Water Management and Dam Division, River DepartmentWater Management and Dam Division, River DepartmentJuly 2009July 2009

Sophistication of Dam Operation for Flood Control by Sophistication of Dam Operation for Flood Control by Use of Rainfall Prediction TechnologyUse of Rainfall Prediction Technology

Climate Change Adaptation Research GroupClimate Change Adaptation Research Group

National Institute for Land and National Institute for Land and Infrastructure ManagementInfrastructure Management

Functions of multipurpose dam洪水調節方式（国土交通省ダム、水資源機構ダム）

その他方式, 11

一定量放流方式,

27

一定率一定量放流方式, 46

自然調節方式, 26

Flood control methods used at multipurpose dams can be broadly categorized into four types: ungated control method, constant volume discharge method, constant-rate constant-volume discharge method and others (e.g., bucket bottom cut method). The method most frequently used by the Ministry of Land, Infrastructure, Transport and Tourism and Japan Water Agency is the constant- rate constant-volume discharge method.

Flood control methods

What is the "constant-rate constant-volume discharge method"?

In the constant-rate constant-volume discharge method, when inflow has exceeded the flood control action level discharge (discharge below which downstream areas and facilities in river areas do not suffer major damage; the dam operating rules define refers to it as "flood discharge"; also known as "harmless discharge"), inflow is reduced at a constant rate until the peak inflow. After the peak inflow, water is released by constant volume.

In this method, the flood control effect of dams can be expected even in cases of medium or small floods. This method, therefore, is suitable for rivers whose downstream channels have not been improved substantially through river improvement projects. Compared with the constant-volume discharge method, however, this method requires a larger flood control reservoir capacity.

Constant-volume discharge method TIme

Discharge

Outflow discharge

Inflow

Discharge cutoff

Constant-rate constant-volume discharge method

Time

Flood control action level discharge =

flood discharge

Flood control methods (MLIT and JWA dams)

Discharge Discharge cutoff

Outflow discharge

Inflow

Other methods

Ungated control method

Constant-volume discharge method

Constant-rate constant-volume

discharge method

Functions of multipurpose dam

If the reservoir level is expected to exceed the surcharge level (upper limit level of flood control capacity) (i.e., if an extreme flood is expected), when the reservoir level has exceeded the 80% level of the reservoir capacity, dam operation procedures are taken so as to make the outflow discharge close to the inflow in order to prevent the reservoir level from exceeding the surcharge level.

＝

These are called particular operating rules.

Besides the flood control methods as mentioned earlier, methods for other purposes such as water supply are stipulated in the operating rules for each dam.

When the occurrence of a flood exceeding the design basis flood for the dam is expected, the following action is taken.…

Note: In this case, the outflow discharge will never exceed the inflow.

Note: The outflow discharge will never exceed the inflow at or around the flood peak discharge.

Functions of multipurpose damExample of operation under the particular operating rulesIf a great flood exceeding the design flood for the flood control plan occurs, the discharge-reducing effect of the dam decreases.

無害放流量

600 [m3/s]

計画最大放流量

2400 [m3/s]

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

7/17 7/18 7/19 7/20 7/21 7/22 7/23 7/24 7/25 7/26 7/27 7/28 7/29

流量

[m3/s]

0

20

40

60

80

100

120

140

160

180

200

雨量

[mm/hr]

実績流域平均雨量[mm] 実績ダム流入量[m3/s]

実績ダム放流量[m3/s]

有効貯水容量

77,500 [千m3]

制限水位容量

5,500 [千m3]予備放流水位容量

3,000 [千m3]0

20,000

40,000

60,000

80,000

100,000

120,000

7/17 7/18 7/19 7/20 7/21 7/22 7/23 7/24 7/25 7/26 7/27 7/28 7/29

ダム

容量

[千m3]

実績ダム容量[千m3]

T Dam: Example during the July 2006 flood

Maximum inflowMaximum outflow discharge

Almost full

Design maximum outflow discharge

Harmless outflow dischargeDis

char

ge

Rai

nfal

l

Observed basin average rainfall [mm]

Actual outflow discharge [m3/s]

Actual reservoir inflow [m3/s]

Res

ervo

ir ca

paci

ty [

1,00

0 m

3] Effective storage capacity

Flood control level capacityFlood-season limited water level capacity

Actual reservoir capacity [1,000 m3 ]

Problems of current dam operating methods(1) In the event of an extreme flood, a outflow discharge higher than the design

maximum outflow discharge is needed under the so-called "particular operating rules." As a result, the damage-mitigating effect is reduced.

