DSPy a system for AI to Write Prompts and Do Fine Tuning
Basin Water Use Accounts
1. Basins water use accounting
Mac Kirby Judy Eastham Mohammed
Kirby, Eastham,
Mainuddin, Mark Thomas, CSIRO
Cali workshop, September 2007
2. Hydrology models
• Water resources
management, flood
forecasting, reservoir
(size) design (specialist Karkheh
hydrology models)
y gy ) Floods,
Floods etc
• Integrated catchment
management (eq WEAP)
• Policy - integrated systems Volta
Mekong
analysis (ie above plus
poverty, economics) Discharge
(water use accounts +
economics) Limpopo
MDB
• Teaching, understanding
(any, simpler better)
( i l b tt )
3. CAVEAT
• I don’t know much about many of the
don t
basins
• I seek input and advice -
– Do the results make sense?
– How accurate is the land use classification
(have irrigation areas been correctly defined)?
– What type of crops / cropping systems?
– Groundwater?
4. Input data
• Rain, Potential ET
• River flows
• Land use
• Catchment boundaries
• River network
Ri t k
• DEM sometimes
• Remote sensed data sometimes
(land use derived from)
• Irrigation water use - crop
coefficients,
coefficients calendar
• Rainfed water use, including
natural / forests - crop
coefficients, calendar
ffi i t l d
6. Karkheh, western
Iran - Flow
Observed and calculated flows
500
Observed and calculated flows Observed flow
400 Calculated flow
400
cm
Observed flow 300
Flow, m
Calculated flow
300
200
Flow, mcm
200 100
0
100
1 49 97 145
Months
0
1 49 97 145
Months
Observed and calculated flows
2000
Observed flow
Calculated flow Observed and calculated flows
1500
cm
4000
Flow, m
1000 3500 Observed flow
3000 Calculated flow
Flow, mc
500 2500
2000
0 1500
1 49 97 145 1000
Months 500
0
Observed and calculated flows 1 49 97 145
3000 Months
2500 Observed flow
Calculated flow
2000
Flow, mc
m
1500
1000
500
0
1 49 97 145
Months
7. Karkheh, dam,
development
• Karkheh dam 2000
• Changes flow to and
use in terminal marshes
• Depends on allocated
use
Discharge
g
3000 Discharge obs
discharge calc
2000
mcm/mo
1000
0
0 50 100 150
Months
8. Limpopo - land and
water use
• Rainfall ~250 - 1050 mm/yr
• Issues include:
–Poverty and food
df d
–Low productivity rainfed agriculture
–Irrigation
development
-ecosystem
degradation,
d d i
loss of
biodiversity
Internet data
13. Water use Sao
Francisco
0.30
Rainfall
0.25 ETo
Evaporation or Rainfall (m
m
0.20
0.15
0.10
0.05
0.00
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
15. Water use Niger
0.4
04
Rainfall
ETo
Evaporation or Rainfall (m)
0.3
0.2
0.1
0.0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
0.4
Rainfall
Evaporation or Rainfall (m)
ETo
0.3
0.2
0.1
0.0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
0.4
04
Rainfall
Evaporation or Rainfall (m)
ETo
0.3
0.2
0.1
0.0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
0.4
Rainfall
Evaporation or Rainfall (m)
ETo
0.3
0.2
0.1
0.0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
16. Water use Nile 0.35
Rainfall
0.30 ETo
Evaporation or Rainfal
0.25
0.20
0.15
0.10
0.05
0.00
Jan Feb M Apr M Jun Jul Aug Sep O N Dec
ar ay ct ov
45000
Calculated
40000
Observed
35000
30000
Flow
25000
20000
15000
10000
0.35
Rainfall
5000 0.30 ETo
al
Evaporation or Rainfa
0.25
0 0.20
1951 1956 1961 1966 1971 1976 1981 1986 1991 1996 0.15
0.10
0.05
0.00
Jan Feb M Apr M Jun Jul Aug Sep O N Dec
ar ay ct ov
0.35
Rainfall
0.30 ETo
Evaporation or Rainfal
0.25
0.20
0.15
0.10
0.05
0.00
Jan Feb M Apr M Jun Jul Aug Sep O N Dec
ar ay ct ov
22. Comparing Basins 2
Water use
Karkheh
Floods,
Floods etc
Volta
Mekong
M k
Discharge
g
rainfed
agriculture
grassland
MDB
woodland + other
irrigated
agriculture Limpopo
net runoff
23. Comparing basins 1
hydrology
1.2
supply lim
ppy it
capacity limit
1
0.8
08
