transcript
- Slide 1
- CALVIN: Optimization of Californias Water Supplies 29 November
2001 10:00am 4:00 pm 744 P St., Sacramento, CA; Auditorium CALVIN
is an optimization model which suggests operations of Californias
inter-tied water supply system which would maximize statewide
economic benefits to agricultural and urban water users within
environmental flow requirements. Morning sessions will focus on
technical aspects, with the afternoon sessions on results, policy,
and management implications. Project descriptions, details, and
results can be found at:
http://cee.engr.ucdavis.edu/faculty/lund/CALVIN/ Tentative Workshop
Agenda 10:00Overview of CALVIN Model (Jay Lund) 10:30Economic
Valuation of Water Uses (Richard Howitt, Mimi Jenkins) 11:00
Managing Data and Model Outputs (Ken Kirby) Model engineering and
economic outputs and how they are used Data Management and
Databases 11:30Model Calibration (Mimi Jenkins) 12:00Limited
Foresight and Carryover Storage (Andy Draper) 12:30Lunch
1:30Limitations of Model and Data (Jay Lund) 1:45Results and
Implications (Various speakers) Alternatives Examined (Randy
Ritzema) Delivery & Economic Performance with Optimized
Operations (Randy Ritzema) Willingness-to-Pay for Additional Water
(Richard Howitt) Water Transfers and Exchanges (Richard Howitt)
2:30Break 2:45Results and Implications (continued) Economic Values
for Facility Expansion (Stacy Tanaka) Economic Costs of
Environmental Flows (Stacy Tanaka) Conjunctive Use (Mimi Jenkins)
Other operational changes (Mimi Jenkins) 3:30Implications for State
Water Policy and Planning (Jay Lund and Richard Howitt)
3:45Continuing Work (Jay Lund) Conclusions 4:00End Agenda
- Slide 2
- CALVIN Economic-Engineering Optimization for Californias Water
Supplies Professor Jay R. Lund Civil & Environmental
Engineering, UC Davis Professor Richard E. Howitt Agricultural
& Resource Economics, UC Davis Web site:
cee.engr.ucdavis.edu/faculty/lund/CALVIN/
- Slide 3
- Real work done by Dr. Marion W. JenkinsDr. Andrew J. Draper Dr.
Kenneth W. KirbyMatthew D. Davis Kristen B. Ward Brad D. Newlin
Brian J. Van Lienden Stacy Tanaka Randy Ritzema Guilherme Marques
Siwa M. Msangi Pia M. Grimes Jennifer L. Cordua Mark Leu Matthew
EllisDr. Arnaud Reynaud
- Slide 4
- Funded by CALFED Bay Delta Program State of California
Resources Agency National Science Foundation US Environmental
Protection Agency California Energy Commission US Bureau of
Reclamation Lawrence Livermore National Laboratory
- Slide 5
- Overview 1) California Water Problems 2) What is CALVIN? 3)
Modeling Approach and Data 4) What is Optimization? 5) Results 6)
Innovations and Uses 7) Major Themes
- Slide 6
- California Water Problems California: an often dry place with a
good climate Wetter winters, very dry summers. Water in
north/mountains, demands in south, central and coast. Groundwater
is 30-40% of supply. Competition for water: Agriculture, urban,
environment
- Slide 7
- Motivation for Project Californias water system is huge and
complex. Water is controversial and economically important. Major
changes are being considered. Can we make better sense of this
system? Understanding from data and analysis Insights from results
Reduce reliance on narrow perspectives How could system management
be improved? What is user willingness to pay for additional water
and changes in facilities & policies? These are not back of the
envelope calculations.
- Slide 8
- What is CALVIN? Entire inter-tied California water system
Surface and groundwater systems Economics-driven optimization model
Economic Values for Agricultural and Urban Uses Flow Constraints
for Environmental Uses Prescribes monthly system operation
- Slide 9
- Approach a) Develop schematic of sources, facilities, &
demands. b) Develop explicit economic values for agricultural &
urban water use for 2020 land use and population. c) Identify
minimum environmental flows. d) Reconcile estimates of unimpaired
1922-1993 inflows, & identify problems therein. e) Develop
databases, metadata, and documentation for more transparent and
flexible statewide analysis.
- Slide 10
- Approach (continued) f) Apply economic-engineering optimization
to combine this information and suggest: 1) promising capacity
expansion & operational solutions, 2) economic value of
additional water to users, 3) costs of environmental water use, and
4) economic value of changes in capacities and policies.
