PhD candidate: Davide Ceresetti
Director: Jean-Dominique CREUTIN Co-director: Gilles MOLINIÉ
Space-time characterization of heavy rainfall events:
Space-time characterization of heavy rainfall events:
Application to the Cévennes-Vivarais region
Application to the Cévennes-Vivarais region
Université de Grenoble
!
Introduction Introduction
!
Methodological development Methodological development
!
Application: Severity Diagrams Application: Severity Diagrams
!
Conclusions Conclusions
OUTLINE OF THE PRESENTATION OUTLINE OF THE PRESENTATION
!"#$
INTRODUCTION
INTRODUCTION
!
Context:
Extreme rainfall in a Mediterranean Mountainous Region
1958-1994:
Daily amount > 190 mm Total: 144 events
Jacq (1994)
Warm humid air from Mediterranean Sea + Orography = Storms
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
$"#$
General overview General overview
General overview General overview
Cévennes-Vivarais: region prone to catastrophic fl ash-fl oods
"#$!%&'(&)*&+!,--,!./001
Social and economic impact (human lives, damages,...)
Social and economic impact (human lives, damages,...)
20+)3/!40516(6057
Specifi c discharge:
5-10 m
3s
1km
289--!:;
="#$
General overview General overview
"#$!%&'(&)*&+!,--,!./001
How can we measure the magnitude of extremes?
How can we measure the magnitude of extremes?
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
="#$
Is it a « hydrological monster »or a regular event?
Is it a « hydrological monster »or a regular event?
Impact of storms at various durations Impact of storms at various durations
#"#$
;>4?40
Spatial and temporal scales are related
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
#"#$
;>4=40
Impact of storms at various durations Impact of storms at various durations
Spatial and temporal scales are related
#"#$
;>4@40
Impact of storms at various durations Impact of storms at various durations
Spatial and temporal scales are related
<,=>?9@"-#
!4A!
Spatial and temporal scales are related
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
#"#$
;>4!A40
Impact of storms at various durations
Impact of storms at various durations
Aim of the study Aim of the study
HOW TO ESTIMATE THE MAGNITUDE OF RAINFALL EVENTS?
(c)
19 September 2000
(a)
22–23 September 1993
(b)
7 September 1998
BCBDE 3 F ,
B5/&'4*+(3'-)&4G0H # @I# A
E)'3143(/314(3-&4*/6'04G77H #JI? J?I= !$KI@
.3L46)-&'4(3-&M3114G77H !#! !!A ?!K
.3L40)+(1:4-&'/&8-':4G77"0H !?# A? @@
!"#$%&'"#()')*$*)+,&-../0&.),1 !"#$%&'"#()')*$*)+,&-../0&.),1 !"#$%&'"#()')*$*)+,&-../0&.),1
Storms over Marseille
A"#$
!"#$%&'"#()')*$*)+,&-../0&.),1
HOW TO ESTIMATE THE MAGNITUDE OF RAINFALL EVENTS?
(c)
19 September 2000
(a)
22–23 September 1993
(b)
7 September 1998
!"#$%&'"#()')*$*)+,&-../0&.),1 !"#$%&'"#()')*$*)+,&-../0&.),1 !"#$%&'"#()')*$*)+,&-../0&.),1
NO DAMAGES NO DAMAGES 60 M€
Classic statistics are unable to detect the more dangerous event
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
A"#$
Aim of the study Aim of the study
Need of a multi-scale descriptor of storms Need of a multi-scale descriptor of storms
Maximum rainfall intensity
:65 :3B
Integration Smoothing Trivial scale pattern
;+(3'-)&N4!@K47-&
O(/3N4!AK4P7?
%&'/&8-':>4?#47740Q!
R"#$
Proposition: transform max intensity into FREQUENCY Proposition: transform max intensity into FREQUENCY
SEVERITY DIAGRAMS: Event magnitude at all scales SEVERITY DIAGRAMS: Event magnitude at all scales
;+(3'-)&N4!@K47-&
O(/3N4!AK4P7?
%&'/&8-':>4?#47740Q!
