uncertainty on hydrological models in climate change scenarios
Post on 10-Jan-2016
37 Views
Preview:
DESCRIPTION
TRANSCRIPT
1
Uncertainty on Hydrological Models in Climate Change Scenarios
S. Jamshid Mousavi
Associate Professor, Civil Engineering Department, Amirkabir
University of Technology (Polytechnic of Tehran), Tehran, Iran
jmosavi@aut.ac.ir
Regional Asian G-WADI Workshop,
June 2011, Tehran, Iran
Outline• Steps in climate change impact studies
• Uncertainty in climate change studies
• Hydrological models and uncertainty sources
• Uncertainty-based calibration/simulation of
hydrological models
• Successive uncertainty fitting (SUFI) approach
• Uncertainty-based calibration of HEC-HMS
model: Tamar Basin experience
• Uncertainty-based calibration of SWAT model: Karkheh River Basin experience
• Climate-change-driven runoff simulation and water allocation at basin scale: Karkheh River Basin experience
2
Downscaling
Calibrated hydrological model
Management and water allocation model
Emission scenarios
GCM outputs
Precipitation, temperature, etc
1-Select a few number of emission
scenarios
2-Take GCMs outputs of metrological
variables
3-Downscale the output of the GCMs
4-Build a calibrated hydrological model
of the basin
5-Simulate hydrologic variables of
interest (runoff) subject to downscaled
climate-change-driven inputs
6-Extend the study chain to water
management system
Typical steps of climate change impact studies on water resources
3
CC-driven simulated runoff scenarios
Downscaled precipitation, temperature, etc
4
Uncertainty on future emission of green gasses: Emission
scenarios
Uncertainty on nature of ocean-atmosphere physical processes
and so structure of GCMs
Uncertainty related to quantification of regional and local effects
reflected in downscaling models
Uncertainty on hydrological models
Uncertainty on behavior of complex socio-economic and
management systems
Uncertainty Issue
5
Approaches to deal with uncertaintyScenario generation-based techniques
Techniques based on probabilistic representation of processes involved (e.g. Reliability Ensemble Averaging [Giorgi and Mearns 2002], Bayesian-based Multi-model Ensembles of GCMs [Tebaldi et al. 2005], …)
6
Uncertainty on Hydrological ModelsSources of uncertainty:
1- Structural uncertainty
2- Data uncertainty
3- Parameter uncertainty Input processing Output
known Unknown
Precipitation Watershed Runoff
Hydrologic Modeling
77
CALIBRATIONModel parameters are determined through model calibration, because parameters may not have exact physical meaning or may not be easily measurable
Manual Calibration: Trial and error-based parameter estimation procedure Time consuming Depends on expertise and judgment of the modeler Does not account for uncertainty
Automatic Calibration: Systematic search procedures for finding parameter values The search procedure is guided based on an objective function
measuring how well any set of parameters perform
16
8
CALIBRATION PROCEDURE
Rainfall-precipitation data
Rainfall-precipitation data
startstart Parameter EstimationParameter Estimation
Simulation in HEC-HMSSimulation in HEC-HMS
Compare HydrographCompare Hydrograph
Error?