→ It is hoped that operation under the particular operating rules is avoided.(2) Even if an extreme flood is expected, preliminary release is not necessarily

carried out.→ There are no established criteria based on rainfall prediction for deciding whether to

carry out preliminary releases. (3) In the constant-rate constant-volume discharge method (a method which is

currently in use at many dams and which does not utilize rainfall prediction), even when a medium or small flood is expected and the reservoir has sufficient flood control capacity, downstream damage might be caused by a outflow discharge exceeding the harmless discharge.

→ If it can predicted that a medium or small flood is coming, flood control can be done by carrying out a outflow discharge within the limit of the harmless discharge.

(4) Since water supply capacity is not currently used for flood control, the effectiveness of the flood control function relying solely on flood control capacity is limited.

→ Water supply capacity can also be utilized by making effective use of rainfall prediction.

(5) Information related to rainfall prediction is lacking, and rainfall predictions are not currently used for flood control operation.

→ The accuracy of rainfall prediction is not clearly known. Operating methods using rainfall predictions have not been clearly indicated.

Purpose of researchFurthermore . . .The frequency of damaging floods has increased in recent years, and flood risk is expected to increase in the coming years under the influence of climate change. River channel improvement has not progressed as expected because of the reduced budget for public works projects. The construction of a new dam takes considerable time and funding.

Therefore . . .It is hoped, in order to make effective use of existing dams, that a large quantity of water is stored in ordinary times and when a major flood is expected, the required capacity for flood control is created by making a preliminary release.

Therefore . . .It is hoped, in order to make effective use of existing dams, that a large quantity of water is stored in ordinary times and when a major flood is expected, the required capacity for flood control is created by making a preliminary release.

○○

Social needsSocial needs

• Mitigation of flood damage in the event of an extreme flood• Flood damage prevention in the event of a medium or small flood• Efficient flood control made possible by reconsidering conventional dam operation• Rational dam operation in coordination with water supply

Merits of dam operation methods using rainfall prediction

Use of rainfall prediction makes the following possible:Use of rainfall prediction makes the following possible:

Flood control capacity is increased by making preliminary releases, and damage is minimized by reducing the maximum outflow discharge by the constant-volume discharge method.

流量In case of extreme In case of extreme floodflood

If prediction is possible, water supply capacity can be used as part of flood control capacity.

New operation method

Outflow discharge by new operation

method

Outflow discharge

under current

operating rules

時間

Inflow

Preliminary release

If flood control capacity is thought to be larger than necessary for flood control, part of flood control capacity can be used to store flood water to eliminate downstream damage.

In case of medium In case of medium or small floodor small flood

Outflow discharge by new operation

method

流量

時間

Inflow

Discharge

Time

Outflow discharge

under current

operating rules

Discharge

Time

Previous study results and bottlenecks・Improvement of rainfall forecasting techniques (FY2007/2008)

A prediction study has been conducted by using a new forecasting model (WRF) which uses JMA predictions as initial values and boundary values. The study has confirmed that rainfall forecasting accuracy can be enhanced.→

There is still a certain degree of uncertainty because there are cases where accuracy is not improved. Quantitative evaluation of accuracy is needed.

・Effective rainfall calculation method considering the water-holding capacity of soil (FY2007/2008)The relationship between the duration of dry weather and the recovery of the water- holding capacity of soil has been formulated, and a runoff prediction method considering the initial rainfall loss has been developed.→

There are still uncertainty factors such as interception by trees and depressions

・Establishing a dam operating method using rainfall prediction (FY2008)Up-to-48-hour rainfall predictions and the reservoir capacity at the time of prediction were compared, and a method for planning releases including preliminary releases has been established.