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
@"#$
S S everity diagrams: a storm comparison tool everity diagrams: a storm comparison tool
(c)
19 September 2000
(a)
22–23 September 1993
(b)
7 September 1998
Weak event Local event Heavy and extended event DS4;ODTBU V%.%EB;4;ODTBU DANGER DANGER
Ramos et al., 2005
J"#$
Improvements proposed in the thesis Improvements proposed in the thesis
BEFORE AFTER
Size of the region 250 km 2 32000 km 2 Involved events Urban fl oods Flash-fl oods
Regional model Point rainfall extremes Spatial rainfall extremes
EMPIRICAL SCALE-INVARIANT MODEL
SPACE-TIME MODEL EMPIRICAL
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
!K"#$
<W5/&&/8QC-53(3-8
X:*()Q.W'W)()1)2-,314SF8/(53')(:
O(/3N4$?KKK4P7
?A larger region A larger region
.3(8/-11/43(/3N4?#K4P7
?Take into account Take into account spatial heterogeneity spatial heterogeneity
!!"#$
Improvements proposed in the thesis
Improvements proposed in the thesis
Mediterranean Sea Rhône River Cévennes Massif
Geographical context Geographical context
Cévennes-Vivarais région
Size 160 x 200 km
2Elevation 0 – 1950 m
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
!?"#$
Geographical context Geographical context
Cévennes-Vivarais région
The region gathers fl at lands , a SE oriented foothill , a mountain ridge and a plateau .
!
!
?
?
$
$
=
=
= $ ? !
C
CD C
CD
!?"#$
Climatic features: average annual rainfall (mm) Climatic features: average annual rainfall (mm)
Mountain ridge:
Over 2000 mm / year
Mediterranean sea shore:
less than 1000 mm / year
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
!$"#$
Measurement network Measurement network
OHM-CV:
OHM-CV: one of the Europe densest rain gauge networks (1/50 km
2) Cévennes- Vivarais Hydro-Meteorological Observatory Radar ARAMIS network Rain gauge network
Hourly (150 gauges, 1993-2008) Daily (225 gages, 1958-2000)
!="#$
Université de Grenoble PART II PART II
METHODOLOGICAL METHODOLOGICAL
DEVELOPMENT
DEVELOPMENT
ACCURATE MODELING OF EXTREMES ACCURATE MODELING OF EXTREMES
TOYTB;49<OVB9N41)&24(/,)(*84353-13F1/
For a reliable magnitude estimation For a reliable magnitude estimation
% ZG%[-H
EXTRAPOLATION
\03'4-84'0/4732&-'+*/
4)M4'0-84)F8/(53'-)&]
!A"#$
SERIES RECONSTRUCTION THROUGH SERIES RECONSTRUCTION THROUGH SCALE-INVARIANCE METHODS SCALE-INVARIANCE METHODS
ZG%[-H
%
YDTOYTB;49<OVB9N4&)47/38+(/7/&'843'4'0/4*/8-(/*48,31/
E
;
!;
$;
?\03'4-84'0/4*-8'(-F+'-)&43'4 444444'0/4*/8-(/*48,31/]
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
!R"#$
For a reliable magnitude estimation For a reliable magnitude estimation
ACCURATE MODELING ACCURATE MODELING OF EXTREMES OF EXTREMES
ROBUST MODELING OF
EXTREMES AT VARIOUS SCALES
TOTB;49<OVB9 YDTOTB;49<OVB9
F GHFI6J
EXTRAPOLATION
SERIES RECONSTRUCTION THROUGH SERIES RECONSTRUCTION THROUGH
SCALE-INVARIANCE METHODS SCALE-INVARIANCE METHODS
GHFK6J
F
;
!;
$!@"#$
For a reliable magnitude estimation For a reliable magnitude estimation
Dealing with ungauged scales: SCALING Dealing with ungauged scales: SCALING
SCALING OF A PROCESS
relation between probability distributions of a process at different scales
%4G77"0H 6G%H
! " " # $%# * # ! " " ! $
44444444444!!=40 44444444444!!?40 44444444444!!!40
=40
?40
!40
ZUS^O^%V%E_4;%9EU%^YE%SD
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
!J"#$
%L3/&!65M3+635L&!0.!'1.7
Prerequisite: Evaluation of rain gauge uncertainties Prerequisite: Evaluation of rain gauge uncertainties
```4``
44`
E(+/4U3-&M3114G77H SF8/(5/*4
U3-&M311 G77H
?K"#$
Evaluation of rain gauge uncertainties Evaluation of rain gauge uncertainties
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
?!"#$
Wind effect Neglectable for high intensities
Bottom hole lamination underestimation in case of very high intensities
Tipping-bucket device
Rain collector
Heavy rainfall Underestimation: 5-10% 5-min rainfall 2-5 % hourly rainfall
?!"#$
Reversal time:
~ 0.2 s in which no water is stored
!
Experimental calibration
!
Numerical Simulation
Evaluation of rain gauge uncertainties
Evaluation of rain gauge uncertainties
Tails behavior Tails behavior
Identifi cation of the behavior of point-rainfall extremes
```4``
44`
Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
%4G77"0H 6G%H
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
??"#$
Open question: how are the distribution tails of rainfall?