EndEnd
Better Estimation of
parameters
Better Estimation of
parameters
Yes
NO
17
9
Difficulties with (Automatic) Calibration TechniquesModel uncertainty because of imperfect model structure due to either parameters
interdependence or conceptual simplifications of physical processes taken place in real natural systems
Input uncertainty due to erroneous or approximate input data (Extending point rain
data to areal data)
Parameter uncertainty and nonuniquness due to non-identifiability feature
a) different parameter sets may not give rise to different model outputs resulting in parameter nonuniquness in inverse-type problems
b) model response surface can be insensitive to a number of parameters in the region of optimum solution
Multiple convergence regions and local optima solutions
Nonconvex shape of the response surface with discontinuous long curved ridges 9
10
Quality of Calibration Results Depend on:The conceptual base and structure of the CRR model
The modeling technique addressing different types of uncertainty depending on quality and quantity of information and data available
The power and robustness of the optimization algorithm
Performance criteria or objective function
10
11
Uncertainty-based Calibration of Hydrological Models
Automatic calibration techniques:
1- Monte-Carlo2- Generalized likelihood uncertainty estimator (GLUE)3- Bayesian techniques4- Parasol5- Successive uncertainty fitting (SUFI)
12
SUFI TechniqueExtensively used in calibration of SWAT hydrological
model at continent, regional and basin scales
13
HMS-SUFI: Uncertainty-based Calibration Technique Step 1. Select an objective function (RMSE in the current application)
More importance is given to some desired discharges like the peak flow
Step 2. Define physically meaningful absolute minimum and maximum ranges for each parameter
Step 3. Carry out a sensitivity analysis with respect to parameters
Step 4. Assign an initial uncertainty range to each parameter used in the first round of Latin hypercube. These ranges are smaller than the absolute ranges
Step 5. Generate n different sets of parameter values using Latin Hypercube sampling technique
Step 6. Evaluate the objective function, g, for each of n generated sets (Run HEC-HMS n times)
Step 7: Calculate the sensitivity analysis matrix J, Hessian Matrix H, Covariance matrix C, and parameter correlations matrix r
Step 8: Calculate uncertainty indicators P-factor, R-factor,.., and new parameter ranges based on best parameter set and matrices C and …
2 2
2
1
*( )(1)i observed simulated
n
ii
C Q QRMSE
C
13
Conceptual basis of SUFI Uncertainty AnalysisInput parameters:
(Uniform distribution)Output variables:
(95% PPU)
14
95PPU:
Uncertainty Bound as 95% of probability distribution
85
-jan
85
-jan
85
-jan
85
-jan
85
-jan
85
-jan
86
-jan
86
-jan
86
-jan
86
-jan
86
-jan
86
-jan
87
-jan
87
-jan
87
-jan
87
-jan
87
-jan
87
-jan
88
-jan
88
-jan
88
-jan
88
-jan
88
-jan
88
-jan
89
-jan
89
-jan
89
-jan
89
-jan
89
-jan
89
-jan
0
200
400
600
800
1000
1200
1400
1600
Month
Riv
er
dis
ch
arg
e(m
3 s
-1)
Jelogir Station
R-factor : Measures total predictive uncertainty in terms of normalized sum of 95% PPU bounds
R-factor : Measures model predictive uncertainty
P-factor: Percent of observed data (runoff) locating within 95PPU bounds of simulated ones
br2
r2 :Correlation coefficient between measured and simulated valuesb: Slope of regression line
Uncertainty measures
0
50
100
150
200
250
300
350
Month
Riv
er
dis
ch
arg
e(m
3 s
-1
)
Ghurbaghestan StationP-factor: 0.78R-factor: 0.91
15
Uncertainty Analysis using SUFI
16
16
HMS-SUFI MODELCase Study:Tamar Subbasin
One of the basins of Gorganrood River located in the North west of Khorasan province in Iran.
Area of Gorganrood River 52826 square kilometers
Area of Tamar basin about 1530 square kilometers
• Considered important because of experiencing severe floods .