→ If rainfall prediction errors are large, there are cases where a large quantity of water is released even though there is still storage capacity or where because storage capacity is not large enough, the reservoir becomes full so that particular operating rules are applied. There is still need to consider actions to be taken when rainfall predictions have turned out to be inaccurate.

Rainfall forecasting technique (What is WRF?)

WRF Weather Research and Forecasting Model

Mesoscale weather model developed by NACR/NCEP (USA) for forecasting and research applications

Though developed from a single project, there are two types, one for forecasting applications and the other for research applications.

Based on the lessons learned from the emergence of many mesoscale models in the United States, the new model has been developed through a concerted effort.

What is a mesoscale model?

Non-hydrostatic model dealing with meteorological phenomena with horizontal scales ranging from 10 to 100 km, such as localized heavy rains and thunderstorms

Capable of 1-kilometer mesh computation

Hydrostatic model dealing with global scale weather phenomena such as typhoons and Bai-u (rainy season) fronts

10-km mesh is the limit because of hydrostatic approximation

Used for forecasting applications because computation can be done quickly

What kind of weather model is used for weather

forecasting?

Computational flow of mesoscale weather model• Define computation domain (nesting)• Enter information in computational domain(automatic downloading from USGS, etc.)

Terrain dataLand use dataVegetation dataSoil data

Download data and prepare initial and boundary value data• Global and regional objective analysis data• Computation results from global and regional weather models

PreprocessingDefine computation domain

Enter initial values and boundary values

Forecasting computation

PostprocessingDisplay computation results

Data assimilation

Specify physical process models• Rainfall models (multiple models)• Radiation models (multiple models)• Boundary layer models (multiple models)

• Although rainfall forecasting accuracy is dependent on the kind of physical process models used, but it is determined mainly by the degree of closeness of the initial values and boundary values to the realities.

Digital national land information is also available.

3D data on wind direction, wind speed, air pressure, air temperature and humidity

Displayed computation results (examples)

3-hour forecasting 12-hour forecasting

Computation results obtained by using RSM data as initial and boundary values (a case in which accuracy is improved)

Forecasting computation by WRF Forecasting computation by JMA's RSM

•Forecasting accuracy of RSM has been improved by using WRF•RSM was fairly accurate in predicting the start and end of rainfall although the predictions were not accurate quantitatively.

Basin of Sameura Dam (August 30, 2004)

Calculation start time differs.

radar-AMeDAS

ground-based observation

radar-AMeDAS

ground-based observation

Flowchart for simulation of dam-based flood control using rainfall prediction

As

the

pred

icte

d in

flow

vol

ume

beco

mes

incr

easi

ngly

gre

ater

than

the

rem

aini

ng

rese

rvoi

r cap

acity

, the

met

hod

of re

leas

e fr

om th

e da

m is

upg

rade

d fr

om (1

) to

(5).

Predicted rainfall data(time horizon: up to 48

hours)

Calculate cumulative rainfall

Predicted total inflow

Rainfall-runoff process model

Total runoff

Predicted total

inflow

Water- holding capacity

Total rainfall

Predicted total rainfall(cumulative rainfall within up to 48 hours)

Remaining reservoir capacity ≥

Total inflowRemaining reservoir capacity is greater than total inflow.

↓(1) All inflow could be intercepted, but water is released at the safe outflow discharge only when the inflow is higher

than the safe outflow discharge.

Volume to be controlled in excess of safe outflow discharge = Total inflow −

Volume that can be released at safe outflow discharge

Calculated volume that can be released at safe outflow discharge = Duration of predicted

effective rainfall × Safe outflow discharge

Remaining reservoir capacity ≥

Volume to be controlled in excess of safe outflow discharge

(2) Constant-volume interception at safe outflow discharge

Calculate preliminary outflow discharge and

volume that can be released in advance.

Preliminary release period = Predicted time of start of flood −

Present time −

Preparation time

Conditions for preliminary release• Outflow discharge restrictions for downstream safety• Discharge capacity of outlet facilities

Volume to be controlled in excess of safe outflow discharge ≤

Volume that can be released in advance Constant-volume interception at the safe outflow discharge can be made by combining it with preliminary release.

