EVALUATION OF ArcSWAT MODEL FOR STREAMFLOW SIMULATION

2012 International SWAT Conference

By

Arbind K. Verma1 and Madan K.Jha2

1. Assistant Professor, SASRD, Nagaland University,

Ngaland (India)

2. Professor, AGFE Department, IIT Kharagpur (India)

2

Introduction

3

Water Availability m3/person/year

Country’s Status

>2000 Adequate

1000-2000 Water-stressed

<500 Water Scarce

Source: Bouwer (2000)

1820

1140

2001 2050Source: Gupta and Deshpande (2004)

Per

cap

ita

wat

er

avai

labi

lity

m3/year

m3/year P

opul

atio

n

1640 million

1028 million

Present and future per capita water availability in India

Introduction (Contd.)

4

Introduction (Contd.)

5

Introduction (Contd.)

Watershed Hydrologic Models

SWAT GWLF WAM

HSPF WEPP ANSWER

MIKE SHE AGNPS HEC-HMS

6

Calibration and validation of ArcSWAT

watershed model for the simulation of streamflow.

Performance evaluation of the model for

simulation of streamflow.

Objectives

7

22°11'51.65"N

8

• Pre Processing of Hydro-meteorological Data, Preparation of model specific weather input files (Daily Rainfall and streamflow data of 8 years- 1998-2005)

• Preparation of Spatial Data for the Study Area:

DEM, Land use/land cover map and Soil map

METHODOLOGY

9

Land use/Land cover map Soil map

10

Contour Map

11

DEM Sub watersheds and Stream reaches

No. of Sub watersheds =15

No. of HRUs =271

12

Calibration and validation of ArcSWAT model for the simulation of stream flow ArcSWAT model set up Sensitivity analysis for 18 streamflow parameters

(LH-OAT Sampling techniques), Calibration (SCEA-UA method) and Validation Performance evaluation of model for simulation of

streamflow at daily and monthly time steps using statistical and graphical indicators

METHODOLOGY (contd.)

13

Model Performance Statistics

Statistical Indicators

Mathematical Expression

RSR

PBIAS

NSE

RMSESTDEVobs

Moriasi et al. (2007)

( ) ( )

( )

1001

1

nQ Qo s ii

nQo ii

−∑

= ∑ =

( )n

o s ii 1n

o oi 1 i

Q QNSE 1

Q Q

=−

=

−= −

−

∑

∑

14

Performance Ratings of a Model for Monthly Time Step

Performance Rating RSR NSE

PBIAS (%) Streamflow Sediment

Very good 0.00 < RSR < 0.50

0.75 < NSE < 1.00

PBIAS < ±10 PBIAS < ±15

Good 0.50 < RSR < 0.60

0.65 < NSE < 0.75

±10 < PBIAS < ±15

±15 < PBIAS < ±30

Satisfactory 0.60 < RSR < 0.70

0.50 < NSE < 0.65

±15 < PBIAS < ±25

±30 < PBIAS < ±55

Unsatisfactory RSR > 0.70 NSE < 0.50 PBIAS > ±25 PBIAS > ±55

Moriasi et al. (2007)

15

RESULTS

Streamflow Simulation

16

Sensitivity Analysis Results for ArcSWAT Streamflow Parameters

Sl.No. Parameters Unit Mean

Sensitivity Index

Assigned Rank

1 Baseflow alpha factor (Alpha_Bf) days 1.4722 1 2 Manning coefficient for channel (CH_N) 1.3714 2 3 Surface runoff lag coefficient (Surlag) 0.2512 3

4 SCS curve number for moisture condition II (Cn2) 0.2141 4

5 Effective hydraulic conductivity in main channel (Ch_K2)

mm/hr 0.1877 5

6 Threshold depth of water in the shallow aquifer required for return flow to occur (Gwqmn)

mm 0.0921 6

7 Soil evaporation compensation factor (Esco) 0.0346 7

8 Groundwater delay (Gw_Delay) (days) 0.0249 8

17

Sensitivity Analysis Results (contd.)

