Page 1
Flood Forecasting &
Decision Support System for Flood
Control and Early Warning
Dr. LIU Zhiyu
Bureau of Hydrology, Ministry of Water Resources of China
Homepage: http://xxfb.hydroinfo.gov.cn/EN/eindex.jsp
Email: [email protected]
International Training Programme 2010
Management of Flood Control and Disaster Mitigation
June 2010 China
Page 2
影响人口
死亡人口
灾害损失
Global Major Disasters
Page 3
Death toll due to flooding
Slides
Droughts
Earthquakes
Epidemics
Extremes tem
peratures
Floods
Wild Fires
Wind Storm
s
Volcanoes
Other N
atural Disasters
Non-N
atural Disasters
TOTAL
Oce
ania
US &
Can
ada
Res
t of A
meric
asEur
ope
AfricaAsi
aTot
al
0
50000
100000
150000
200000
250000
300000
350000
400000
450000
Type of disaster
Nu
mb
er
of
peo
ple
kill
ied
Oceania
US &
Canada
Rest of
Americas
Europe
Africa
Asia
Total
Total number of people reported killed, by continent and by type of phenomenon (1990 to 1999, source: IFRC World
Disaster Report 2000)
Page 4
Floods
Seasonal (riverine) flood monsoon rainfall, snowmelt, etc.
Flash flood heavy short-term storm, snowmelt, etc.
Urban floodDebris & mud flow (often as a flash flood)
Snowmelt flood Ice jam flood Glacier lake outburst flood Dam-break flood
Page 5
Flood Risk Management
Structural structuresConstruction of dams and river embankments River improvement works, floodway, flood
retarding basin, etc. Slope protection (erosion control)High-floored houses, shelters, riverside forests
and bamboos, etc.
Non-structural structures Legal framework, coordination among stakeholders Land use management flood monitoring, forecasting & early warning
systems Preparedness –hazard mapping, improving
communication, education to create awareness Insurance and mutual aid
Page 6
Data Collecting
Decision
Making
Meteorology
Hydrological
Engineering
Drought
Damage
Loss
Forecasting &
Prediction
Non-structure Measure for Flood Management
4 step
(洪水管理4环节)
Em
erg
en
cy
Re
sp
on
se
Activitie
s
+ rebuilding
预测预报 调度决策
救灾重建
Flood forecasting plays a key role in flood management
Page 7
Questions:
1. WHAT IS REAL-TIME FLOOD FORECASTING ?
2. UNDER WHICH CONDITIONS IS IT POSSIBLE ?
3. WHAT ARE COMMON APPROACHES OF FF ?
4. WHAT IS THE STATE-OF-THE-ART OF FF?
5. China NFFC’s decision support systems?
Page 8
WHAT IS REAL-TIME
FLOOD FORECASTING?
Page 9
•A Flood Forecasting System is a tool aimed at reducing
uncertainty on the future evolution of a flood event..!!!
•It is the reduction of uncertainty that allows for more reliable
decisions.
•Real-time flood forecasting provides a possibility of providing
the authorities in charge of the management of catastrophic
situations such as floods, timely and sufficiently accurate
information on the possible future evolution of the events
within a pre-determined time horizon.
Page 10
“Timely” means
early in time to allow for the implementation of the
possible interventions.
“Sufficiently accurate” means
to be associated to a measure of its uncertainty
and within a tolerable error band (e.g.10-20%).
“Pre-determined time horizon” means
time horizon must be sufficiently long (e.g. 12hr)
to allow for the implementation of decisions.
Page 11
UNDER WHICH CONDITIONS
IS IT POSSIBLE ?
Page 12
Large-size catchments (>10,000 km2) - flood forecast can be
performed up to 12 hours in advance, only by upstream level
and/or discharge measurements.
89/63
Page 13
Medium-size catchments (between 10,000 and 1,000 km2),
flood forecast can be performed if a telemetering rain gauge
network is available.
超声波水位计
Page 14
Small-size catchments (< 1,000 km2), - flood forecast is
possible only if a precipitation forecast is available.
Very small-size or urban catchments (< 200 km2) - it is
impossible to provide an accurate real-time forecast, radar
precipitation estimates are essential.
Radar DataPrecipitation
Estimates
Page 15
WHAT ARE COMMON APPROACHES
OF FLOOD FORECASTING?
Page 16
A schematic of a flood forecasting system (NERC, 1994)
• a precipitation forecasting
model (deterministic and/or
stochastic)
• a catchment model
(deterministic and/or
stochastic)
• a flood routing model
• A flood plain model
• A geographical information
system (GIS)
• A geo-referenced data
bank
• An expert system shell
Page 17
Fiver steps taken in operational flood forecasting
(1) In the synoptic weather situation, when rainstorms occur, the flood is
roughly estimated and forecasted according to the quantitative
precipitation forecasting, or the possible rainfall.