16
17
Gorgan-Roud Basin
1717
1818
Case Study-Tamar Subbasin Lack of sufficient data at hydrometric and rainfall stations makes modeling the basin’s
response to floods challenging
4 reliable flood events were available the first 3 of which were used for calibration and the last one for verification
18
1919
Observed Flood Events
Event Date Peak Flow (m^3.s) Duration(hr)
1 19Sep2004 128 20
2 6May2005 299 30
3 9Aug2005 783 19
4 8Oct2005 120 13
Date, peak flow and duration of flood events
19
2020
Hydrograph of first event0 5 10 15 20 25 30 35 40 450
5
10
15
20
25
30
35
40
Rai
n(m
m)
0 5 10 15 20 25 30 35 40 450
20
40
60
80
100
120
140
Event 1
Time(hr)
Q(M
3 /s)
Time 0 equals to (19Sep2004, 18:00)
20
2121
Hydrograph of Second event
Time 0 equals to (06May2005, 01:00)
0 5 10 15 20 25 30 35 40 450
5
10
15
20
25
30
35
40
Rai
n(m
m)
0 5 10 15 20 25 30 35 40 450
50
100
150
200
250
300
350
event 2
Time(hr)
Q(c
ms)
21
2222
Hydrograph of Second event0 5 10 15 20 25 30 35 40 450
5
10
15
20
25
30
35
40
Rai
n(m
m)
0 5 10 15 20 25 30 35 40 450
100
200
300
400
500
600
700
800
event3
Time(hr)
Q(c
ms)
Time 0 equals to (09Aug2005, 20:00)
22
2323
Hydrograph of Second event0 5 10 15 20 25 30 35 40 450
5
10
15
20
25
30
Rai
n(m
m)
0 5 10 15 20 25 30 35 40 450
20
40
60
80
100
120
140
event 4
Time(hr)
Q(c
ms)
Time 0 equals to (08Oct2005 ,
23
24
HEC-HMS STRUCTURE
HEC-HMS Components
Basin module Meteorological module Control Specification
Loss module Transformation module
Base flow module
Routing module
248
25
BASIN MODULE
25
Loss Method Routing (for Reach ) Base Flow Method Transform Method
Deficient and Constant Wave kinematics Bounded Recession Clark Unit hydrograph
Exponential Lag Constant Monthly Mod Clark
Green-Ampt Muskingum Linear Reservoir Kinematic Wave
Gridded Deficient and Constant
Muskingum-Cunge Nonlinear Boussinesq SCS Unit Hydrograph
SCS Curve number Modified Puls recession Snyder Unit hydrograph
Gridded SCS Curve Number Straddle Stagger Specified Unit hydrograph
Initial and ConstantUnit hydrograph S
hydrograph
Soil moisture Accounting
Smith Parlange
Gridded Soil moisture Accounting
25
BASIN MODULE
26
26
Model of Tamar Basin in HEC-HMS
Divided into 7 sub-basin with 3 routing reaches
SCS-CN Method Estimating hydrologic losses
Clark Method Transforming rainfall to runoff
Flood routing in reaches Muskingum method.
No base flow
26
27
27
Tamar Basin in HEC-HMS
27
2828
Calibration Parameters Loss method: SCS-CN Method
1- Curve number: 7 parameters (1 for each subbasin)2- Initial abstraction: 7 parameters (1 for each subbasin)
Transformation Method: Clark 1-Time of concentration: estimated by SCS synthetic unit hydrograph approach2- Storage Coefficient (R):
Routing Method: Muskingum (K,X)
Total No. of Calibration Parameters=2428
29
Manually-calibrated parameter values for each event and their upper and lower limits (suggested by IWRI (2008)
Upper limit
Lower limit
Event-1 Event-2 Event-3 Event-4 Upper
limit
Lower
limit
Event-1
Event-2
Event-3 Event-4
Curve number
Subbasin1 91 60 66.1 86.3 80.6 91
Constant
