1. Green and Blue Water
a model & data based analysis of
water scarcity, productivity (and trade)
with a focus on the CPWF basins
Holger Hoff1,2, Dieter Gerten1 , Jens Heinke1
1:Potsdam Institute for Climate Impact Research
2: Stockholm Environment Institute & Resilience Centre

2. The LPJmL eco-hydrological & crop model
biophysical processes at plant level
CO2 Temperature Precipitation Radiation
e.g. crop Interception
production, Photosynthesis
biofuels, carbon  Transpiration
sequestration Carbon /
Water Balance
Water resources
Crop yields
Evaporation
Surface flow
Subsurface flow
consistent simulation of water resources,
plant water use and productivity
.

3. The LPJmL eco-hydrological & crop model
landscape level
livestock
household
precipitation partitioning into green & blue
industry
conveyance losses
withdrawals evaporation
interception lakes
irrigation
reservoirs
transpiration
Irrigation….. figure
demand von Stefanie
return flows
evaporation
surface runoff
plant water supply
subsurface runoff
consistent calculation of green & blue virtual water contents
(crops and other ecosystem services) AND water availability

4. The LPJmL eco-hydrological & crop model
reservoirs
Biemans et al 2011

5. current blue water availability per capita
m3 cap-1 yr-1
adding green water (evapotranspiration from cropland)

6. current green & blue water availability per capita
m3 cap-1 yr-1
discussion: country-wise aggregation of water for food
potential for expanding cropland ?

7. current crop water productivity (kcal m-3)
kcal m-3
Sri Lanka 2129 kcal m-3
India 1648
Bangladesh 2697
Pakistan 1218
water productivity -> weighted water availability

8. maximum calorie production (kcal cap-1)
agricultural water productivity
for current water availability & productivity
kcal cap-1 day-1

9. CPWF basins
current green & blue water availability per capita
6000
5000
4000
3000
m3 cap-1 year-1
blue water
green wate
2000
1000
0
Ganges Indus Limpopo Mekong Niger Nile Sao Francisco Volta
Ganges Indus Limpopo Mekong Niger Nile Sao Yellow
Volta Yellow
Francisco
.

10. CPWF basins
current crop water productivity (kcal m-3)
2500
2000
1500 factor 4
kcal m-3
1000
500
0
Ganges Indus Limpopo Mekong Niger Nile Sao Volta Yellow
Francisco
discussion: potential for closing the yield gap ? .
.

11. CPWF basins
maximum crop production (kcal cap-1)
for current water availability & productivity
12000
10000
8000
kcal cap-1 day-1
6000
4000
3000
2000
0
po
o
lta
w
s
g
s
er
ile
sc
du
ge
on
l lo
ig
po
Vo
N
ci
an
In
ek
Ye
N
m
an
G
M
Li
Fr
o
3000 a tipping point? Sa
30 year averages .

12. CPWF basins
inter-annual variability of green & blue water availability
LPJmL vs CPWF data (1951-2000)
0.5
0.45
0.4
0.35
0.3
CP blue
CoV
0.25 LPJ blue
0.2 CP green
0.15 LPJ green
0.1
0.05
0
Ganges Indus Limpopo Mekong Niger Nile Sao Volta Yellow
Francisco
3000 a tipping point?
.

13. CPWF basins
factoring in variability
14000
12000
10000
kcal cap-1 day-1
8000
6000
4000
3000
2000
0
po
o
lta
w
s
g
s
er
ile
sc
du
ge
on
l lo
ig
po
Vo
N
ci
an
In
ek
Ye
N
m
an
G
M
Li
Fr
o
3000 a tipping point? Sa
using 10th percentile instead of mean green & blue water availability
. .

14. CPWF basins
the future (2050)
14000
12000
10000
8000
kcal cap-1 day-1
6000
4000
3000
2000
0
po
o
lta
w
s
g
s
er
ile
sc
du
ge
on
l lo
ig
po
Vo
N
ci
an
In
ek
Ye
N
m
an
G
M
Li
Fr
o
Sa
discussion: 1) future crop water productivity (incl. CO2 effect)?
2) how much of the additional pressure is due to climate change?
. .

15. future water demand
3,5
population change only
3
– relative to year 2000 Ganges
2,5 Indus
Limpopo
2 Mekong
Niger
1,5 Nile
Sao Francisco
1 Volta
Yellow
0,5
0
1990 2000 2010 2020 2030 2040 2050 2060
discussion: future populations and diets ?
.

16. future (blue) water availability
1,2
1 Ganges
Indus
0,8 Limpopo
Mekong
0,6 Niger
Nile
0,4 Sao Francisco
climate change only Volta
0,2 – relative to year 2000 Yellow
0
1990 2000 2010 2020 2030 2040 2050 2060
.
– 30 year averages, not accounting for changing variability, monsoon etc
.

17. next steps: simulating impacts of variability,
e.g. monsoon changes
.
Bondeau pers comm

18. next steps: simulating land use change effects
via “moisture recycling”
Nikoli et al
.
Indus „precipitationshed“
.

19. next steps: simulating land use change effects
via “moisture recycling” (external driver)
Nikoli et al
80
70
60
50
% internally generated
40
precipitation
30
20
internal
10
terrestrial
0
ao Francisco
Niger
Volta
Nile
Limpopo
internalprecipitation
Indus
basin
Yellow
Mekong
Ganges
originating from
terrestrial ET
LPJmL simulations of ET changes with land use change

20. next steps: (“real”) water footprints
of food production (and trade)
plus other ecosystem services
e.g. carbon sequestration
%

21. LPJ-based crop virtual water contents
plus ComTrade data
-> virtual water im- / exports
consistent with water availability (“footprints”)
Sri Lanka 100m3 cap-1 net import
Cyprus
India 5m3 cap-1 net export
Bangladesh 30 m3 cap-1 net import
Pakistan 2 m3 cap-1 net export
discussion: what happens under increasing future water scarcity?
see MENA example
.

22. LPJ-based crop virtual water contents
plus ComTrade data
-> virtual water im- / exports
consistent with water availability (“footprints”)
1200 correlations of per capita
net VW imports and
VW imports (m**3) per capita &
1000 - blue water availability:
Cyprus
- 0.51
- blue plus green water avail: - 0.64
800 - water-limited potential
kcal production: - 0.79
600
year
400
200
0
0 500 1000 1500 2000 2500 3000 3500
water-limited potential kcal production per capita & day
.

23. 1) Effects of international trade on local water resources
a) in water scarce (MENA) countries
m3 cap-1 yr-1
.