(2) Flood forecasts are successively made with the rainfall-runoff model
duration by duration according to rainfall.
(3) When flooding has appeared in the upper reach, the flood will be
forecasted for the lower reach based on the corresponding water level (or
discharge).
(4) If there is much difference between the observed values and forecasted
values, the forecasts should be real-time adjusted.
(5) If taking structural measures, such as flood diversion or flood storage,
or if dyke break occurs, for flood control policy-making, regulation of
flood forecasts should be made.
Page 18
Principle of operational flood forecasting
•Different Approaches,
• Group Consultation, and
• Comprehensive Analysis
Page 19
Empirical forecasting
schemes
Mathematic models for
flood forecasting
Tw
o m
ajor ap
pro
aches o
f FF
降雨径流相关 上下游水位(流量)相关
Rs
RI
E P and EM
IM
RB
Qs
CI
CG
KG
RG
QI
QG
K B,WM
R
1-FR FR
W
WU
WL
WD
S
SM
EX
KI
UH
∑Q
EU
EL
ED
UM
LM
C
XINANJIANG MODEL新安江蓄满产流模型
Operational flood
forecasting
WHAT ARE COMMON APPOACHES OF FLOOD FORECASTING?
Page 20
Empirical flood forecasting methods
o API Model (P~Pa~R + UH)
P
t初损后渗法
P
tP~Pa~R
Q
t
实测
计算
P~Pa~R Unit Hydrograph
Page 21
Empirical flood forecasting methods
上游
实测
演算
Q
t
Q
t
实测
演算
涨落急剧型 涨落平缓型
Page 22
),(
),(
),(
)(
1
1
1
1
PZfZ
QZfZ
ZZfZ
ZfZ
UtDt
UtDt
UtDtDt
UtDt
Improved accuracy
Empirical flood forecasting methods
o Corresponding stage (discharge) method
Page 23
Resultant discharge method
)(
)(
1
1
Utm
UtDt
QfZ
QfQ
Resultant discharge (m3/s)upQ Peak stage at Pingle station (m)
Page 24
Stage (discharge) fluctuating rate method
),(
),(
1
1
tttt
tt
QZfZ
KZfZ
Page 25
Multi-factors combined axes correlation method
),,(
),,(
0
0
TZPfZ
PZPfZ
m
cm
1
2
3
4
5
6
7
72
60
48
3624
12
57 58 59 60 61
Mean area precipitation(mm)
150 100 50
Peak stage at Yangshan station(m)
0 59 60 61 62 63 64 65 66
Page 26
Rainfall runoff correlation method
),(
]),[(
am
am
PPfQ
QPPfQ
P+
Pa (m
m)
Runoff (mm) Peak discharge(m3/s)
Page 27
Mathematic models for flood forecasting
Statistical
(black
box)
Lumped
Conceptual
(grey box)
Distributed
Physically-based
(white box)
Deterministic Stochastic
Watershed Hydrological Models
Joint
Stochastic-Deterministic
Semi-Distributed
Conceptual
(grey box)
Page 29
Mathematic models for flood forecasting
Page 30
• Tank model
• Snowmelt-runoff model (SRM)
• Inflow-storage-outflow (ISO) function models
• Conceptual watershed model for flood forecasting (China)
• Conceptual watershed model (the HBV model)
• Sacramento soil moisture accounting model (NWSRFS-SAC-SMA)
• Snow accumulation and ablation model (NWSRFS-SNOW-17)
• Synthesized constrained linear system (SCLS)
• Non-linear rainfall runoff model – URBS
And, the combined streamflow forecasting and routing models
• Streamflow synthesis and reservoir regulation (SSARR)
• Manual calibration program (NWSRFS-MCP3).
HOMS website: http://www.wmo.int/web/homs/projects/HOMS_EN.html
Models for forecasting streamflow from
hydromeorological data
Page 31
The main factor is:
• the understanding and the correct definition of the
purposes, for which the method or the model will be used.
•the availability of data, that heavily conditions the selection
of the type of modelling approach, which ranges from the
simple statistical methods to the extremely detailed physical
process models.
Choosing an appropriate flood forecasting model
Page 32
The following additional factors have to be considered when selecting a
model:
1) Forecasting lead time vs time of concentration (or travel time when
dealing with routing problems)
2) Robustness of the approach, in the sense that, when dealing with
real time forecasting, sudden instabilities or large forecasting errors must be
avoided at all cost, even by resorting to slightly less accurate approaches.
3) Computational time, in that the forecast must be made available in
time to the flood managers and dependent responders, to guarantee the
effectiveness of their decisions. Frequently this requirement discourages the
use of sophisticated and accurate, but time consuming, approaches.