Subbasin1 0.65 0.2 0.34 0.47 0.53 0.28
Subbasin2 91 61 63.2 85.5 76.3 91 Subbasin2 0.65 0.2 0.50 0.15 0.52 0.58
Subbasin3 87 58 80.1 82.1 70.2 87 Subbasin3 0.65 0.2 0.53 0.53 0.63 0.49
Subbasin4 85 60 78.1 64.85 60.6 85 Subbasin4 0.65 0.2 0.25 0.49 0.61 0.64
Subbasin5 84 50 51 84 84 82.9 Subbasin5 0.65 0.2 0.25 0.32 0.64 0.58
Subbasin6 91 70 86.3 80.2 79.9 91 Subbasin6 0.65 0.2 0.43 0.53 0.63 0.47
Subbasin7 91 70 77.2 72.1 80.0 91 Subbasin7 0.65 0.2 0.43 0.51 0.64 0.61
Initial abstraction
Subbasin1 0.25S 0.15S 0.24S 0.16S 0.22S 0.15S
Muskingum
X
Reach 1 0.5 0.2 0.47 0.48 0.49 0.42
Subbasin2 0.25S 0.15S 0.18S 0.2S 0.20S 0.24S
Subbasin3 0.25S 0.15S 0.17S 0.24S 0.22S 0.13S Reach 2 0.5 0.2 0.21 0.41 0.40 0.42
Subbasin4 0.25S 0.15S 0.25S 0.22S 0.21S 0.24S
Subbasin5 0.25S 0.15S 0.21S 0.23S 0.20S 0.21S Reach 3 0.5 0.2 0.25 0.46 0.47 0.48
Subbasin6 0.25S 0.15S 0.19S 0.24S 0.23S 0.16S
Subbasin7 0.25S 0.15S 0.22S 0.13S 0.19S 0.17S
29
30
Results and AnalysisSingle-event Calibration Scenario-Event 1
30
31
Results and AnalysisSingle-event Calibration Scenarios-Events 2 and 3
32
Results and AnalysisJointly-calibrated events: Scenario 0
33
Results and AnalysisJointly-calibrated events scenario with bigger weights assigned to Event
1 in the Obj. Function (Scenario 0W)
33
34
Results and AnalysisJointly-calibrated events with 31 parameters and increased Ia values of event-1 to 0.45
(Scenario 1)
34
35
Results and AnalysisJointly-calibrated events with 31 Parameters and decreased Ia values of event-3
to zero (Scenario 2)
35
36
Jointly-calibrated events with 31 Parameters and adjusted Ia values of event-1 and event-3 (Scenario 3)
we were keen to see if there is another set of parameters with Ia values closer to initially-set bounds. The 31-parameter problem was run with Ia lower bounds of event-1 as 0.35 (instead of 0.45 in scenario 6) and those of other events (1 and 2) as 0.05 (instead of zero in scenario 7).
36
37
Results and AnalysisNonuniqueness of Parameter Sets
The only difference between scenarios 1-3 is in their Ia lower and upper bounds
What about the other parameter values? Were they same?
Answer: NoNonuniqueness
37
38
Verification Analysis Simulating Event 4 by 9 sets of Parameter Values Obtained in Calibration
Stage
Before recalibration of Ia After recalibration of Ia
38
39
Simulated objective function and Ia values of all recalibrated parameters (Screening Step)
Parameter sets
Event-1 Event-2 Event-3 Mean Events
Scenario 0 Scenario 0W
Scenario 1 Scenario 2
Scenario 3
Simulated RMSE 21.6431 22.8237 21.4624 38.0544 21.4112 21.6312 21.5938 21.5938 24.6156
Sub-basin’s 1 Ia coefficient 0.4005 0.2989 0.5201 0.3008 0.4567 0.2938 0.4070 0.95 0.2205
Sub-basin’s 2 Ia coefficient 0.3579 0.2586 0.6050 0.3013 0.4545 0.2941 0.4227 0.95 0.3092
Sub-basin’s 3 Ia coefficient 0.3380 0.2894 0.4394 0.2631 0.4584 0.2100 0.4538 0.9500 0.2309
Sub-basin’s 4 Ia coefficient 0.4052 0.2747 0.5936 0.3141 0.3902 0.3200 0.4263 0.9500 0.3570
Sub-basin’s 5 Ia coefficient 0.3563 0.3628 0.4826 0.2964 0.3606 0.2710 0.3968 0.0054 0.3581
