1

IMPROVING THE SNOWMELT MODELLING IN MESOSCALE HYDROLOGIC MODEL (mHM) with SATELLITE BASED DYNAMICALLY CALCULATED

DEGREE-DAY FACTOR

Birkan ERGUC Zuhal AKYUREK

Birkan ERGUC

November 2019

Reading, UK

AGENDA

• Motivation

• Objective

• Study Areas

• Methodology

• Results

• Discussion and Future Works

2METU - GGIT

MOTIVATION

• Fresh water sources are under stress

• Better management for mountain watersheds

• More realistic modelling

• Snow-related dynamics dominant in mountains

• Need for more realistic snow modelling

3METU - GGIT

Taurus Mountains, TurkeyEncyclopædia Britannica

OBJECTIVE

• Achieve More Realistic Snow Modelling:

• Degree-day factor

• Based on satellite imagery

• Spatio-temporally dynamic

• Test results

• mHM hydrologic model

• Karanlıkdere basin

• Bahcelik basin

4METU - GGIT

Word Art https://wordart.com/create

1 / 1 23.11.2019, 16:04

STUDY AREA: Karanlikdere• Karanlıkdere Basin

• Mountain basin

• 1295 - 2332m elevation

• ~30 km2 area

5METU - GGIT

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Monthly Average Flow (m3/s)

STUDY AREA: Bahcelik• Bahcelik Basin

• Mountain basin

• 1392 - 2705m elevation

• ~2500 km2 area

6METU - GGIT

0

2

4

6

8

10

12

Monthly Average Flow (m3/s)

METHODOLOGY

• Degree-day Factor Conventionally:

• Simple computation

• Reduced data need

• Relatively good performance

• Generally accepted constant in basin

• Spatially and/or Temporally

• Followed Methodology1 (He et al., 2014):

• Daily average temperature

• Daily precipitation

• MODIS Snow cover data

7METU - GGIT

1 He, Z. H., et al. "Estimating degree-day factors from MODIS for snowmelt runoff modeling." Hydrology and Earth System Sciences18.12 (2014): 4773-4789.

METHODOLOGY (Cont.)

8METU - GGIT

𝑉𝑠 = 𝑆𝐷 ∙ 𝑆𝐶𝐴 (1)

VS : Volume per area of snow (mm3 / mm2)

SD : Snow depth (mm)

SCA: Snow-covered area (%)

𝑝𝑠 .𝑑𝑉𝑠

𝑑𝑡= ቐ

𝑃, 𝑇 < 𝑇𝑆 𝐴𝑐𝑐𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑜𝑛𝑃𝑠 −𝑀, 𝑇𝑆 ≤ 𝑇 ≤ 𝑇𝑅 𝐴𝑐𝑐𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑜𝑛 + 𝐴𝑏𝑙𝑎𝑡𝑖𝑜𝑛

−𝐷𝐷𝐹 𝑇 − 𝑇𝑚 , 𝑇 > 𝑇𝑅 𝐴𝑏𝑙𝑎𝑡𝑖𝑜𝑛(2)

𝑝𝑠 : Snow density (%)

VS : Volume per area of snow (mm3 / mm2)

P : Daily precipitation (mm)

Ps : Daily snowfall(mm)

M : Daily snowmelt depth (mm)

Ts : Temperature threshold for snowfall (oC)

TR : Temperature threshold for rain (liquid) (oC)

Tm : Temperature threshold for melting (oC)

METHODOLOGY (Cont.)

9METU - GGIT

• Preparing MODIS Data

• Pixel based approach

METHODOLOGY (Cont.)

10METU - GGIT

• MODIS data process: Karanlikdere Basin

METHODOLOGY (Cont.)

11METU - GGIT

• MODIS data process: Bahcelik Basin

METHODOLOGY (Cont.)

12METU - GGIT

• Daily Degree-day Factor Calculation:

• At days of ablation

• TR = 2,5 oC

• TS = TM = 0 oC

• SCA = From MODIS data (daily)

• 𝑝𝑠 = Month Snow Density (%)

December 0,25

January 0,25

February 0,27

March 0,30

𝑝𝑠 .𝑑𝑉𝑠𝑑𝑡

= −𝐷𝐷𝐹 𝑇 − 𝑇𝑚 , 𝑇 > 𝑇𝑅

−𝐷𝐷𝐹 𝑇(𝑡) − 0𝑝𝑠 . 𝑉𝑠 𝑡 − 𝑉𝑠(t − 1) =

𝐷𝐷𝐹(𝑡) = −𝑝𝑠 . 𝑉𝑠 𝑡 − 𝑉𝑠(t − 1)

𝑇(𝑡)Daily

METHODOLOGY (Cont.)

• Mesoscale Hydrologic Model (mHM)

• 102 – 104 km2 basins

• Distributed model

• Tested on 400+ basin

• Utilize sub-grid variability

• Multi-parameter Regionalization

• Use data at finest scale available

• Link to global parameters

• Max. 53 calibration parameters

• Open source

13METU - GGIT

METHODOLOGY (Cont.)

