Wuletawu Abera Ph.D. defense

Modelling Water Budget at basin scale Using JGrass-

NewAge system

April 28, 2016, Trento University

PhD candidate: Wuletawu Abera Worku

Farm

er P

ray

for

rain

, in

afa

r re

gio

n

Supervisor: Prof. Riccardo Rigon

Friday, 29 April 16

Global Water scarcity

Introduction: water emergency

- Water for food (agriculture) - Water for Energy - Ecosystem

4 billion people faces water scarcity

Mekonnen and Hoekstra Sci. Adv. 2016

Wuletawu AberaFriday, 29 April 16

curse of hydroclimatic variability

intraannual (e.g. seasonal and monthly)

interannual (year-to-year)

unpredictable timing and intensity of extremes

Introduction: Hydroclimatic variability

+/-+/-

-

- Climate change

Aggravate:

Wuletawu Abera WorkuFriday, 29 April 16

Coping strategies

2. Infrastructure e.g.- desalination - Transfer canal and pipes - Storage dams

3. Hydrological Information

1. Institutions (legal and policy)

Hall et al, 2015, Pedro-Monzonís et al, 201

3I

Introduction: coping strategy

California Aqueduct


- Hydrometeorological data

Hydrological information

- Hydrological models

Introduction: the challenge

Hydrological systems is highly variable in space and time


6

ext

Introduction: Grand challenge

Wuletawu Abera Worku

Modern day societal demand of hydrological information

Friday, 29 April 16

7

ext




Friday, 29 April 16

8

ext




Friday, 29 April 16

9

ext




Friday, 29 April 16

Not just discharge:➡Precipitation➡Evapotranspiration➡Storage



Friday, 29 April 16

11

At each HRU and links



Friday, 29 April 16

Water Budget!

- Spatially and temporally continues WB closure

- Develop methods and procedures

Te Precipi =rainfall + snow melting

Te Runoff contribution from upstream

Te Evapotranspiration

Te Runoff loss to downstream

Introduction: Water budget modelling

Jk(t) +

m(k)X

i

Qki(t)� ET k(t)�Qk(t) =@Sk(t)

@t

Te Storage


The Modelling Framwork: JGrass-NewAge system

Luca


Formettal et al, 2014

Friday, 29 April 16

-small basin (116km2)

-snow dominated, alpine

-topographically complex

-gauged (12 meteo, 3 hydrometers)

-1994-2012 simulation

Posina basin Upper Blue Nile (UBN) basin

-176000km2

-Topographically, sociopolitically complex

-Data scarce

1994-2009

The Study basins


Water Budget! Basin partitioning procedures and connection

Formetta et al. 2015

The uDig Spatial Toolbox for hydro-geomorphic analysis

Wuletawu Abera, Andrea Antonello, Silvia Franceschi, Giuseppe Formetta, Riccardo Rigon

Book chapter, British Society for Geomorphology, 2014

Jk(t) +

m(k)X

i

Qki(t)� ET k(t)�Qk(t) =@Sk(t)

@t


16

HRU-channel partition

42 HRUs402 HRU

Water budget closure for each HRUs


Water Budget!

17

Geostatistics (Kriging):

•Experimental semivariogram

•Theoretical semivariogram: Exp, Gau, Sph, Lin

•Kriging estimates: OK, LOK, DK, LDK

Optimal VGM model and parameters are fitted to each time steps

4 kriging X 4 VGM model = 16 data sets

Improving spatial field of input from meteo data: Posina basin

Jk(t)� ET k(t)�Qk(t) =@Sk(t)

@t


Water Budget!

18

based on discharge

Formetta et al. 2014; Hall et al.,2006; Li et al., 2012; Mou et al., 2008; He et al., 2014

MOD10A1 and MYD10A1

Modelling the input component: Posina Basin

Of the total J(t), how much are rainfall and snowfall ?


19

The effect of semivariogram model is minimal

The difference is between Local & universal

Cross validation analysis


20

Results: Space-time Precipitation of each HRU + errors


Water Budget!