Sub-basin’s 6 Ia coefficient 0.3846 0.3113 0.4698 0.2875 0.3474 0.2312 0.2258 0.6586 0.2564
Sub-basin’s 7 Ia coefficient 0.3155 0.3906 0.4749 0.2913 0.4426 0.1979 0.1796 0.8876 0.1246
Min Ia 0.3155 0.2586 0.4749 0.2631 0.3474 0.1979 0.1796 0.0054 0.1246
Max Ia 0.4052 0.3906 0.6050 0.3141 0.4584 0.3200 0.4538 0.95 0.3581
40
Comparison of parameter values, other than Ia s, of the 6 screened parameter sets (See nonuniquness)
Parameter values
Event-1 Event-2 Scenario-0 Scenario-0W Scenario 1 Scenario 3
CN1 65.97 78.69 82.57 77.70 83.60 78.10
CN2 84.34 72.56 83.59 82.72 78.76 76.31
CN3 76.03 67.82 67.28 60.91 68.66 58.30
CN4 60.05 60.84 62.11 60.01 61.09 60.96
CN5 58.39 79.92 77.11 63.42 72.94 71.47
CN6 88.88 76.43 74.35 76.37 73.76 70.19
CN7 83.61 75.07 72.34 70.32 79.52 71.04
SC1 0.5199 0.4171 0.2785 0.2232 0.2879 0.2355
SC2 0.2984 0.5614 0.4742 0.4091 0.3182 0.3575
SC3 0.3095 0.2776 0.5406 0.5968 0.4700 0.4476
SC4 0.5987 0.5660 0.5961 0.5988 0.2696 0.4446
SC5 0.4968 0.2005 0.2640 0.2302 0.3039 0.3102
SC6 0.5213 0.3188 0.5529 0.2593 0.5114 0.5380
SC7 0.4201 0.3144 0.3512 0.2559 0.3877 0.2502
X1 0.4464 0.3245 0.3226 0.2012 0.2381 0.4232
X2 0.2540 0.4333 0.4564 0.3781 0.3563 0.3831
X3 0.4063 0.3692 0.3345 0.3912 0.3186 0.4711
41
Dealing with Uncertainty Uncertainty measures and the best obj. function obtained from another SUFI run with the new parameter bounds derived from 6 screened parameter sets : event 4
(values within parenthesis are for a smoothed hydrograph) Iteration 1 2 3 4 5 6 7 8 9 10
P-factor% 89.47 (94.7
4)
84.21 (78.9
5)
73.68 (84
.21)
68.42 (68.
42)
47.3 (68
.42)
31.58 (47.
37)
36.84 (42.1
1)
26.32 (36.8
4)
15.79 (26.
32)
15.78 (26.
32)
R-factor 1.6559 (1.65
38)
1.0795 (0.81
3)
0.8849 (0.
4887)
0.7457 (0.3
653)
0.5234 (0.
3430)
0.2096 (0.2
267)
0.1767 (0.16
76)
0.1296 (0.11
40)
0.1129 (0.1
110)
0.0704 (0.0
837)
Par-factor
0.2047 (0.20
47)
0.1285 (0.10
98)
0.0783 (0.
0741)
0.0571 (0.0
566)
0.0395 (0.
0433)
0.0289 (0.0
319)
0.0227 (0.02
25)
0.0149 (0.01
54)
0.0117 (0.0
125)
0.0085 (0.0
090)
Best sol’s RMSE
14.45 (9.17
06)
13.62 (9.20
93)
13.74 (8.
7700)
13.36 (7.6
496)
13.22 (7.
4637)
13.05 (7.2
927)
12.78 (7.13
73)
12.69 (7.09
89)
12.62 (7.0
065)
12.56 (6.9
803)
42
Parameter Bounds and Simulated Discharges in iterations 1 and 3 of SUFI with
Newly-Selected Parameter Bounds:Event-4
42
43
Measuring Input Parameter Uncertainty
)17(2/)minmax(
)minmax(
1
1
m
jjj
m
jjj
bb
bb
factorPar
43
45
Iteration’s 4 simulated discharges of the events in SUFI for jointly –calibrated eventswith fixed parameter intervals except initial abstraction ranges to find de-calibrated Ia values
45
46
ConclusionsThe proposed 3-stage procedure
1- Obtain different parameter values by calibrating events either separately or jointly to come up with candidate parameter sets entering in verification stage
2- Recalibrate the parameters reflecting initial basin conditions and then screen out the above candidate parameter sets based on how well they perform in verification and also how physically meaningful their recalibrated parameters are