14METU - GGIT

Data Type Data Source

Meteorological Data State Meteorological Service (MGM) Joint Research Center (JRC)

Hydrometric Data State Hydraulic Works (DSI)

Elevation Data General Directorate of Mapping (HGM)

Land Use Data Ministry of Agriculture and Forestry

Geological Data DG Mineral Research and Exploration (MTA)

Leaf-Area_Index Data MODIS Land Cover Type Product

Soil Data Harmonized World Soil Database

Satellite Data MODIS 500m resolution Snow Products

• Data Sources:

METHODOLOGY (Cont.)

• mHM Calibration

• Modelling grid: 1 x 1 km

• Sub-grid: 100 x 100 m

• Meteo-grid: 25 x 25 km

• Dynamically Dimensioned Search (DDS) algorithm

• 7500 iteration

• Objective Function:

• Simulation period:

• Karanlikdere Basin: 1.12.2014 - 30.09.2017 – Warm up: 180 days

• Bahcelik Basin: 1.10.2003 – 30.04.2017 – Warm-up : 365 days

15METU - GGIT

min𝑍 = 1 −1

2𝑁𝑆𝐸 − ln 𝑁𝑆𝐸

Z : Objective Function

NSE : Nash-Sutcliffe efficiency

RESULTS

• Daily Degree-day factors for Karanlikdere Basin

16METU - GGIT

-15

-10

-5

0

5

10

15

20

25

30

10/1

/13

11/1

/13

12/1

/13

1/1/

142/

1/14

3/1/

144

/1/1

45/

1/14

6/1

/14

7/1/

148

/1/1

49

/1/1

410

/1/1

411

/1/1

412

/1/1

41/

1/15

2/1/

153/

1/15

4/1

/15

5/1/

156

/1/1

57/

1/15

8/1

/15

9/1

/15

10/1

/15

11/1

/15

12/1

/15

1/1/

162/

1/16

3/1/

164

/1/1

65/

1/16

6/1

/16

7/1/

168

/1/1

69

/1/1

610

/1/1

611

/1/1

612

/1/1

61/

1/17

2/1/

173/

1/17

4/1

/17

5/1/

176

/1/1

77/

1/17

8/1

/17

9/1

/17

10/1

/17

11/1

/17

12/1

/17

1/1/

182/

1/18

3/1/

184

/1/1

85/

1/18

6/1

/18

7/1/

188

/1/1

89

/1/1

8

0

1

2

3

4

5

6

7

8

9

10

Sn

ow

dep

th (

dm

)T

emp

erat

ure

(C

)

Date (days)

DD

F (

mm

/day

/C)

DDF

snowdepth (dm) DDF Avg. DDF tavg

Avg. calculated DDF = 2.431 | mHM calibrated DDF (forest): 2.395

RESULTS (Cont.)

• Daily Degree-day factors for Bahcelik Basin

17METU - GGIT

-30

-10

10

30

50

700

0.5

1

1.5

2

10/1

/02

5/1/

03

12/1

/03

7/1/

04

2/1/

05

9/1

/05

4/1

/06

11/1

/06

6/1

/07

1/1/

08

8/1

/08

3/1/

09

10/1

/09

5/1/

10

12/1

/10

7/1/

11

2/1/

12

9/1

/12

4/1

/13

11/1

/13

6/1

/14

1/1/

15

8/1

/15

3/1/

16

10/1

/16

5/1/

17

12/1

/17

7/1/

18

Sn

ow

dep

th (

cm)

Avg

.Tem

p(C

)

Deg

ree-

Day

Fac

tor

(mm

/day

/C

DDF at Grid 64200

DDF Snowdepth(cm) Avg.DDF Avg.Temp

0

1

2

3

4

5

6

7

01-

01

01-

11

01-

21

01-

31

02-

10

02-

20

03-

01

03-

11

03-

21

03-

31

04

-10

04

-20

04

-30

05-

10

05-

20

05-

30

06

-09

06

-19

06

-29

07-

09

07-

19

07-

29

08

-08

08

-18

08

-28

09

-07

09

-17

09

-27

10-0

7

10-1

7

10-2

7

11-0

6

11-1

6

11-2

6

12-0

6

12-1

6

12-2

6

Avg

. D

egre

e-D

ay F

acto

r

Average DDF for 2002 – 2018 (at Grid 64200)

Avg. calculated DDF = 0.2826 | mHM calibrated DDF (pervious): 0.1237

RESULTS (Cont.)

• Flow simulations for Karanlikdere Basin

18METU - GGIT

DDF Calibrated mHM NSE: 0,45785 KGE: 0.292mHM NSE: 0.45078 KGE: 0.281

-16

-6

4

14

24

34

44

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

12

/1/1

4

1/1

/15

2/1

/15

3/1

/15

4/1

/15

5/1

/15

6/1

/15

7/1

/15

8/1

/15

9/1

/15

10

/1/1

5

11

/1/1

5

12

/1/1

5

1/1

/16

2/1

/16

3/1

/16

4/1

/16

5/1

/16

6/1

/16

7/1

/16

8/1

/16

9/1

/16

10

/1/1

6

11

/1/1

6

12

/1/1

6

1/1

/17

2/1

/17

3/1

/17

4/1

/17

5/1

/17

6/1

/17

7/1

/17

8/1

/17

9/1

/17

Pre

cip

itat

ion

(m

m),

Avg

.Tem

p.

