

It’s not about usual “academic/applied sciences” discussion, but about systemic approach.


The MINIMAX MODEL (1)There is a real danger in only minimizing the

“maximum” (extreme floods) without really improving the global system and the medium or moderate floods. However, and because of the media coverage, this is the actual trend,

To find a good compromise between Prevention and Prevision is a real challenge,

Flood forecasting and warning have to be incorporated within the global flood management scheme.


MINIMAX MODEL (2)

e

0

50

100

150

200

250

300

3501971

1976

1981

1986

1991

Années

Q (m3/s)

Intermediate Range

Structural Methods Range

Crisis management Range

DAILY FLOWS

Extreme events


FLOOD/WARNING SYSTEMS ARE USERS ORIENTED (1)

The USERS are both :

For the scientists, the users are technical Services and institutional bodies : They operate the networks, our models, … (all technical systems), and design and deliver appropriate messages to

End users (ie individuals, citizens, …, nothing to do with technical people) who are the recipients of the messages (dissemination response) and whose comfort, quality of life ,and sometime just life, depend of them.


FLOOD/WARNING SYSTEMS ARE USERS ORIENTED (2)

If the research and technical object to consider is the global warning system, who is in charge of the this global picture ?

If we consider only the hydrological/meteorological part of the system, what is the maximum possible benefit for the global system, due to improvements of these “hydromet” components, and who is in charge of the global evaluation, to draw future priorities for research and transfer ?

Hydromet components : Robustness or accuracy ?


Forecasts at downstream site derived from :

Observations/measurements of water stages at downstream and upstream sites

Measurements of (catchments) and upstream reach flows

Observation of rainfall (snow melting)

Forecasts of rainfall

Nowcasting and forecasts from other basins

flood forecasts, lead time and time to peak


TENTATIVE S/T LOCKS AND CRITICAL COMPONENTS IDENTIFICATION

For the headwater catchments, the nowcasting, the rainfall observations and forecasts, the snow melting modelling are, at least, critical components,For the downstream forecasting sites, flow routing in main channels (including upstream reaches) are critical components,At basin scale, the improvement of runoff production(robust across scale) and water transfers controlled by topography and soil is important. All spatialized relevant information (geology, terrain, land use and land cover -DOT) is also important, and the incorporation of embedded meteorological information at different scales as well.


RAINFALL OBSERVATION AND MEASUREMENTS


RAINFALL OBSERVATION AND MEASUREMENTS

It’s now obvious that hydrological radar will be soon the most appropriate device in rainfall measurements, in cooperation with the ground systems,

The capability to feed hydrological models with quantified spatially consistent rainfall is critical, even for global rainfall-runoff models, and not only for distributed physically based models (robustness and uncertainty ?)

That doesn’t mean that only distributed physically based models are relevant modelling components


0.0

0.4

0.8

1.2

02 - DEC 03 - DEC01 - DEC

Rai

nfal

l ( m

m )

RAINFALL KNOWLEDGE

0

2468

1012

02 - DEC 03 - DEC01 - DEC

Flow

( m

3s-

1)

Observed

Forecast (known rainfall)

Forecast (rainfall stopped)

Forecast (same rainfall)


abcourbes des dˇbits prˇvus � 2 heurestraces des prˇvisions de dˇbits de 1 � 9 heures

WITH AND WITHOUT RAINFALL FORECASTSON A SMALL BASIN

J. Lavabre and Al



Hydrological Model Choice?

Thanks to Bob Moore for the picture


LE MODELE HYDROLOGIQUE GR3H

figure 4 – Modèle GR3H

A

P

S = k.A

Ps = (1-k²).P Pr = k².P

Q = QR + Qp

0.1 Pr

C

3/CHU1

2C

3/2C

HU2

B RQp

QR = RDEB-RFIN

0.9 Pr

444111BRR DEBFIN

+=

C

lnA=5.3 fixé : lnB=5.43 ; lnC=-1.15 ; Nash=94

lnA=5.3 et lnB=4.4 fixés : lnC=1.01 ; Nash=83

a

bfigure 5 – Perte en calage (crue juin 2000).