3- Calculate the new parameter ranges from the screened parameter values and re-run the SUFI model to find narrower parameter ranges, as the final parameter ranges, at which uncertainty indicators are still acceptable. Also check if the final ranges perform well in simulating all calibration events (Backward step). This needs again re-calibration of initial-basin-condition parameters with respect to calibration events 46
Karkheh River Basin Study
47
1- Climate Change Impact on Surface Runoff
2- Climate Impact on Water Allocation
Downscaling
Calibrated hydrological model
Management and water allocation model
Emission scenarios
GCM outputs
Precipitation, temperature, etc
1-Select a few number of emission
scenarios
2-Take GCMs outputs of metrological
variables
3-Downscale the output of the GCMs
4-Build a calibrated hydrological model
of the basin
5-Simulate hydrologic variables of
interest (runoff) subject to downscaled
climate-change-driven inputs
6-Extend the study chain to water
management system
Typical steps of climate change impact studies on water resources
48
CC-driven simulated runoff scenarios
Downscaled precipitation, temperature, etc
÷One of the most important basins in
Iran in terms of Surface and
groundwater resources, agriculture
potential, hydropower generation,…
16000 MCM of potential storage
capacity 40% percent of which has
been constructed
49
Karkheh River Basin
Area: 4763592 Hectar
50
Karkheh River Basin
Tangmashoureh
Garshaگ
Seymareh
Koranbuzan
Sazbon
ROR Karkeh
Karkheh
51
Location of Dams in Karkheh
Basin
ScenarioA2
B1A1B
Emission Scenarios
A1
World’s Rapid but Uniform Economy Grow with
Tendency to Increasing Consumption
International Cooperation toward Clean Technologies
and Environment ProtectionBalanced Utilization of Fossil and No-fossil Fuels as Energy
Resources
Nonuniform Rapid Economy Grow with Variable Regional
Technologies along with Increasing Consumption
52
Downscaling
CGCM 3.1 T & P outputs were related to basin’s
meteorological stations data from 1980-2002 to estimate
Near (2010-2040) and Far (2070-2100) future downscaled
outputs of the emission scenarios
Calibration Period: 13 yearsValidation Period: 10 Years
T:
: P
53
1 2 3 4 5 6 7 8 9 10 11 120
2
4
6
8
10
12
14Hamedan. Station
A1B (2070-2100)
A1B (2010-2040)
A2 (2070-2100)
A2 (2010-2040)
B1 (2070-2100)
B1 (2010-2040)
Obs. (1982-2002)
Month
Ave
rage
Pre
cipi
tati
on (m
m)
1 2 3 4 5 6 7 8 9 10 11 120
1
2
3
4
5
6
7
8
9
10Ahwaz Station
A1B (2070-2100)
A1B (2010-2040)
A2 (2070-2100)
A2 (2010-2040)
B1 (2070-2100)
B1 (2010-2040)
Obs. (1982-2002)
Month
Ave
rage
Pre
cipi
tati
on (m
m)
54
Downscaled/Historical Precipitation at Hamedan and Ahwaz Stations
1 2 3 4 5 6 7 8 9 10 11 120
10
20
30
40
50
60
Hamed fo. Station
A1B (2070-2100) A1B (2010-2040) A2 (2070-2100) A2 (2010-2040)
B1 (2070-2100) B1 (2010-2040) Obs. (1982-2002)
Month
Max
. Tem
pera
ture
(ْC)
1 2 3 4 5 6 7 8 9 10 11 120
10
20
30
40
50
60 Ahwaz Station
Month
Max
. Tem
pera
ture
(ْC
)
55
Downscaled/Historical Temprature at Hamedan and Ahwaz Stations
SWATSoil & water Assessment Tool
56
.
