By

Narendra Kumar Tiwary

(Executive Engineer, Water Resource Department, Bihar, India

Ph.D Scholar, IIT Delhi)

Prof. A.K.Gosain, IIT Delhi

Prof.A.K.Keshari ,IIT Delhi

Civil Engineering Department

Indian Institute of Technology, Delhi

Use of New Technologies & Softwares in Flood sector

Drainage Map of the Bagmati Basin

±

Legend

Drainage lines

Basin boundary20 0 20 40 6010

Km

IntroductionFlood is associated with a serious loss of life, property and damage to utilities.

05-08-2014 3

Event-1 The man-made embankments of river Kosi failed

and Flooded north Bihar, India during 18 August 2008

434 Dead bodies were found until 27 November 2008

05-08-2014 4

Kosi

Burhi

GandakGanda

k

05-08-2014 7

Event-2 In June 2013, a multi-day cloudburst centered on the

North Indian state of Uttarakhand caused devastating floods and landslides in the country's worst natural disaster since the 200 4 tsunami more than 5,700 people were "presumed dead. Destruction of bridges and roads left about 100,000 pilgrims and tourists trapped in the valleys

05-08-2014 8

Event-2

05-08-2014 9

Dehradun -Capital of Uttarakhand

05-08-2014 10

Challenge for Water Resource Engineers and Scientists Today, with modern equipments and radars we should

have been able to predict the floods and expected inundation much earlier and stopped these human disasters

05-08-2014 11

Important Questions to be answered

Where the floodplain and flood-prone areas are?

How often the flood plain will be covered by water?

How long the flood-plain will be covered by water?

At what time of year flooding can be expected?

05-08-2014 13

To modify SWAT model for real time flood forecasting and apply

modified SWAT model for real time flood forecasting of river basins

using web based data,

To enhance the forecast lead time by incorporating the rainfall

forecast issued by IMD and assess its accuracy.

To map expected inundation, and flood zonation by integrating

SWAT outputs into HEC GeoRAS and HEC-RAS models

Objective of the study

Real Time Flood Forecasting and Flood Plane Zoning System

COPONENTS The Real Time Flood Forecasting and Flood Plain Zoning

System usually consists of one or more of the following components:

1. Rainfall Runoff Model.

2. Runoff Routing Model.

3. Error Analysis and Updating Technique

4. Hydrodynamic model for channel routing

(HEC-RAS)

5. Generating TIN and delineating flood plain

05-08-2014 14

METHODOLOGY USED IN PRESENT STUDY1

.

1. Watershed delineation using SRTM

DEM

2. Formation of HRUS

3. Writing input tables

4. Runoff generation using Green Ampt

Method

5. Developing subroutines and

modifying SWAT for real time flood

forecasting application

Models for Forecasting Statistical/ Black box

Correlation

UH Based

Stochastic (ARMAX)

Conceptual (Water Balance)

SWAT

HEC-HMS

Methods for Flood Forecasting

Simple correlations

Use very little data

Only gauge data is sufficient

Work very well for reaches that have very

small contribution from rain

Physically Based ModelsTo use the equations of mass, energy and momentum to

describe the movement of water over the land surface and through the unsaturated and saturated zones.

05-08-2014 18

Ways to enhance the forecasts

Enhanced forecasts require data on

rainfall

To incorporate the contribution of

the catchment

To incorporate the impact of spatial

variability

Forecast lead time increases on

account of catchment lag

Addition to lead time is possible

through precipitation forecasting

Typical flood forecasting situation

Effective Lead Time

Total Forecast Lead Time

Time used by

observation and

forecasting

Flo

od

pro

du

cin

g s

torm

Tim

e e

lapsed in

obser.

Tra

nsm

issio

n T

ime

Data

pro

cessin

g

Fore

cast fo

rmula

tion

Tra

nsm

it ion tim

e

Rain

beginsRain

ends

Forecast reaches

userFlood reaches

outflow station

Rain

forecast

05-08-2014 21

Gamma Distribution UH Method

05-08-2014 22

05-08-2014 23

05-08-2014 24

05-08-2014 25

START

Read Input Data

Initialize Parameters

Daily Loop

Yes

End

Simulation

Last hr of Last Day

Simulation

If eventstatus=1 Perform Hourly Simulations

If eventstatus=2 Forecast hourly

Route

Command

Subbasin

Command

Read Sub –daily Precipitation

Write Outputs

to File to be Read by Another SWAT Run

Generate Max/Min Temperature

No

No

Compute and Print Final Basin Statistics

Capture Catchment Characteristics,Save values of Soil Water,Reach Storage,Base flow for next SWAT run