Upper bounded (Weibull)
Exponential (Gumbel)
Hyperbolic (Fréchet)
44
441)24 446G%H
1)24G%H
$4O''(3,'-)&4;)73-&8 Tails behavior
Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
?$"#$
'3P/&4M()74./8'(/a4?KK@
Tails behavior
Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model One has 3 possibilities:
1) Extract Maxima
2) Peaks over Threshold 3) Work on distributions
N&1OL&!73)'/&!76P&
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
?="#$
At various duration --> tail behavior of point rainfall series
Ceresetti et al, 2010, WRR
Tails behavior
Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
Rigorous method
!
K-S test for lower bound x
min!
Estimator for power-law slope
Goldstein et al, 2004, Clauset et al., 2007
L
7-&Straight line in log-log Power-law Fréchet distribution
ZS\ BUQV
O\
?#"#$
DUAL BEHAVIOR: Need of a GENERALIZED model for EXTREMES DUAL BEHAVIOR: Need of a GENERALIZED model for EXTREMES
Ceresetti et al, 2010, WRR
Tails behavior
Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
Flat lands Flat lands
Mountainous region Mountainous region
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
?A"#$
Gumbel
Fréchet
```4``
44`
Tails behavior Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
%4G77"0H ZG%H
Construction of a scaling model for point rainfall maxima
?R"#$
Menabde et al, 1999
Veneziano et Furcolo, 2002
%4G77"0H ZG%H
! " " # $%# * # ! " " ! $
44444444444!!Z;b4G!4c//PH 44444444444!!Z;b4G!4*3:H 44444444444!!Z;b4G?40H
Annual Maxima Distribution is scale-invariant in the range 2 h – 1 week
Tails behavior Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
?@"#$
%4G77"0H ZG%H
& " # $ '%# ( & " ! $ '
44444444444!!Z;b4G=40H 44444444444!!Z;b4G?40H 44444444444!!Z;b4G!40H
4441)2 E-7/4G0H
!40 ?40 =40
4441)2
7)7/&'8 4444[%
de
b-(8'47)7/&'4Gd>!H 9/,)&
*47)7
/&'4Gd>?H E0-(*
47)7/&'4Gd>$H
Scaling moment linearity in log-log vs scale
! " " # $%# * # ! " " ! $
PROBABILITY DISTRIBUTION STATISTICAL MOMENTS
Gupta et al., 1990
Tails behavior
Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
?J"#$
Extreme distribution defi ned through moments
Tails behavior Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
Moments scaling Extreme distribution scaling
EXY9
ZG%f-H
%
R4;O_9 !4;O_ !4XSYU
B g EU
O ZS
VO E% SD
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
$K"#$
Tails behavior Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
IDF: Intensity – Duration – Frequency curves
Q0R!S Q0R!6!
H))TUJ
8U 8!V&&W
IDF: implicit scale-invariance relations for one T R
$!"#$
!K4:(8
!KK4:(8 5
8 3
8--%;b %
& ' ()
% * " (
)$ &
)+"()$Tails behavior Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
IDF: Intensity – Duration – Frequency + GEV Extremes
Q0R!S Q0R!6!
H))TUJ
8U 8!V&&W
8
IDF + GEV Model for all scales and all T R
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
$!"#$
!K4:(8
!KK4:(8 5
8 3
8--X!GSY!H)3B6)3J
TBC
%;b %
& ' ()
% * " (
)$ &
)+"( )$Tails behavior Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
GEV simple-scaling IDF model
X!GSY!H)3B6)3J
Q0R!S Q0R!6!
H))TUJ
8U 8!V&&W
8
Use of daily series to estimate extreme rainfall down to 1 hour
$!"#$
*
&%+ ,
&-./" 0 $, -
&-./. ) -
&-./. / " ! ).$0" &
&
-./$
)+
.)7/&'848,31-&2
-
&"
% 1
&-./"" 0 $.
"/ " ! ) " .$)/ " ! ).$
"" &
&
-./$
)+.
&%.