↓(3) Constant-volume interception not exceeding safe

outflow discharge during flood period + Preliminary release

Calculate volume that cannot be controlled by

preliminary release

Calculate the constant volume outflow discharge through

convergence analysis so that the remaining reservoir

capacity is used up.

Required outflow discharge ≤

Design maximum outflow discharge

(4) Constant-volume interception at a rate higher than the safe outflow discharge + Preliminary release

(5) Constant-volume interception at a rate higher than the design maximum outflow discharge +

Preliminary release (for protection from beyond- design-basis flood). The outflow discharge,

however, must not be excessively high.

Evaluating rainfall prediction errors in simulationThe prediction accuracy of JMA's VSRF, MSM and RSM at 214 rain gauge stations in the catchments of dams in 7 river systems in Japan has been evaluated.

Study on usability of weather forecast data (NILIM)

Regression coefficient = Regression coefficient = Predicted rainfall / Measured rainfallPredicted rainfall / Measured rainfall

Expectable maximum measured Expectable maximum measured rainfall = Predicted rainfall / 0.7 = rainfall = Predicted rainfall / 0.7 = 1.43 1.43 **

Predicted rainfallPredicted rainfall

Expectable minimum measured Expectable minimum measured rainfall = Predicted rainfall / 1.4 = rainfall = Predicted rainfall / 1.4 = 0.71 0.71 **

Predicted rainfallPredicted rainfallFig. 1 Upper and lower limits of the regression coefficient for cumulative rainfall determined from past study results

Use of past study research (analysis using JMA forecasts)It is necessary to make forecasting by using the newly developed WRF to accumulate

data and evaluate accuracy.

Use of past study research (analysis using JMA forecasts)It is necessary to make forecasting by using the newly developed WRF to accumulate

data and evaluate accuracy.

Upper limit of regression coefficient = 0.4 / 48 × Rainfall duration + 1.0

Reg

ress

ion

coef

ficie

nt fo

r con

tinuo

us ra

infa

ll Regression coefficient: 0.8

Lower limit of regression coefficient = 0.7

predicted rainfall

Rainfall duration

Measured rainfall = Predicted rainfall / 0.8

measured rainfall

Surface soil moisture condition RG = Surface soil moisture condition RG = Maximum waterMaximum water--holding capacity holding capacity ＊＊

expexpααTT

WaterWater--holding capacity during the rainfall holding capacity during the rainfall = Maximum water= Maximum water--holding capacity holding capacity −−

RR GG

Calculating effective rainfallInitial rainfall loss, primary runoff ratio, saturation rainfall and maximum rainfall loss are determined on the basis of total runoff and total rainfall data for single rainfall events at each dam.

The recovery of the maximum water- holding capacity is determined, by assuming it to be an exponential function, from the discharge reduction condition after the second inflection point during the discharge reduction period during a flood.

Effective rainfall at K DamEffective rainfall at K Dam

初期損失雨量 40 [mm]

一次流出率 0,4

飽和雨量 210 [mm]

最大表土保水

能力

102 [mm]

Fig. 2 Total rainfall and direct runoff at K Dam

(colored according to time from initial stage of rainfall)

Primary runoff ratio

Saturation rainfall

Maximum water- holding capacity of surface soil

Initial rainfall loss

3-5 days

5-10 days

10-20 days20 days or more

Less than 3 days

Dire

ct ru

noff

[mm

]

Total rainfall [mm]

Direct runoff = Surface runoff + Early intermediate runoff Intermediate runoff +

Groundwater runoffGroundwater runoff

1st inflection point

Inflow at

K Dam[mm/ hr]

Rainfall[mm]

2nd inflection point

Aug

. 30,

198

8

18:0

0

Aug

. 31

06:0

0

12:0

0

18:0

0

Sep

t. 1

06:0

0

12:0

0

18:0

0

Fig. 3

Calculation of reduction factor (example)

06:0

0

12:0

0

18:0

0

Sep

t.2

12:0

0

In (total inflow) [mm/ hr]

cumulative rainfall

• River use downstream and the rate of water level rise

• Time required for river patrol and the scope of patrol

• Determination of release pattern taking into consideration river users' ability to escape

Proposed procedure for safety checks for preliminary release -draft-

10cm

30cm

50cm

70cm

90cm

流速 V (m/s)