21

��

��

��

��

��

��

��

��,��,�� ,��,�� ,��,��

�

��

��

��

��

��

��

�

��

��

��

��

��

��

�

��

��

��

��

��

��

��

��

�

!�� "#��$�"�%%&��'�

(��)��*�

��

��

��

��

��

��

��+ ��+ ��+ ��+ ��+ ��+

�

�

�

�

�

�

Results: Snowfall and rainfall separation

r=0.6AI=60%



@t

Friday, 29 April 16

Water Budget!

22


Results: Snowfall and rainfall separation

Estimating water budgets at the basin scale with JGrass-NewAge system, Part I: water

inputs, their variability and uncertainty

Wuletawu Abera, Giuseppe Formetta, Marco Borga, Riccardo Rigon

submitted, HESS, 2016

Friday, 29 April 16

Water Budget!

23

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●●●

●●

●

●●

●

●

●

●●

●●

●

●

8

9

10

11

12

13

36 38 40Long

Lat

0

1000

2000

3000

4000Altitude (m)

Gauge stations are very scarce • 1 station/5000km^2 • Poor maintenance

Option: Satellite Rainfall Estimation products

Spatial J(t) info using Kriging is elusive t

Improving rainfall spatial field when in-situ observation is absent: UBN basin


@t


Water Budget!

24

GOF between SREs and in-situ data

- correlation -RMSE-BIAS

Five High temporal and spatial SREs1. CMORPH2.TRMM3.TAMSAT4.SM2R-CCI5.CFSR

Which satellite product ?


Knoche et al. 2014

Friday, 29 April 16

Water Budget!

25

Ecdf Matching

�

�

��

��

��

��

��

�

� ��

� ��

Panofsky and Brier, 1986; Michelangeli et al. 2009

BIAS correction


Water Budget!

26

Results: SREs GOF comparison


Water Budget!

27

��

��

��

�

�

��

��

��

��

��

�

��

��

��

��

��

��

��

�

�

��

��

��

��

�

��

��

��

��

��

��

��

��

�

�

��

��

��

��

�

��

��

��

��

��!�

��!��

��"�

Results: Spatial distribution of GOF


Water Budget!

28

Results: Annual rainfall volume difference

��

�

��

��

��

� ��

��

��

��

�� !"

��#��$

%��%

�&�'(

��)�� *��+)),

��-��*�-��

��.�� .��

~2700

~1600

~600

~2000


29

Comparative evaluation of different satellite rainfall estimation products and bias correction

in Upper Blue Nile (UBN) basin

Wuletawu Abera, Luca Brocca, Riccardo Rigon

Journal of atmospheric Research, 2016

Results: BIAS correction procedure improvement


Water Budget!

30

HYMOD

T Van Delft et al. 2009 ext

Improving output components: Rainfall-runoff Model


@t

Calibration of model parameters: Particle swarm; Luca


Water Budget!

31

Storage information from HYMOD Text

Rn is modulated using SWRB and LWRB components (Formetta et al.2013)

Improving output components: ET


@t

P r i e s t l e y -

T a y l o r ( P T )

formulation

ET (t) = ↵C(t)

Cmax

�(t)

�(t) + �R

n

The problem is how to estimate/calibrate ext


Water Budget!

32

3. Use of GRACE

2. Budyko assumption

s(t)� s(0) =

Z T

0J(t)�Q(t)� ↵AET (t)ds

s(TB)� s(0) = 0

This becomes measured

ds(t)

dt= J(t)�Q(t)� ↵ET (t)

The only unknown

1.Literature e.g Cristea et al. 2012, (0.6 to 2.4); Problem?

Three options

↵(TB) =

R TB0 J(t)�

R TB0 Q(t)dt

R TB0 ET (t)dt


Water Budget!

33

Results: ET component for Posina basin.....

Alpha-estimation


Water Budget! Results: ET component for Posina basin.....

ET-hourly


Water Budget!

35

Results: Q component for Posina basin.....

�

��

��

��

� ��

��

��

��

��

��

� �

� �

� �

! �

��

� ��

��

��

��

��

��

�

�

��

��

��

��

� ��

��

��

��

��

��

�

�

�

�

� ��

��

��

��

��

� �

� �

� "

� �

� ��

��

��

��

��

�

�

�

�

"

� ��

��

��

��

��

#��$%�"

#��$%&

#��$%��

#��$%�!