First Hour of First Day Of Simulation and want real time forecast

Hour loop

Import catchment characteristics saved in previous run

Routing Command loop

add

Command

save

Command

Add waterRoute water hourly

through channel

Yes

No

05-08-2014 26

05-08-2014 27

Call

openwth

Call std2

Call simulate

Call finalbal

Call writeaa

Call peatw

Call std3

Call xmon

Call clicon

Call command

Call writed

Call xmon

Call writem

Call tillmix

Call

PmeasCall Tmeas

Call

SmeasCall

HmeasCall

WmeasCall Pgen

Call Weatgn

Call Tgen

Call Clgen

Call Wndgen

Call Slrgen

Call Rhgen

Call Pgen

Aunif/Dstn1

Call

Aunif/Dstn1

Call Tgen

Call Clgen

Call Rhgen

Ee/Aunif/Dstn1Call Wndgen Aunif

Call Aunif/Dstn1

Call

Ee/Aunif/Dstn1Call Aunif

Call Weatgn Aunif/Dstn1

Call Slrgen

Call Subbasin

Call

Route

Call Recmon

Call Recyear

Call Save

Call Recday

Call Reccnst

Call Structu

Call Routres

Call Transfer

Call Addh

Call Saveconc

Call Varinit

Call

AlbedoCall Solt

Call Surface

Call Percmain

Call Etpot

Call Etact

Call Fert

Call Crpmd

Call Graze

Call Nminrl

Call Nitvol

Call Pminrl

Call Gwmod

Call Apply

Call Washp

Call Decay

Call Pestlch

Call Enrsb

Call

PestyCall Orgn

Call Psed

Call Nrain

Call Nlch

Call Solp

Call Subwq

Call Bacteria

Call Urban

Call Pothole

Call Filter

Call Latsed

Call Gwnutr

Call Surfstor

Call

SubstorCall Wetlan

Call Hrupond

Call Irrsub

Call Autoirr

Call Watuse

Call Watbal

Call Sumv

Call Virtual

Call Operatn

SWU (Soil Water

Uptake)

Call Curno

Grow (Plant Growth)

Tstr (Temp Stress)

Nup (Nitrogen Uptake)

Npup (Phosphorus Uptake)

Anfert (Autofert)

Call Nfix Call Nuts

Call Nfix

Call canopyint

Call snom

Call leaf_litter *

Call crackvol

Call volq

Call crackflow

Call surfst_h2o

Call alph

Call pkq

Call tran

Call eiusle

Call ysed

Call percmicro

Call

percmacro

Call Ee

Call Expo

Call Erfc

Call Theta

Call Regres

Call pond

Call Irrigate

Call Irrigate

Call SbsdayCall Sweep

Call impndday

Call Bsbday

Call alph

Call pkq

Call ysed

Call enrsb

Call pesty

Call orgn

Call psed

Call Tair

Call rchinit

Call rtover

Call rtday

Call rtmsk

Call rthourly

Call rthmsk

Call rtsed

Call watqual

Call noqual

Call rtpest

Call rtbact

Call irr_rch

Call rchuse

Call reachout

Call resnut

Call resinit

Call irr_res

Call res

Call lakeq

Call Borwet *

Call Curno

Call Curno

Call Plantop

Call Dormant

Call Harvkillop

Call Harvestop

Call Killop

Call Tillmix

Call impndmon

Call bsbmonCall rchmon

Call

writea

Call sbsmon

Call impndyr

Call rchmyr

Call sbsyr

Call bsbyr

Call

headout

Call header

Call swbl

Call vbl

Call

sbsaaCall

impdaaCall

rchaaCall

bsbaaCall

stdaa

Call

resopmrb

Call

sat_excess

Call

bmp_eff

Call dailycn

Call surq_daycn

Call surq_greenampt

Not

currently

active

Call rchday

Call curno Call ascrv

Call date_and_time

Call initial

Call std1

Call

readwwq

Call readinpt

Call

readsubCall readwgn

Call readswq

Call readrte

Call opensub

Call

readurban

Call readpnd

Call readwus

Call readres

Call readcrop

Call

readtillCall readpest

Call

readfert

Call readhru

Call readchm

Call readmgt

Call readsol

Call readgw

Call

soil_chemCall soil_phys

Call rteinit

Call h2omgt_init

Call hydroinit

Call impd_init

Call getallo Call

caps

Call

openfile

Call caps

Call readfig Call caps

Call readfile Call

caps

Call

readbsnCall

ascrv

Call

readcod

Call

gcycl

Call caps

Call

readlwqCall ascrv

Call

caps

Call

curno

Call

ttcoef

Call

allocate_parms

Call

Zero0Call

zero1Call

zero2

SWAT FlowChart

05-08-2014 29

05-08-2014 30

05-08-2014 31

05-08-2014 32

05-08-2014 33

05-08-2014 34

05-08-2014 35

Model Set up for study areaData sets used

The SWAT model requires data on terrain, land use, soil, and weather for assessment of inflows and outflows of reaches. Following datasets have been used for setting up of the SWAT model:

(1) DEM – SRTM (90 m resolution)

(2) Landuse – Global USGS (2 M)

(3) Soil – FAO Global soil (5 M)

(4) Rain gauges /Temperature gauges – IMD

(5)Stream Gauges – CWC05-08-2014 36

Study Area

INPUT DEM

05-08-2014 38

05-08-2014 39

05-08-2014 40

05-08-2014 41

Division into subbasins

05-08-2014 42

Soil Details

05-08-2014 43

SEQN SNAM NLAYERS HYDGRP TEXTURE SOL_AWC1 SOL_K1 CLAY1 SILT1 SAND1

3663 3663 2 C LOAM 0.117 35.65 24 35 42

3664 3664 2 C LOAM 0.157 28.52 17 36 47

3682 3682 2 D LOAM 0.175 7.77 24 36 40

3684 3684 2 C LOAM 0.175 13.92 22 38 41

3695 3695 2 C LOAM 0.175 14.96 20 40 40

3743 3743 2 D LOAM 0 6.48 18 44 38

3761 3761 2 C LOAM 0.175 24.73 20 34 46

3808 3808 2 D LOAM 0.175 6.17 21 35 44

3851 3851 2 D CLAY_LOAM 0.137 7.17 27 35 37

Table-6.2 Soil properties

GAUGE STATIONS

05-08-2014 44

05-08-2014 45

05-08-2014 46

05-08-2014 47

05-08-2014 48

05-08-2014 49

05-08-2014 50

y = 108.73x - 4204.1R² = 0.8808

-500.00

0.00

500.00

1000.00

1500.00

2000.00

0.00 20.00 40.00 60.00

dis

ch

arg

e in

m3

/s

Stage in M

discharge-stage for Hayaghat

Series1

Linear (Series1)

05-08-2014 51

y = 4E-09x3 - 1E-05x2 + 0.0182x + 38.219R2 = 0.9605

0.000

10.000

20.000

30.000

40.000

50.000

60.000

0.00 500.00 1000.00 1500.00 2000.00

Sta

ge in

M

Discharge in Cumecs

Stage-discharge for Hayaghat

Series1

05-08-2014 52

y = 4E-07x3 - 0.0002x2 + 0.0567x + 44.834R² = 0.936

44.000

45.000

46.000

47.000

48.000

49.000

50.000

0.000 50.000 100.000 150.000 200.000 250.000 300.000

Sta

ge i

n M

Discharge in Cumecs

Stage-Discharge for Benibad

Results

05-08-2014 53

Calibration - Output

Days

Dis

charg

e m

3/s

Calibration Results Seventeen parameters were selected for sensitivity

analysis, calibration, validation and uncertainty analysis to be carried out by SWAT-CUP4 using SUFI algorithm. On the basis of global sensitivity analysis it can be concluded that most sensitive parameter is CN2.mgt followed by GW_DELAY, CH_N2.rte, ESCO.hru, HRU_SLP.hru, GWQMN.gw, GW_REVAP.gw and SURLAG.bsn. P-factor and r-factor for calibration were found to be 0.74 and is 0.44 respectively, which are very much within the range recommended for a perfect model. Seventy four percent observed and simulated values lie in 95PPU. P-factor and r-factor for validation period are 0.63 and

05-08-2014 55

Validatio for monthly flow

05-08-2014 56

Model Performance Test

The correlation statistics (R2) was determined to measure the linear correlation between the actual and the predicted values and it was found to be 0.93 which indicates the best performance of the model.

R 2 = = 0.96

To estimate the efficiency of the fit, the Nash-Sutcliffe coefficient (Nash) (Nash and Sutcliffe, 1970) was used. It was found to be 0.96 which indicates the best performance of the model.

Nash = 1 - = 0.7505-08-2014 57

Daily Flow Valdation

05-08-2014 58

Flow Hydrograph Daily Simulation

05-08-2014 59

0.0

20.0

40.0

60.0

80.0

100.0

120.0

140.0

160.00

200

400

600

800

1000

1200

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Dis

cha

rge

in

cu

me

cs

1-July Time in days 31-July

Flow Hydrograph of Hayaghat of July-04

RF at Simra

RF at Kathmandu

RF at Nagarkot

RF at Benibad

RF at Hayaghat

HayaOBS

HayaSim

Valdation for daily flow 2010

05-08-2014 60

Validation for hourly flowNE = 0.72

05-08-2014 61

Validation Graph

05-08-2014 62

ERROR ANALYSIS Time series analysis of errors in hourly

simulations/forecasts has been performed. An ARIMA model developed for forecasting error has been deployed for correcting simulated/forecasted hourly flows. Integration of simulation capability of SWAT model and error forecasting capability by time series analysis has been done for solving the problem of real-time flood forecasting