&-./TBC
%;b %
& ' ()
% * " (
)$ &
)+"( )$Need of a regional model for IDF relations
Mountainous region
Flat lands
Tails behavior
Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
$?"#$
GEV simple-scaling IDF model:
Rainfall Tr=100 years ,!U 9!U
8,!U ,>!U
ROBUST:
Obtained using daily data
(> 50 years of data at 225 gages)
.3L-7+7N b13'413&*
.3L-7+7N 7)+&'3-&
(-*2/
Daily data hides information on infra-daily scale Tails behavior
Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
$$"#$
Ceresetti et al, 2011, Submitted to WRR
```4``
44`
Need to model extremes in space Tails behavior
Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
$="#$
Tails behavior
Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
%4G77"0H ZG%H
! " " # $%# * # ! " " ! $
44444444444!!Z;b4G=KK4P7
?H 44444444444!!Z;b4G=K4P7
?H 44444444444!!Z;b4G!4P7
?H
RADAR IMAGERY spatial scale-invariance
detected in the range 1-400 km
2RADAR Few events, not enough data
Solution 1: Statistics on radar data
$#"#$
Tails behavior
Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
Solution 2: Interpolation of point data
Signifi cant underestimation of maxima in coarse networks
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
$A"#$
Tails behavior
Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
Solution 2: Interpolation of point data
Spatial undersampling Underestimation maxima 20-50%
%6)O/3(&1!.6&/1 F5(&+'0/3(605!0.!'065(!)&37O+&)&5(7
$R"#$
ARF computed from historical series 1993-2008
ARF: Areal Reduction Factor
8 OUb
O4GP7
?H
Example: rainfall fi eld" (- " 2 '( $% " (- " #$ ( $1 2)3 (- " 2 '( $
Tails behavior Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
2)3 % " 2
" # 2 !
Solution 3: Semi-empirical model based on gages
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
$@"#$
Tails behavior
Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
Solution 3: Semi-empirical model based on gages
We can build AREAL REDUCTION FACTOR
Dynamic scaling model for ARF
(
2 4 % #5+16
Dynamic scaling ratio
De Michele et al., 2001
$J"#$
ARF in Cévennes- ARF in Cévennes- Vivarais region Vivarais region
Duration has lower infl uence in mountain
C+&3!HW)
,J CNY
Flat Lands
?0 =0 @0 !?0 ?=0
Tails behavior Tails behavior Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
=K"#$
C+&3!HW)
,J CNY
Mountainous region
?0 =0 @0 !?0 ?=0
Point rainfall model Point rainfall model h Spatial rainfall model Spatial rainfall model
9!U Y/3(!/3517
=!"#$
Point rainfall model Point rainfall model Spatial rainfall model Spatial rainfall model
Regional model for assessing the magnitude of extremes Regional model for assessing the magnitude of extremes
h
9!U Y/3(!/3517
Severity Diagrams Severity Diagrams
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
=!"#$
APPLICATION:
APPLICATION:
SEVERITY DIAGRAMS
SEVERITY DIAGRAMS
Storm comparison
Use of Severity Diagrams Use of Severity Diagrams
Observed storm Virtual storm
(numerical simulation)
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
=$"#$
From rainfall fi elds to severity diagrams From rainfall fi elds to severity diagrams
O (/ 34 GP 7
?H
O,,+7+13'-)&4*+(3'-)&4G0H
Multi-step process involving historical series
9/5/(-':4;-32(374G:(8H
Z6)&!HUJ
U3-&423+2/4*3'3
U3*3(4*3'34)(48-7+13'/*47)*/1
=="#$
!
%(0+)!L0)'3+6705
!
:01&/!&M3/O3(605
!
N&3/#(6)&!)056(0+65R
Severity Diagrams computation Severity Diagrams computation
AREAL REDUCTION FACTOR INTENSITY- DURATION- FREQUENCY
9 Z O E %O V 4U O %D bO V V 9 B C B U %E _
Assign to each spatial rainfall observation Assign to each spatial rainfall observation a frequency value (severity) a frequency value (severity)
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
=#"#$
Application of severity diagrams Application of severity diagrams
U/86/,'4')4U37)84QQe413(2/(4*)73-&4c-'0
!
4X/'/()2/&/-':4)M4(3-&M3114/L'(/7/4F/035-)(4GM13'413&*8a47)+&'3-&IIH
!