Karkheh
ET/ETo
0.6 Indus
M
Floods, etc
ekong
Volta Ganges
0.4 Yellow Sao Francisco
Limpopo N iger
Volta
Nile
0.2 Volta Mekong
g
Karkheh
0
Discharge
0 0.5 1 1.5
P/ETo MDB
Limpopo
24. Comparing Basins 2
1.2
supply limit
capacity limit
1
Water use0.8
ET/ETo
Karkheh
0.6 Indus
Volta Mekong
Ganges
Yellow Sao Francisco
0.4
04
Limpopo N iger
Floods,
Floods etc
N ile
0.2
Karkheh
0
0 0.5 1 1.5
P/ETo
Volta
Mekong
M k
Discharge
g
rainfed
agriculture
grassland
MDB
woodland + other
irrigated
agriculture Limpopo
net runoff
27. Water Stress Index distribution
(Irrigation only) on the Indus &
Ganges rivers
Indus River
Ganges River
1.0
Severe > 0 4
0.4
Water Stress Index ΣA/Q
Medium 0.2-0.4 0.8
Low < 0.2
0.6
0.4
r
0.2
0.0
1.0 0.8 0.6 0.4 0.2 0.0
Source Relative distance from estuary Estuary
28. Current Water Stress distribution
on the Indus River
CURRENT
1.0
Severe > 0.4
mulativ e Wate r
Medium 0.2-0.4 0.8
Low < 0.2
Stress Index
0.6
Irrigation 0.4
Domestic
Cum
S
Industrial 0.2
0.0
1.0 0.8 0.6 0.4 0.2 0.0
Source Relative distance from ocean Estuary
29. Current & future water use
trends in response to population
go t
growth in t e Indus Basin
the dus as
Current 2025
2% 4%
5%
12%
84%
93% Irrigation
Domestic
Industrial
30. Projected Water Stress
distribution on the Indus River
(2025 population)
( 0 5 popu at o )
2025
1.4
Severe > 0.4 Irrigation 1.2
Medium 0 2 0 4
0.2-0.4
mualtiv Wate r
Domestic
D ti
Low < 0.2 1.0
Stress Index
Industrial
08
0.8
ve
0.6
Cum
0.4
S
0.2
0.0
00
1.0 0.8 0.6 0.4 0.2 0.0
Estuary
Source Relative distance from ocean
31. Comparing current and
projected Water Stress
distribution on the Indus River
1.4
14
1.2
W a te r S tre s s In d e x
Severe > 0.4
(a ll w a te r u s e s )
Medium 0.2-0.4 1.0
10
Low < 0.2
0.8
0.6
06
t
Current 0.4
2025 0.2
02
0.0
1.0
10 0.8
08 0.6
06 0.4
04 0.2
02 0.0
00
Relative distance from ocean
Source Estuary
32. Comparing water stress of current
and future population (2025) in the
Indus Basin
300
Current
250
2025
ple
Millio peop
200
150
on
100
50
0
Low Medium Severe
Degree of water stress
33. Water stress indices
seasonal effects Percentage of the population of the Indus Basin
living under varying degress of water stress:
Current population and climate
• Whole Indus
M edium
High
80%
Low
70%
60%
50%
40%
30%
20%
10%
0%
Percentage of the population of the Indus Basin
Jan Feb M Apr M
ar ay Jun Jul Aug Sep O N Dec
ct ov
living under varying degress of water stress: 2025
population and future climate
M edium
80% High
70% Low
60%
50%
40%
30%
20%
10%
0%
Jan Feb M Apr M
ar ay Jun Jul Aug Sep O N Dec
ct ov
34. Other stress indices
• Dams - storage volume as a fraction of annual flow
– Lake Volta ~ 3 x annual flow
– Karkheh ~ 1 x annual flow
– Mekong ~ 0 05 x annual flow
0.05
• Drought indices - recurrence
• Linked to population
• I di to be used with care - th M
Indices t b d ith the Murray-D li
Murray-Darling
would show as very stressed on most indices, but
we don’t propose to send international aid to
don t
relieve poverty
35. Conclusions -
tentative thoughts
• Want local input on data - land use, etc.
• Can do water use accounts for many basins - but
data availability varies
• Basins vary greatly in water availability, variability,
l l bl bl
management
• B i likely vary greatly in vulnerability to change
Basins lik l tl i l bilit t h
36. Conclusions 2
Basins compared
• Water use accounts give a foundation for comparing
basins
• CPWF basins show range from dry to wet
• Range of water opportunities and constraints
R f t t iti d t i t
• Implications for transportability of results across
basins of similar hydrology type?
• Foundation for economic analyses, developing other
models eg WEAP
• Role for analogous studies on food and poverty?