Sl.No. Parameters

Unit Mean Sensitivity

Index

Assigned Rank

9 Soil conductivity (Sol_K) mm/h 0.0150 9

10 Available water capacity of the soil layer (Sol_Awc)

mm/mm 0.0110 10

11 Soil depth (Sol_Z) 0.0077 11

12 Groundwater ‘revap’ coefficient (Gw_Revap) 0.0070 12

13 Plant evaporation compensation factor (Epco) 0.0066 13

14 Threshold depth of water in the shallow aquifer for ‘revap’ to occur (Revapmin)

(mm) 0.0061 14

15 Maximum canopy index (Canmx) 0.0041 15 16 Soil albedo (Sol_Alb) 0.0000 17 17 Leaf area index for crop (Blai) 0.0000 17

Initial and Calibrated Parameter Values for Streamflow Simulation

Parameter Lower boundary Upper boundary Initial Value Calibrated Parameter value

Alpha_Bf 0 1 0.048 0.99947 Canmx 0 100 0 91.80400 Ch_K2 0.01 500 0.01 76.476000 Ch_N 0.01 0.3 0.014 0.01003 Cn2* -25 25 Soil /LULC data -17.18700 Epco 0 1 0 0.97161 Esco 0 1 0.95 0.02797

Gw_Delay 0 500 31 28.69900 Gw_Revap 0.02 0.2 0.02 0.03377

Gwqmn 0 5000 0 65.50400 Rchrg_Dp 0 1 0.05 0.00034 Revapmin 0 500 1 1.80870 Sol_Awc* -25 25 Soil data 20.47100

Sol_K* -25 25 Soil data 24.94900 Sol_Z* -25 25 Soil data 23.26100 Surlag 1 24 4 17.92800

19

ArcSWAT Model Performance Statistics for Simulation of Daily Streamflow

Statistical Indicators

Daily Streamflow Calibration Period

(1999-2003) Validation Period

(2004-2005) Observed Simulated Observed Simulated

Mean(m3/s) 33.41 32.82 34.34 34.90 STDEV(m3/s) 53.61 50.84 65.56 66.09

ME (m3/s) 0.59 -0.66 MAE (m3/s) 9.72 10.84

RMSE (m3/s) 20.10 25.93 PBIAS 1.14 -1.94 RSR 0.37 0.39 NSE 0.86 0.84 R2 0.86 0.85

20

ArcSWAT Model Performance Statistics for Simulation of Monthly Streamflow

Statistical Indicators

Monthly Streamflow Calibration Period

(1999-2003) Validation Period

(2004-2005) Observed Simulated Observed Simulated

Mean(mm) 49.07 48.52 50.04 51.01 STDEV(mm) 64.15 60.12 61.95 63.55

ME (mm) 0.56 -0.97 MAE (mm) 8.09 9.88

RMSE (mm) 13.52 15.07 PBIAS 1.14 -1.94 RSR 0.21 0.24 NSE 0.95 0.93 R2 0.96 0.94

21

Observed and SWAT Simulated Daily and Cumulative Streamflow Hydrographs

0

200

400

600

800

1000

1/1/

1999

1/1/

2000

1/1/

2001

1/1/

2002

1/1/

2003

Date

Stre

amflo

w (m

3 /s)

0

1000

2000

3000

4000

5000

6000

Cum

ulat

ive

Stre

amflo

w

(MC

M)

Observed (daily)Simulated (daily)Observed (cumulative)Simulated (cumulative)

(a) Calibration

0

200

400

600

800

1000

1/1/

2004

4/1/

2004

7/1/

2004

10/1

/200

4

1/1/

2005

4/1/

2005

7/1/

2005

10/1

/200

5

Date

Stre

amflo

w (m

3 /s)

0

500

1000

1500

2000

2500

Cum

ulat

ive

Stre

amflo

w

(MC

M)

Observed (daily)Simulated (daily)Observed (cumulative)Simulated (cumulative)

(b) Validation

Total Streamflow (MCM)

Obs.=5217.38

Sim.=5178.41

Total Streamflow (MCM)

Obs.=2134.57

Sim.=2176.05

22

Comparison of Observed and SWAT Simulated Daily Streamflows Hydrographs for Calibration Years

1999-2003

0

100

200

300

400

500

600

1/1 3/1 5/1 7/1 9/1 11/1 D t

Stre

amflo

w (m

3 /s) 0

20

40

60

80

100

Rai

nfal

l (m

m)

Mean RainfallObservedSimulated

(a) 1999

0

100

200

300

400

500

600

1/1 3/1 5/1 7/1 9/1 11/1 D t

Stre

amflo

w (m

3 /s) 0

20

40

60

80

100R

ainf

all (

mm

)

Mean RainfallObservedSimulated

(b) 2000

0

100

200

300

400

500

600

1/1 3/1 5/1 7/1 9/1 11/1 Date

Stre

amflo

w (m

3 /s) 0

20

40

60

80

100

Rai

nfal

l (m

m)