#��$%�

#��$%�

CalKGE=0.71

ValKGE=0.63

KGE=0.73

KGE=0.62


36

��

��

��

��

��

��

��

��

��

Results: Spatial and temporal dynamics of water balance for Posina basin


@t


37

�

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

� ��

��

��

�

��

�

�

��

��

��

��

��

Results: Basin scale monthly water budget for Posina basin

Water budget forecast based on only Precip

for 2012


38

�

��

��

��

��

��

��

�

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

�

!"

#

$�%�� %�� !"��#��

&'(� � )��'��

Results: Basin scale annual water budget for Posina basin

Annual WB (1994-2011)

Annual variability of J

Q follows J

ET is less variable

J: 1730 +/- 344 Q 76.5%ET 30%

ds/dt -4.5%

Estimating water budgets at the basin scale with JGrass-NewAge system, Part II: water

outputs, and Storage components

Wuletawu Abera, Giuseppe Formetta, Marco Borga, Riccardo Rigon

submitted, HESS, 2016


Water Budget!

39

��

Water budget when data is scarce: UBN basin

Independent data/estimation/satellite are used to verify the

results:

Q: split time series data/internal sites

ET: MODIS MOD16 satellite data

ds/dt: GRACE satellite


Water Budget!

40

� ��

��

�

�

�

�

��

�

�

�

�

��

��

��

��

� ��

�

! �

Results: Output component (ET)

MOD16 underestimate


@t

Yilmaz et al., 2014; Knipper et al., 2016; Ramoelo et al., 2014


Water Budget!

41

�

�

��

��

��

��

��

��

��

��

��

��

� �

��

��

�

��

��

Results: Output component (ET)

GOF between MOD16 and NewAge ET


Water Budget!

42

Results: Output component (Q)

KGE=0.93 KGE=0.91KGE=0.55 KGE=0.81

KGE=0.38

KGE=0.58KGE=0.72KGE=0.55


Water Budget!

43

��

�

��

��

��

��

��

��

��

��

��!!�"� #��$%$�&�

Results: Output component (ds/dt)


@t

-seasonality+

-amplitude


Water Budget!

44

�

�

�

��

��

��

�

�

��

��

��

�

�

��

��

��

�

�

��

��

��

��

��

��

�

��

��

��

��

��

��

��

�

��

��

��

��

��

!"��

#$

�

��

��

Results: Water budget for UBN basin


Water Budget!

45

�

��

��

��

��

��

�

��

�

��

��

��

� !

"#$"%

&�%

�'�

��

��

�

#$

!

&

Results: Mean monthly basin scale Water budget closure


Water Budget!

46

�

�

�

��

��

��

��

��

��

� ��

�

�

��

��

��

��

��

��

� ��

�

�

��

��

��

��

��

��

��

��

��

� �

�

Results: Water budget long term annual mean

Spatially distributed annual water budget


Water Budget!

47

�

��

��

��

��

��

��

�� !��

�

Results: Mean annual basin scale Water budget closure

Water budget modelling of Upper Blue Nile basin using JGrass-NewAge model system and

Satellite data

Wuletawu Abera, Giuseppe Formetta, Luca Brocca, Riccardo Rigon

To be submitted soon, HESS


Water Budget!

48

- New simplified methods of separating rainfall from snowfall using MODIS data - Evaluation of different SREs data is very crucial, and the difference between could be as high as 2700 mm per year. Procedures are presented.

- The Adige (HYMOD) component is effectively calibrated in different basin (Posina and UBN), and the performance ranges from very high to acceptable.

-The Budyko assumption used to reformulate PT ET model, with less uncertainty

- Effectively employed different Remote sensing data for water budget closure

Conclusions and contributions...


Water Budget!

49

Finally, but most importantly, I am grateful for ....


Water Budget!

50

TextTextT

Thank you for your attention

[email protected]


mailto:[email protected]

mailto:[email protected]

Wuletawu Abera Ph.D. defense

Education