4<)3(8/(4(3-&M3114&/'c)(P
\/4,3&4+8/4)+(4863,/Q'-7/47)*/14')4,)76+'/49BCBU%E_4;%OTUO.9 M)(43&:48')(74),,+((-&24-&4'0/4<W5/&&/8
Applications:
1. Evaluation of meso-scale deterministic simulations (MesoNH) 2. Evaluation of the variability of Ensemble simulations (AROME)
=A"#$
Evaluation of deterministic simulations performance: 2005, Sep 06 Evaluation of deterministic simulations performance: 2005, Sep 06
Wrong Maximum Location - Rainfall Underestimation – Different space-time scales Wrong Maximum Location - Rainfall Underestimation – Different space-time scales
4&+&7&((6!&(!3/[!,-88[!7O*)6((&1!(0!\CY
]*7&+M&1!+365.3//
:&702^!+365.3//
SF8/(5/*48/5/(-':
9-7+13'/*48/5/(-':
Small scale max ~ 500 yrs 3-4 hours / 0-100 km2
Large scale max ~ 300 yrs 7-10 hours / 0-30 km2
Small scale max ~ 50 yrs 3-6 hours / 0-50 km2
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
Deterministic simulation performance: 2003, Dec 03 Deterministic simulation performance: 2003, Dec 03
4&+&7&((6!&(!3/[!,-88[
!7O*)6((&1!(0!\3Y
]*7&+M&1!+365.3//
:&702^!+365.3//
SF8/(5/*48/5/(-':
9-7+13'/*48/5/(-':
Maximum Severity: ~500 yrs Time scale: 9-14 h
Spatial scale: 0-200 km2
Maximum Severity: ~500 yrs Time scale: 14-18 h
Spatial scale: 200-500 km2
Severity: an effective multiscale diagnostic
=R"#$
Evaluation of ensemble simulations variability Evaluation of ensemble simulations variability
Determine the variability of the members
.)*/1N4OUS.B4S6/(3'-)&314.)*/14G.W'W)Qb(3&,/H
Z()F3F-1-8'-,4/&8/7F1/N42()+64)M48-7+13'-)&84)F'3-&/*46/('+(F-&2
F56(63/!40516(6057
_0O513+`!40516(6057
?KK@a4D)5/7F/(4K!49')(7
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
=@"#$
Application: effect of initial conditions Application: effect of initial conditions
Space-time scales OK, LOW magnitude
S^9BUCB;
9BCBU%E_ ZBUES^9QOUS.B
i! ZBUES^9QOUS.B
i?
ZBUES^9QOUS.B
i$ ZBUES^9QOUS.B
i= ZBUES^9QOUS.B
i#
=J"#$
Université de Grenoble PART IV PART IV
CONCLUSION AND CONCLUSION AND
PERSPECTIVES
PERSPECTIVES
Conclusion Conclusion
.O%D4UB9YVE9N
!
4j+3&'-M-,3'-)&4)M4'0/4/(()(4,)77-''/*4F:4(3-&423+2/4*/5-,/8
!
4j+3&'-M-,3'-)&4)M4'0/4+&*/(/8'-73'-)&4*+/4')4863'-314+&*/(83761-&24
!
4O88/887/&'4)M4'0/4'3-14F/035-)(4)M4(3-&M3114/L'(/7/84
!
4U/2-)&31-k3'-)&4)M46)-&'"863'-314(3-&M3114/L'(/7/84G%;bhOUb4,+(5/8H
!
4;/'/,'-)&4)M4'0/4863'-3143&*4'/76)(3148,31/Q-&53(-3&,/4(3&2/8
!
4B8'-73'/4'0/4732&-'+*/4)M40/35:4(3-&M3114/5/&'84-&4'0/4<W5/&&/84(/2-)&
!
4B531+3'-)&4)M46/(M)(73&,/4)M47/'/)()1)2-,3147)*/18
#!"#$
Perspectives Perspectives
!
4U3-&423+2/4+&,/('3-&'-/8N4*/8-2&4)M434&/c4(3-&423+2/4&/'c)(P4GX:./g46()l/,'H
!
4U/2-)&314%;bhOUb4,+(5/8N48/5/(-':4,03(3,'/(-k3'-)&4)M4M+'+(/4/5/&'8
!
49/5/(-':4*-32(3784)M4/&8/7F1/48-7+13'-)&8N4-76()5/7/&'4)M4/&8/7F1/48-7+13'-)&84
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
#?"#$
;-88/7-&3'-)&4)M434(-2)()+84M(37/c)(P4-&4c0-,04 6(/,-6-'3'-)&4-848//&438434,)&'-&++74)M48,31/8
#$"#$
<3&4c/43661:4'0-84M(37/c)(P4')4)'0/(453(-3F1/8]
Université de Grenoble Université de Grenoble Université de Grenoble
EXODm4_SYn
;-88/7-&3'-)&4)M434(-2)()+84M(37/c)(P4-&4c0-,04
(3-&M3114-848//&438434,)&'-&++74)M48,31/8
%&'()*+,'-)&
%&'()*+,'-)& ./'0)*)1)2-,314*/5/1)67/&'8./'0)*)1)2-,314*/5/1)67/&'8 9/5/(-':4;-32(3789/5/(-':4;-32(378 <)&,1+8-)&<)&,1+8-)&