Choosing an appropriate flood forecasting model
Page 33
WHAT ARE THE STATE-OF-THE ART
OF FLOOD FORECASTING?
Page 34
IMPROVEMENT OF RAINFALL MEASUREMENT
New approach: multi-sensor rainfall data assimilation
using Bayesian approaches with an aim to improve
rainfall measurements
Page 35
RAINGAUGES METEO SATELLITEMETEO RADAR
Direct measurement of precipitation
Point measurement
At point quantitatively good while spatially low quality
Indirect measurement of precipitation
Good spatial description but quantitatively poor
Precipitation intensity
Wind and positioning
Evaporation
Clutter o Ground Clutter (non-meteorological echoes)
Attenuation
Beam Blocking
Orographic Enhancement
Bright Band, snow, etc.
Presently poor algorithms for QPS
Several under development
Discontinuous in time
Indirect measurement of precipitation
Good spatial description but coarse and quantitatively poorer than RADAR
Page 36
RAINGAUGES RADAR & SATELLITE BAYESIAN COMBINATION
BLOCK KRIGING OF THE RAINGAUGES + RADAR
UPSCALING
BK GAUGES + RADAR at satellite scale
SATELLITE
KALMAN FILTER
BK GAUGES + RADAR + SATELLITE at satellite scale
DOWNSCALING (KALMAN SMOOTHING)
BLOCK KRIGING OF THE RAINGAUGES + RADAR + SATELLITE
y yP~
Page 37
2.8
0.0
April 15 1998 – @ 19:00
BLOCK KRIGING
BK + RADAR BK + SATELLITE BK+RADAR+SATELLITE
RADAR SATELLITE
Page 38
BAYESIAN COMBINATION & TOPKAPI
13-18 April 1998
Discharge at CASALECCHIO
0
50
100
150
200
250
300
15 21 27 33 39 45 51 57 63 69 75 81 87 93 99 105
time (h)
Dis
char
ge
(m3/
s)
OBSERVED
BK
RADAR
BK+RADAR
BK+SATELLITE
BK+RADAR+SATELLITE
RADAR
UNDER-ESTIMATION
Page 39
Discharge at CASALECCHIO
0
500
1000
1500
2000
2500
3000
1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 103 109 115 121 127 133 139
time (h)
Dis
ch
arg
e (
m3
/s)
OBSERVED
BLOCK KRIGING
RADAR
BK + RADAR
BK + SATELLITE
BK + RADAR + SATELLITE
Discharge at CASALECCHIO
0
100
200
300
400
500
600
700
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67
time (h)
Dis
ch
arg
e (
m3
/s)
RADAR OVER-ESTIMATION
BAYESIAN COMBINATIONS & TOPKAPI
13-22 November 2000
Page 40
IMPROVEMENT OF HYDROLOGICAL MODELS
The new millennium: the derived from topography
models, e.g.
THE SCN + GIS MODEL
THE TOPMODEL MODEL
THE TOPKAPI MODEL
THE LISFLOOD MODEL
THE tRIBS MODEL
Page 41
Chq
x
qr
t
h
Point eq.
Watershed eq.
s
T
T
sss
sVba
t
V
s
T
T
ooo
oVba
t
V
c
T
T
ccc
cVba
t
V
Distributed Model Lumped Model
Large Scale eq.
trhh tx 0,0,
*
0,0, trhcq tx trhh tx 0,0,
*
0,0, trhcq tx
Integrated eq.