Calibration (1990-2002)
br2: 0.46P-factor: 71%R-factor: 0.83
Validation (1980-1990)
br2: 0.42P-factor: 65%R-factor: 0.74
Vlidation Calibration
90
-jan
90
-jan
91
-jan
91
-jan
92
-jan
92
-jan
93
-jan
94
-jan
94
-jan
95
-jan
95
-jan
96
-jan
97
-jan
97
-jan
98
-jan
98
-jan
99
-jan
99
-jan
00
-jan
01
-jan
01
-jan
02
-jan
02
-jan
0
50
100
150
200
250
300
350
400
450
Series1
Observed
MonthR
ive
r d
isc
ha
rge
(m3
s-1
)
Watershed area: 2781 km 2
𝜙: 0.46P-factor: 0.75R-factor: 0...84
95
-jan
95
-jan
95
-jan
95
-jan
96
-jan
96
-jan
96
-jan
96
-jan
97
-jan
97
-jan
97
-jan
97
-jan
98
-jan
98
-jan
98
-jan
98
-jan
99
-jan
99
-jan
99
-jan
99
-jan
0
50
100
150
200
250
300
350
400
450
95PPU
Observed
Month
Riv
er
dis
ch
arg
e(m
3 s
-1)
Watershed area: 2781 km 2
𝜙: 0.40P-factor: 0.77R-factor: 0.72
Polchehr Station 8 Hydrometric Stations used
Sample Results of the Model Calibration Using SUFI
59
1980-2002
2070-2100 2010-2040
Max. Temperature (C)
61
Precipitation (mm)
1980-2002
2070-2100 2010-2040
62
Runoff (m3)1982-2002
2070-2100 2010-2040
63
Garsha Seymare Tang-mashure Krkhe0
500
1000
1500
2000
2500
3000
3500
4000
4500
A1B (2073-2099)A1B (2013-2039)A2 (2073-2099)A2 (2013-2039)B1 (2073-2099)B1 (2013-2039)Historic (1985-2002)
Dam
Wat
er a
vail
able
(M
CM
)
Average Annual Runoff @ Grasha, Seymareh and Karkheh Dams Basins
30%9%B1 (2013-2039)A2 (2013-2039)
A2 (2073-2099)
B1 (2073-2099)46%
10%
64
4163 MCM
1590 MCM
Jan Feb Mar Apr May Jun Jul Aug Sep Nov Oct Dec0
100
200
300
400
500
600
700
800
A1B (2073-2099) A1B (2013-2039) A2 (2073-2099) A2 (2013-2039)
B1 (2073-2099) B1 (2013-2039) Historic (1985-2002)
Month
Ava
ilab
le W
ater
(M
CM
)
A1B (2073-2099)A1B (2013-2099)
Monthly Distribution of Runoff @Karheh Dam Site
65
66
Summary Results of CC-Driven Runoff Scenarios
67
Scenario
Region
B1
( 2013-2039)
B1
( 2073-2099)
A2
( 2013-2039)
A2
( 2073-2099)
A1B (2013-2039)
A1B
( 2073-2099)
13 10.4 31.8 45.7 22.3 35.42 North
Diff. Discharge (%)-30 -32 -18.5 -9.3 -25.18 -17 South
6.09 6.4 10.93 19.7 8.54 12.18 North
Diff. Precipitation (%)-97.91 -97.9 -97.84 -97.58 -97.84 -97 South
1.5 2.51 1.74 4.3 2.7 2.51 North Diff. Max. Temperature
(0C)0.98 1.00 0.99 1.047 1.03 0.98 South
Summary Results of CC-Driven Runoff ScenariosHistoric (1985-2002)
Region
85.49 NorthDischarge (m3/s)
132 South
409 NorthPrecipitation (mm)549.3 South
22.2 NorthMax. Temperature (0C)31 South
6.16 NorthMin. Temperature (0C)14.47 South
2. Climate Change Impact Assement on water Allocation
68
MODSIM
River Basin Management Decision Support
69
S275762 ha
24.05
PS1
PS21645 ha
11270 ha21993 ha
6000 ha
24000 ha
14000 ha
11000 ha
20%
15%
35000 ha
15%
S1611787 ha
outlet18907 ha
68000 ha
45333 ha
AMC
2750 ha30%
30%
85 15%
510
100
70
30%
30%
30%
30%
100
5019 ha
0.12
10%
10%
93
20%
35000 ha
3950 ha
13000 ha
14660 ha
60000 ha
2000 ha
S1611787 ha
15%
AMC
5019 ha
1645 ha
)(
:
:
:
5
6000 ha
-
-
- -
-
20 Agricultural nodes
Type No. Demand Type
Prioاrity
1 Domestic 1
2 Environment 2
3 Agriculture 3
4 Hydropower 4
: Target Reservoir storage: 5
Schematic of Karkheh Water Resource System
7 Reservoirs
71
Dams and Powerplants Characteristics
73
A1B (2073-2099)
A2 (2073-2099)
B1 (2073-2099)
Historic (1985-2002)0
5
10
15
20
25
Potation
Scenario
Flo
w (
MC
M)
A1B (2073-2099)
A1B (2013-2039)
A2 (2073-2099)
A2 (2013-2039)
B1 (2073-2099)
B1 (2013-2039)
Historic (1985-2002)0
1000
2000
3000
4000
5000Enviromental
Scenario
Flo
w (
MC
M)
A1B (2073-2099)
A1B (2013-2039)
A2 (2073-2099)
A2 (2013-2039)
B1 (2073-2099)
B1 (2013-2039)
Historic (1985-2002)0
1000
2000
3000
4000
5000
Agriculture
Scenario
Flo
w (
MC
M)
Annual Allocated Water to Different Demand types ???