RADAR NETWORK, NEW CONCEPTS

New (X-band) Radar are now available, with doppler and multi-polarization facilities


HYDROMET COUPLING ACROSS SCALE

24 hours precipitation forecastcomputed by ALADIN model


Hypsometric curve

Elevation zones

Observed flowModelled flow

PACK Snowmelt ModelThanks to Bob Moore

Model simulation


ROUTING AND INUNDATION MODELLING


ROUTING and INUNDATION MODELLINGIn main rivers, the channel flow routing is critical to

improve the forecasts. One robust possibility is to use Kinematic wave based models (with lateral flows) as the CEH KW model for example

cuxqct

q =∂∂+∂

∂ 1 4 n-1320

21 43 n-1 n

2tu

The difficulty is in using the results of these models to draw the flood plains according to the inundation modelling needs. The hydrodynamics based models are more efficient, but less robust and, to a certain extend, data driven


The 1D hydrodynamics models are robust enough to be used as a component incorporated in operational systems. The main questions are about the needs in geometric data to feed the models, and the operational staff skills in operating them,

The flood plains mapping is much more easy from these models if, and only if, only one geometric database is used,

The geometric characteristics of the basins are more and more needed in hydrological modelling, generally speaking, the use of DOT is one way of improvements.

ROUTING and INUNDATION MODELLING


1D HYDRAULICS/GEOMATICS MODELLING


Flood simulation maps

300 m3/s 500 m3/s 900 m3/s

Flooded area




0

100

200

300

400

500

600

700

800

0 1000 2000 3000 4000 5000 6000

Dis charge [m3.s -1]

Wat

er le

vel [

cm]

hydrometric measurementra ting curve before floodrating curve after flood

DISCHARGE MEASUREMENT IS STILL A CHALLENGE


RUNOFF PRODUCTION


Improved process representation relevant to scales beyond the point : hillslope, grid, catchment, …, using with a better “yield” the available DTM and DEM,Water tracking on slopes, runoff-production is dominated by sub-horizontal water transfers controlled by topography and soil,Better use of spatial dataset support : terrain, soil, geology, land use, land cover, weather variables, …- Improved model transfer to ungauged catchments across

scales- Whole catchment models, linking rainfall-runoff and

hydrodynamic river models- Atmospheric/hydrological model coupling

RUNOFF PRODUCTION MODELLING IMPROVEMENTSvs ROUTING IMPROVEMENTS


Many relevant parameters are derived from spatially consistent information


Soil humidity expressedby API index on 6 August

Soil humidity expressedby API index on 11 August

Soil moisture (API index) on August 6th

Soil moisture (API index) on August 11th


0

20

40

60

80

100

120

140

160

180

200

1.8.2

002

2.8.2

002

3.8.2

002

4.8.2

002

5.8.2

002

6.8.2

002

7.8.2

002

8.8.2

002

9.8.2

002

10.8.

2002

11.8.

2002

12.8.

2002

13.8.

2002

14.8.

2002

15.8.

2002

16.8.

2002

17.8.

2002

18.8.

2002

19.8.

2002

dai

ly t

ota

l in

mm

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

max

.dis

char

ge

m3/

s

Praha

Ús tí

Praha

Ús tí n.L

80 m m

110 m m

SPACE AVERAGE of DAILY PRECIPITATION on BASIN and FLOW PEAK DOWNSTREAM


THE GROUND NETWORKS ARE CRITICAL COMPONENTS, and the RESEARCH BASINS,

DEVOTED TO FLOOD FORECASTING AS WELL

Real-Collobrier research basin

J. Lavabre


RISK ASSESSMENTAll the technical components have the unique

purpose to provide the Service in charge of public warning with relevant information.

Messages design and dissemination, and dissemination response, are the most important tasks of a flood forecasting system which must be first a flood warning system

The answer is not in hydrological modelling


ARE YOU TALKIN’ TO ME ?

Documents

Hydrological Models for Flood Forecast · P. Givone, Cemagref - Direction Scientifique PREVISION DES CRUES 10 TENTATIVE S/T LOCKS AND CRITICAL COMPONENTS IDENTIFICATION For the headwater