水深 H

0.5 1.0 1.5 2.0 2.5 3.0

成人男性

成人女性

子供・高齢者

対象

断面

のH-V

● cm

■ cm

▲ cm

安全性を考慮した放流パターン

0

2

4

6

8

10

12

14

0:00 1:00 2:00 3:00 4:00 5:00

豊平

峡ダ

ム放

流量

(m3/s)

経過時間

本検討パターン

現行の放流の原則

90min

10min

現行のパターンに戻る

初期流量は水深0.05m時の流量

0.00m3/sを設定）

安全性を考慮した放流パターン

1時間30分経過まで0.00m3/s放流

1時間40分経過まで0.03m3/s放流

1時間50分経過まで0.11m3/s放流

2時間00分経過まで0.23m3/s放流

以後2m3/sまで0.7m3/s/10min

0.7m3/s以上8m3/sまで2.1m3/s/10min

8m3/s以上18m3/sまで3.5m3/s/10min

18m3/s以上30m3/sまで5.0m3/s/10min

30m3/s以上50m3/sまで6.2m3/s/10min

60m3/s上限

< Important considerations>

Critical depth Critical flow velocity

Maximum velocity

Adult males

Adult females

Children/ the elderly

Source: Suga et al.: Experiment on safe evacuation behavior (wading through water) in case of flood

Water depthH

H-V

for c

ross

se

ctio

n of

in

tere

st

Adult males

Adult femalesChildren/the elderly

Flow velocity V

Release pattern chosen taking safety into consideration0.00 m3/s until 90 minutes pass0.03 m3/s until 100 minutes pass0.11 m3/s until 110 minutes pass0.23 m3/s until 120 minutes passThereafter 0.7 m3/s/10 min until 2 m3/s2.1 m3/s/10 min at 0.7 m3/s–8 m3/s3.5 m3/s/10 min at 8 m3/s–18 m3/s5.0 m3/s/10 min at 18 m3/s–30 m3/s6.2 m3/s/10 min at 30 m3/s–50 m3/s60 m3/s: upper limit

Out

flow

dis

char

ge a

t H

ohei

kyo

Dam

Return to currently used pattern

Discharge (0.00 m3/s) corresponding to a water depth of 0.05 m is used as the initial discharge.

Release pattern chosen taking safety into consideration

pattern under consideration

currently applied release principle

Time

Adult males

Adult females

Children／the elderly

Range in which safe evacuation is difficult

Range in which safe evacuation

is possible

Result of simulation of new operation method

In 22 out of the 30 flood cases in which water was actually released at out higher than harmless discharge, it would be possible to reduce water at outflow discharge lower than harmless discharge.

In 13 out of the 30 flood cases in which water was actually released at outflow discharge higher than harmless discharge, preliminary releases would be carried out.

In 8 out of these cases, it would be possible to release water outflow discharge lower than harmless discharge. In the remaining 5 cases, peak discharge would be reduced.

Results obtained in 64 flood cases at 7 dams最大放流量別ケース数の比較

26

6

4

2

34

56

0

10

20

30

40

50

60

実績 降雨予測を

活用した操作

ケー

ス数

無害放流量以下の放流ケース

ただし書き操作ケース

無害放流量以上の放流ケース

実績で無害流量以上の放流ケースにおけるシミュレーションでの事前放流実施状況

4

211

8

14

0

5

10

15

20

25

30

降雨予測を活用した操作

ケー

ス数

事前放流無しで無害放流量以下

事前放流により無害放流量以下

事前放流無しで計画最大放流量以上（流量は低減）事前放流により計画最大放流量以上（流量は低減）事前放流無しで計画最大以下

事前放流により計画最大放流量以下

Comparison of the number of cases by maximum discharge

Num

ber o

f cas

es

below harmless discharge case

particular operating rule case

above harmless discharge case

Actual resultsOperation using rainfall prediction

Below harmless discharge case without preliminary releaseBelow harmless discharge case with preliminary release

Above design maximum outflow discharge case without preliminary release (discharge reduced)

Above design maximum outflow discharge case with preliminary release (discharge reduced)

Below design maximum outflow discharge without preliminary release

Below design maximum outflow discharge with preliminary release

Num

ber o

f cas

es

Operation using rainfall prediction

Simulated preliminary releases in past above harmless discharge cases

It has been confirmed that preliminary releasing is effective to a certain degree against an extreme flood. A preliminary release, however, should be carried out in view of such factors as prediction errors and the possibility of the recovery of water supply capacity and the exceeding of flood control capacity.