TOPKAPI Model and parameter lumping
Page 42
+
=
+
TOPKAPI: minimum data requirements
Data
Model
D.E.M. Soil Types Map Land Uses Map
TOPKAPI is developed and calibrated from widely available maps
TOPKAPI is one of the first distributed models that can run in real time
TOPKAPI can be directly linked to Radar,Satellite, Meteorological Models to provide reliable flood forecasts
Page 43
the Yihe River basin
A=5 318 km2
the Huaihe River basin A=10
900 km2
the Yangtze River basin
A=15 500 km2
Chinese basins
Page 44
Soil
DEM
Landuse
Huaihe River
A=10,900km2 at the Xixian station
GLDAS global soils dataset of Reynolds :
http://www.ngdc.noaa.gov/seg/eco/cdroms/reynolds/reynolds/reynolds.htm
UMD 1km Global land cover :
http://www.geog.umd.edu/landcover/1km-map/meta-data.html
USGS GTOPO30:
http://edcwww.cr.usgs.gov/landdaac/gtopo30/gtopo30.html
Flood Forecasting for the Huaihe River basin by
using TOPKAPI
Page 45
0
1000
2000
3000
4000
5000
6000
7000
80001998-5-1 0:00
1998-5-8 0:00
1998-5-15 0:00
1998-5-22 0:00
1998-5-29 0:00
1998-6-5 0:00
1998-6-12 0:00
1998-6-19 0:00
1998-6-26 0:00
1998-7-3 0:00
1998-7-10 0:00
1998-7-17 0:00
1998-7-24 0:00
1998-7-31 0:00
1998-8-7 0:00
1998-8-14 0:00
1998-8-21 0:00
1998-8-28 0:00
1998-9-4 0:00
1998-9-11 0:00
1998-9-18 0:00
1998-9-25 0:00
1998-10-2 0:00
1998-10-9 0:00
1998-10-16 0:00
1998-10-23 0:00
1998-10-30 0:00
日期和时间
流量 (m3s-
1)
0
20
40
60
80
100
120
140
160
流域平均降水
计算
实测
Calibration
Page 46
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
100002002-5-1 0:00
2002-5-8 0:00
2002-5-15 0:00
2002-5-22 0:00
2002-5-29 0:00
2002-6-5 0:00
2002-6-12 0:00
2002-6-19 0:00
2002-6-26 0:00
2002-7-3 0:00
2002-7-10 0:00
2002-7-17 0:00
2002-7-24 0:00
2002-7-31 0:00
2002-8-7 0:00
2002-8-14 0:00
2002-8-21 0:00
2002-8-28 0:00
2002-9-4 0:00
2002-9-11 0:00
2002-9-18 0:00
2002-9-25 0:00
2002-10-2 0:00
2002-10-9 0:00
2002-10-16 0:00
2002-10-23 0:00
2002-10-30 0:00
日期和时间
流量 (m3s-
1 )
0
20
40
60
80
100
120
140
160
180
200
流域平均降水
计算
实测
Validation
Page 47
Flood Forecasting for the Dongjin reservoir by
using TOPKAPI
A=1080km2 at the Dongjin reservoir
Page 48
SoilDEM
Landuse
Digital Soil Map of the World and Derived Soil Properties on CD-ROM
(FAO, 1998)
USGS Global Land Cover Characteristics Data Base :
http://edcdaac.usgs.gov/glcc/global_int.html (2000)
USGS HYDRO1k data set
http://edcdaac.usgs.gov/gtopo30/hydro/ (2000)
Reservoir-inflow
In-reservoir
Grid map of
“reservoir-
inflow” cells
Page 49
Calibration
0
500
1000
1500
2000
2500
6-7
6-10
6-13
6-16
6-19
6-22
6-25
6-28 7-1
7-4
7-7
7-10
7-13
7-16
7-19
7-22
7-25
7-28
7-31 8-3
8-6
8-9
Time(△t=3hr)
discharge(
m3/s)
Cal.
Obs.
Page 50
Validation
0
200
400
600
800
1000
1200
4-1
4-2
4-4
4-5
4-7
4-8
4-10
4-11
4-13
4-14
4-16
4-17
4-19
4-20
4-22
4-23
4-25
4-26
4-28
4-29 5-1
Time(△t=3hr)
Discharge(
m3/s) Obs.
Cal.
Page 51
Flood event statistics (time step=3 hrs)
Page 52
Principal calibrated model parameters
Page 53
USE OF METEOROLOGICAL QUANTITATIVE
PRECIPITATION FORECASTS
Page 54
(2) Application of NWP/QPF into Flood Forecasting
QPF FFS
FFS
reportingdata
Decision-
making
QPF
Page 55
NWP
20×20km,
3-hour,
up to 5 days
Page 56
Flood
forecasting
(with rainfall
forecasts)
Forecasted Observed
Page 57
China NFFC’s
Decision Support System?
Page 58
MISS — Meteorological Information Service System
(“天眼”气象信息综合业务系统)
TCS — Tropic Cyclones System (热带气旋系统)
NCHIOS — National Comprehensive Hydrological
Information Operational System
(全国水情综合业务系统)
FFS — Flood Forecasting System (洪水预报系统)
ISS — Information Service System
(水文信息综合服务系统)
Operational Systems
Page 60
Rainstorm Monitoring: auto warning
Page 61
自动报警
Flood Monitoring: auto warning
Page 62
库容曲线
统计信息
Integrated Information Enquiry
Page 63
历史过程比较
测站视频
Integrated Information Enquiry
Page 64
Comprehensive Hydrological Analysis
Page 65
Forecasting-based Flood Early Warning
Page 66
Forecasting-based Flood Early Warning
Page 67
Flash flood forecasting
and warning
Page 68
6/22/2010 68
Flash flood forecasting
and warning
Page 69
National Flood Forecasting System (NFFS)
Page 70
National Flood Forecasting System (NFFS)
Page 71
The empirical relationship method
Page 72
Automatic calibration of model parameters
Page 73
Real-time operational flood forecasting
Page 74
Interactive flood forecasting program
Page 75
Various forecasts by different forecasters
Page 76
Forecasts are made along the river trees