Environment:
A1B (2073-2099) 5%
B1 (2013-2039) 11%
Agriculture:
A1B (2073-2099) 22%
A2 (2073-2099) 76%
Domestic:
A1B (2073-2099) 3%
B1 (2013-2039) 7%
A2 (2073-2099) 7%
75
16.5
1836 2196
Agriculture Enviromental Potation0
10
20
30
40
50
60
70
80
A1B (2073-2099) A1B (2013-2039) A2 (2073-2099) A2 (2013-2039)
B1 (2073-2099) B1 (2013-2039) Historic (1985-2002)
Demand
Per
cen
tage
of
sup
ply
(%
)
Long-term (Annual) Percentage of Meeting Demands
76
1 2 3 4 5 6 7 8 9 10 11 120
102030405060708090
100
Agriculture
A1B (2073-2099) A1B (2013-2039) A2 (2073-2099) A2 (2013-2039)
B1 (2073-2099) B1 (2013-2039) Historic (1985-2002)
Month
Per
cen
tsge
of
sup
ply
(%
)
Significant decrease in meeting Agr. Demand except in Autumn Season
Monthly Percentage of Meeting Agricultural Demand
77
0 10 20 30 40 50 60 70 80 90 1000
50
100
150
200
250
Exceedance Probability (%)
Ene
rgy
(GW
H)
A1B (2073-2099)A1B (2013-2039)
A2 (2073-2099)
A2 (2013-2039)
B1 (2073-2099)
B1 (2013-2039)Historic (1985-2002)
Total MonthlyEnergy Duration Curve of the System
(GWh)
30Firm EnergyA1B (2073-2099)
78
1597
1278
958
639
319.5
Ene
rgy
(GW
H)
asd
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
Scenario
En
ergy
(G
WH
)
Annual Total Energy Generated
(GWh)
A1B (2073-2099): 26%
B1 (2073-2099): 7%
79
3195
A1B (2073-2099)
A1B (2013-2039)
A2 (2073-2099)
A2 (2013-2039)
B1 (2073-2099)
B1 (2013-2039)
Historic (1985-2002)0
30
60
90
120
150
180Karkhe Jarayani
Scenario
Ene
rgy
(GW
H)
A1B (2073-2099)
A2 (2073-2099)
B1 (2073-2099)
Historic (1985-2002)0
200
400
600
800
1000
1200
Seymare
Scenario
Ene
rgy
(GW
H)
733
Annual Energy Generated at two Sites (GWh)
ROR Karkheh
Seymareh
80
531
1 2 3 4 5 6 7 8 9 10 11 120
100
200
300
400
500
600
700
A1B (2073-2099) A1B (2013-2039) A2 (2073-2099) A2 (2013-2039) B1 (2073-2099)
B1 (2013-2039) Historic (1985-2002)
Month
En
ergy
(G
WH
)Monthly Distribution of Total Energy Generated (GWh)
Increase in Fall and Winter Months
Decrease in Spring and Summer Months
81
82
Historic
(1985-2002)
B1
(2013-2039)
B1
(2073-2099)
A2
(2013-2039)
A2
(2073-2099)
A1B
(2013-2039)
A1B
(2073-2099)
3529.3
(GWh) 0 -3 4 14 1 14
Difference Percentage of Total Energy Generation
Compared to Historical Scenario
83
Thank You for Your Attention
top related