(C)

Q (

m3

/s)

Flow Results for DDF-fed mHM Model - Bahcelik Basin

Precipitation Qobs mHM DGF Fed DGF calibrated Avg.Temp.

RESULTS (Cont.)

• Flow simulations for Bahcelik Basin

19METU - GGIT

-20

-10

0

10

20

30

40

50

600

10

20

30

40

50

60

200

310

01

200

3122

920

04

032

720

04

06

2420

04

09

2120

04

1219

200

5031

820

050

615

200

509

1220

051

210

200

60

309

200

60

60

620

06

09

03

200

612

01

200

7022

820

070

528

200

708

2520

071

122

200

80

219

200

80

518

200

80

815

200

811

1220

09

020

920

09

050

920

09

08

06

200

911

03

2010

013

120

100

430

2010

072

820

1010

2520

110

122

2011

04

2120

110

719

2011

1016

2012

011

320

120

411

2012

070

920

1210

06

2013

010

320

130

40

220

130

630

2013

09

2720

1312

2520

140

324

2014

06

2120

140

918

2014

1216

2015

031

520

150

612

2015

09

09

2015

120

720

160

305

2016

06

02

2016

08

3020

1611

2720

170

224

Avg

. P

reci

pit

atio

n (

mm

), A

vg.T

emp

erat

ure

(C)

Q (

m3/

s)

Flow Results for DDF-fed mHM Model - Bahcelik Basin

Avg. Precipitation Qobs mHM DDF Calibrated Avg. Temp

DDF Calibrated mHM NSE: 0.07003 KGE: 0.21543mHM NSE: 0.11149 KGE: 0.25381

RESULTS (Cont.)

• Snowpack Simulations: Karanlikdere Basin

20METU - GGIT

-16

-6

4

14

24

34

44

0

50

100

150

200

250

300

350

400

450

500

12/1

/14

1/1/

15

2/1/

15

3/1/

15

4/1

/15

5/1/

15

6/1

/15

7/1/

15

8/1

/15

9/1

/15

10/1

/15

11/1

/15

12/1

/15

1/1/

16

2/1/

16

3/1/

16

4/1

/16

5/1/

16

6/1

/16

7/1/

16

8/1

/16

9/1

/16

10/1

/16

11/1

/16

12/1

/16

1/1/

17

2/1/

17

3/1/

17

4/1

/17

5/1/

17

6/1

/17

7/1/

17

8/1

/17

9/1

/17

Pre

cip

itat

ion

(m

m),

Avg

. T

emp

. (C

)

Sn

ow

pac

k (

mm

)

Observed vs Simulated SWE (mm)

Precipitation Obs. SWE (mm) mHM DDF Fed Avg. Temp.

RESULTS (Cont.)

• Snowpack Simulations: Bahcelik Basin

21METU - GGIT

-20

-10

0

10

20

30

40

500

20

40

60

80

100

120

140

160

180

200

10/2

/03

2/2/

04

6/2

/04

10/2

/04

2/2/

05

6/2

/05

10/2

/05

2/2/

06

6/2

/06

10/2

/06

2/2/

07

6/2

/07

10/2

/07

2/2/

08

6/2

/08

10/2

/08

2/2/

09

6/2

/09

10/2

/09

2/2/

10

6/2

/10

10/2

/10

2/2/

11

6/2

/11

10/2

/11

2/2/

12

6/2

/12

10/2

/12

2/2/

13

6/2

/13

10/2

/13

2/2/

14

6/2

/14

10/2

/14

2/2/

15

6/2

/15

10/2

/15

2/2/

16

6/2

/16

10/2

/16

2/2/

17

Avg

.Tem

p.(

C),

Pre

cip

itat

ion

(m

m)

Sn

ow

pac

k (

mm

)

Observed vs Simulated SWE

Precipitation AgroGrid SWE (mm) DDF Calibrated mHM Calibrated Avg.Temp.

RESULTS (Cont.)

22METU - GGIT

• Bahcelik Basin Simulated SWE vs MODIS Imagery

DISCUSSION and FUTURE WORKS

• Results

• Slight differences in terms of flow simulation

• Improvement on SWE simulations

• Consistency between average calculated DDF and mHM

calibrated DDF

• Error Sources

• Data quality (meteorological and flow data)

• Relatively short period of availability

• Size of basin for Karanlikdere

• Future Works

• Add spatial snow distribution as calibration parameter to mHM

• Improve meteorological data with ground-based observations

• Investigate other remote sensing snow data

23METU - GGIT

24

Thank You.

Kackar Mountains, Turkey