Mean RainfallObservedSimulated

(c) 2001

0

100

200

300

400

500

600

1/1 3/1 5/1 7/1 9/1 11/1 D t

Stre

amflo

w (m

3 /s) 0

20

40

60

80

100

Rai

nfal

l (m

m)

Mean RainfallObservedSimulated

(d) 2002

0

100

200

300

400

500

600

1/1 3/1 5/1 7/1 9/1 11/1 Date

Stre

amflo

w (m

3 /s) 0

20

40

60

80

100

Rai

nfal

l (m

m)

Mean RainfallObservedSimulated

(e) 2003

23

Comparison of Observed and SWAT Simulated Daily Streamflows Hydrographs for Validation Years

2004-2005

0

200

400

600

800

1000

1/1 3/1 5/1 7/1 9/1 11/1 Date

Stre

amflo

w (m

3 /s)

0

20

40

60

80

100

Rai

nfal

l (m

m)

Mean RainfallObservedSimulated

(a) 2004

0

200

400

600

800

1000

1/1 3/1 5/1 7/1 9/1 11/1 Date

Stre

amflo

w (m

3 /s) 0

20

40

60

80

100

Rai

nfal

l (m

m)

Mean RainfallObservedSimulated

(b) 2005

24

Observed and SWAT Simulated Monthly Streamflow Hydrographs

0

50

100

150

200

250

300

Jan-

99

Jul-9

9

Jan-

00

Jul-0

0

Jan-

01

Jul-0

1

Jan-

02

Jul-0

2

Jan-

03

Jul-0

3

Month

Mon

thly

Stre

amflo

w (m

m)

0

150

300

450

600

Mon

thly

Rai

nfal

l (m

m)

Mean Rainfall Observed Simulated

`

(a) Calibration

0

50

100

150

200

250

Jan-

04

Apr-0

4

Jul-0

4

Oct

-04

Jan-

05

Apr-0

5

Jul-0

5

Oct

-05

Month

Mon

thly

Stre

amflo

w (m

m)

0

150

300

450

600

Mon

thly

Rai

nfal

l (m

m)

Mean RainfallObservedSimulated

(b)Validation

25

Scatter Plots of Observed and Simulated Streamflow

R2 = 0.94

0

50

100

150

200

250

300

0 50 100 150 200 250 300Observed Monthly Streamflow

(mm)

Sim

ulat

ed M

onth

ly S

tream

flow

(m

m)

Data Points1:1 LineLinear

(b)Validation

R2 = 0.96

0

50

100

150

200

250

300

0 60 120 180 240 300

Sim

ulat

ed M

onth

ly S

tream

flow

(m

m)

Data Points1:1 LineLinear

(a)CalibrationR2 = 0.86

0

100

200

300

400

500

600

0 100 200 300 400 500 600

Sim

ulat

ed D

aily

Stre

amflo

w

(m3 /s

)

Data Points1:1 LineLinear

(a)Calibration

R2 = 0.85

0

200

400

600

800

1000

0 200 400 600 800 1000Observed Daily Streamflow (m3/s)

Sim

ulat

ed D

aily

Stre

amflo

w (m

3 /s)

Data Points1:1 LineLinear

(b) Validation

Daily Monthly

26

Bar Plots of Observed and Simulated Annual Streamflow

0

400

800

1200

1600

2000

1999 2000 2001 2002 2003 2004 2005

Year

Tot

al S

tream

flow

(MC

M) Observed

Simulated

27

Conclusion

• Alpha_Bf was found to be the most sensitive parameters followed by Ch_N, Surlag, Cn2, Ch_K2, Gwqmn, Esco, Gw_delay, Sol_k, Slope, Sol_Awc, Sol_Z, Gw_revap, Epco, Revapmin, Canmax, Slsubbsn, and Biomix.

• Values of PBIAS, NSE, and R2 during model

calibration and validation at daily and monthly time steps were found to vary from -1.94 to 1.76, 0.84 to 0.95, and 0.86 to 0.96 respectively.

Based on the analysis of the results obtained in this study, the following conclusions could be drawn

28

Conclusion (contd.)

• Lower values of PBIAS and RSR coupled with higher values of NSE and R2 indicated that the SWAT model simulated streamflow within accepatable level of accuracy.

• Diffrerent graphical techniques used to evaluate model

performance showed that there is a reasonably good agreement between observed and simulated daily and monthly streamflows as well as observed and simualted annual streamflow volumes for both calibration and validation period of model simulation.

• Overall, it can be concluded that the ArcSWAT model

simulated streamflow satisfactorily. Therefore, ArcSWAT model can be used for future studies on wtershed modeling in the Upper Baitarni river basin.

29

THANK YOU