At a dam where the constant-rate constant-volume discharge method is used, the decision to narrow down options to constant-volume discharge at the harmless discharge, which uses the flood control capacity more, should be made carefully. →

The relationship between forecasting accuracy and flood

control needs to be determined.

In predicting reservoir inflow, it is desirable to use a high-accuracy runoff analysis model, such as a distributed runoff model.

There are dams incapable of carrying out effective preliminary releases because of the lack of capacity of outlet facilities where preliminary releases would in theory be effective. → There is a need to reconsider existing reservoir outlet facilities.

EvaluationEvaluation Dam operation using WRF

Determining outflow discharges in cases where flood control capacity is not large enough even after a preliminary release is carried out

Policies needs to be determined as to how to use reservoir capacity in view of WRF errors

Problems to be addressedProblems to be addressed

Confirming WRF errors in order to optimize flood control so that downstream damage is minimized

Conducting runoff analyses using distributed runoff models, etc., to forecast reservoir inflow more accurately and perform more appropriate flood control operation taking into consideration the safety of downstream areas

< Improving operating methods>

< Improving forecasting techniques>

It is necessary to develop operating rules or operating manuals for new flood control methods and conduct studies on where the responsibility for operation lies and on actions to be taken dealing with risks.

< Redesigning institutional systems>

Research scheduleFY2008

Verifying the accuracy of WRF-based rainfall forecasting applied to historical floods and evaluating its usefulness in flood control

Verifying the accuracy of WRFVerifying the accuracy of WRF--based rainfall based rainfall forecasting applied to historical floods and forecasting applied to historical floods and evaluating its usefulness in flood controlevaluating its usefulness in flood control

Studies on criteria for carrying out preliminary releases taking into consideration the magnitudes of rainfall events, forecasting errors and characteristics of dams

Studies on criteria for carrying out preliminary Studies on criteria for carrying out preliminary releases taking into consideration the releases taking into consideration the magnitudes of rainfall events, forecasting magnitudes of rainfall events, forecasting errors and characteristics of damserrors and characteristics of dams

Forecasting reservoir inflow by making combined use of a distributed runoff model and rainfall forecasting, and evaluating the accuracy of forecasting

Forecasting reservoir inflow by making combined use Forecasting reservoir inflow by making combined use of a distributed runoff model and rainfall forecasting, of a distributed runoff model and rainfall forecasting, and evaluating the accuracy of forecastingand evaluating the accuracy of forecasting

Predicting rainfall by using an improved WRF (accumulation of data)

Verifying prediction accuracy

Predicting rainfall by Predicting rainfall by using an improved using an improved WRF (accumulation of WRF (accumulation of data)data)

Verifying prediction Verifying prediction accuracyaccuracyStudies on the type and accuracy of rainfall prediction

information necessary for dam operation Studies on the type and accuracy of rainfall prediction Studies on the type and accuracy of rainfall prediction information necessary for dam operationinformation necessary for dam operation

Studies on risk management using rainfall prediction and studies on dam operating rules using rainfall prediction by WRF Studies on risk management using rainfall prediction and studiesStudies on risk management using rainfall prediction and studies on dam on dam operating rules using rainfall prediction by WRFoperating rules using rainfall prediction by WRF

Improving WRF by using GSM20, precipitable water, etc.

Improving WRF by Improving WRF by using GSM20, using GSM20, precipitableprecipitable water, etc.water, etc.

Establishing a concept for dam operation methods taking rainfall prediction into consideration

Establishing a concept for dam operation Establishing a concept for dam operation methods taking rainfall prediction into methods taking rainfall prediction into considerationconsideration

Validating dam operation methods using rainfall predictionValidating dam operation methods using rainfall predictionValidating dam operation methods using rainfall prediction

FY2009

FY2010

FY2011