05-08-2014 63

Error 2004 Simulation

05-08-2014 64

-400

-300

-200

-100

0

100

200

300

400

500

1

16 31 46 61

76 91

106

121

136

151

166

181

196

211

226

241

256

271

286

301

316

331

346

361

376

391

40

6

421

436 451

46

6

48

1

49

6

511

526

541

556

571

586

Dis

cha

rge

in

cu

me

cs

4-July-04 Time in hours 28-July-04

ERROR 4-July-28-July-04

ERROR

Error 2004 hourly simulation

05-08-2014 65

-60

-40

-20

0

20

40

60

80

100

120

1

18 35 52 69

86

103

120

137

154

171

188

205

222

239

256

273

290

307

324

341

358

375

392

40

9

426 44

3

46

0

477

49

4

511

528

545

562

579

596

FIRST DIFF OF ERROR 4-July-28-July on y-axix and time in hrs on X-axix

FIRST DIFF

Error analysis of 2004 flood event

ARIMA Model Parameters for error analysis for hourly simulation of year 2004

ARIMA Parameters

B1(LAG1)= 1.3153

B2(LAG2)= -0.0162

B3(LAG3)=-0.3013

05-08-2014 66

Error Analysis There are 600 observed and simulated values. Four

hundred data have been used for determining ARIMA model parameter and two hundred forecasted errors have been determined. Nash Sutcliffe coefficient before correction for forecasted period of error was 0.64 and after error correction it is 0.795.ARIMA model parameters are given below.

b = 1.3153, -0.0162, -0.3013

Z(t)= 1.3153*Z(t-1)-0.0162*Z(t-2)-0.3013*Z(t-3)

05-08-2014 67

05-08-2014 68

0

100

200

300

400

500

600

700

800

900

1 11 21 31 41 51 61 71 81

91

101

111

121

131

141

151

161

171

181

191

201

211

221

231

241

251

261

271

281

291

301

311

321

331

341

351

Dis

cha

rge

in

cu

me

cs

2-Aug Time in hour 22Aug

Hourly Simulation 2nd August-22nd August-05

OBS

SIMCorrected

SimOld

Hourly 2005 ARIMA Model Parameters for error analysis for hourly

simulation of year 2005

ARIMA Parameters

B1(LAG1)= 0.6056

B2(LAG2)= 0.1792

B3(LAG3)=0.0988

05-08-2014 69

05-08-2014 70

05-08-2014 71

Real time flood forecasting graph

05-08-2014 72

Real time forecasting graphs

05-08-2014 73

0.00E+00

5.00E+00

1.00E+01

1.50E+01

2.00E+01

2.50E+01

3.00E+01

3.50E+01

4.00E+01

4.50E+010

200

400

600

800

1000

1200

1400

1600

1 8 15 22 29 36 43

50 57 64 71 78 85

92

99

106

113

120

127

134

141

148

155

162

169

176

183

190

197

204

211

20 July time in hours 28 July

Real time forecasts issued 0n 20th July and 23 rd July'04

rainfall

Obs

Forecast-205-9

Forecast-202-5

Forecast-202-5

Real time forecast

05-08-2014 74

Forecast generated at Benibad for July 2004 event

05-08-2014 75

Forecast generated at Benibad for July 2004 event

05-08-2014 76

Real time flood forecasting Graph

05-08-2014 77

Real time flood forecasting of Haya

05-08-2014 78

Hourly Discharge

05-08-2014 79

SUMMARY AND CONCLUSIONS

05-08-2014 80

(1)Increase in lead time(cathment response delay+travel time in channel)

(2)Application possible in ungauged cathments also

(3)Sediments,water quality and land use can be modelled on real time basis

simultaneously

(4)Physically based

(5)Easy in adaptability

(6)Capable of modelling snow melt, percolation, lateral flow and base flow

(7)Parsimonious

(8)GIS based

(9)Inundation mapping and flood zonation easily possible

(10)Easy updating of catchment characteristics

(11)Capable to incorporate rain fall forecast of IMD and increase lead time

(12)Increase in accuracy

(13)Capable of generating synthetic sub-hourly hydrographs

CONCLUSIONS

05-08-2014 81

As outcome of the present research work it can be

concluded that Modified SWAT model combined

with time series error forecasting ARIMA model and

HEC-RAS hydraulic model can be used as an

efficient tool for real-time flood forecasting and

inundation mapping. Modified SWAT model

reproduces stream flow hydrographs reasonably

under multiple storm events and can be used as a

real-time flood forecasting tool with enhanced lead

time. Rainfall forecast can further increase the

forecast lead time.

Thank You

END

05-08-2014 82