- Slide 11
- Model Schematic Over 1,200 spatial elements 51 Surface
reservoirs 28 Ground water reservoirs 24 Agricultural regions 19
Urban demand regions 600+ Conveyance Links
- Slide 12
- CALVINs Demand Coverage
- Slide 13
- Economic Values for Water Willingness to pay Agricultural Urban
Operating Costs
- Slide 14
- Agricultural Water Use Values 0 10,000 20,000 30,000 40,000
50,000 60,000 70,000 050100150200250300350400 Deliveries (taf)
Benefits ($ 000 ) March August June July May April September
October 0 1,000 2,000 3,000 51015 October February January
- Slide 15
- Urban Water Use Values 0 5,000 10,000 15,000 20,000 25,000
30,000 35,000 40,000 45,000 50,000 202530354045505560 Deliveries
(taf) Penalty ($000) Winter Summer Spring
- Slide 16
- Operating Costs Fixed head pumping Energy costs Maintenance
costs Groundwater recharge basins Wastewater reuse treatment Fixed
head hydropower Urban water quality costs
- Slide 17
- Environmental Flow Constraints Minimum instream flows Rivers
and lakes Delta outflows Wildlife refuge deliveries Sources: DWRSIM
PROSIM - CVPIA Various local studies
- Slide 18
- Hydrology Surface & Groundwater 1921 - 1993 historical
period: Monthly unimpaired inflows Surface inflows from State &
Federal data Groundwater from Federal & local studies Need to
reconcile conflicting data!
- Slide 19
- Policy Constraints All Cases Environmental flows Flood control
storage Current Policy Base Case DWRSIM surface operations CVGSM
pumping and deliveries Unconstrained economic case Only environment
& flood control
- Slide 20
- Model Inputs Schematic and facility capacities Agricultural
water values Environmental Flow Constraints Urban water values
Operating costs Hydrology: Surface & ground water Policy
Constraints
- Slide 21
- Data Flow for the CALVIN Model
- Slide 22
- Database and Interface Tsunami of data for a controversial
system Political need for transparent analysis Practical need for
efficient data management Databases central for modeling &
management Metadata and documentation Database & study
management software Planning & modeling implications
- Slide 23
- What is Optimization? Finding the best decisions within
constraints. Best implies performance objective(s). Constraints
limit the range and flexibility of decisions. Some constraints are
physical. Other constraints are policy. Optimization can identify
promising solutions.
- Slide 24
- Network Optimization In Words Decisions: Water operations and
allocations Best performance: (1) Minimize sum of all costs over
all times Subject to some constraints: (2) Conservation of mass (3)
Maximum flow limits (4) Minimum flow limits
- Slide 25
- Network Flow with Gains - Math Minimize: (1) Z = c ij X ij, X
ij is flow from node i to node j Subject to: (2) X ji = a ij X ij +
b j for all nodes j (3) X ij u ij for all arcs (4) X ij l ij for
all arcs c ij = economic costs (ag. or urban) b j = external
inflows to node j a ij = gains/losses on flows in arc u ij = upper
bound on arc l ij = lower bound on arc
- Slide 26
- Some Results Delivery, Scarcity, and Cost Performance
Willingness to pay Water transfers and exchanges Economic Value of
Facility Changes Costs of Environmental Flows Conjunctive Use and
other Operations
- Slide 27
- CALVINs Innovations 1) Statewide model 2) Groundwater and
Surface Water 3) Supply and Demand integration 4) Optimization
model 5) Economic perspective and values 6) Data - model management
7) Supply & demand data checking 8) New management options
- Slide 28
- Themes 1.Economic scarcity should be a major indicator for
Californias water performance. 2.Water resources, facilities, and
demands can be more effective if managed together, especially at
regional scales. 3.The range of hydrologic events is important, not
just average and drought years. 4.Newer software, data, and methods
allow more transparent and efficient management.
- Slide 29
- More Information... Web site:
cee.engr.ucdavis.edu/faculty/lund/CALVIN/
- Slide 30
- Slide 31
- Water Demand Functions Urban demands are generally demands for
direct consumption. Agricultural demands are derived demands that
depend on the cost of other inputs and the value of the output.
Environmental demands are represented by constraints. The
elasticity of demand measures the responsiveness of the quantity
demanded to changes in the price ( cost ) of water. An elastic
demand has an elasticity greater than one, and is relatively
responsive to price changes. Direct statistical estimation of
demands is prevented by the absence of price variation in
Californian agricultural or urban water
- Slide 32
- Calvins Economic Requirements Calvin needs a stepped Penalty
function for each delivery node on the schematic. The penalty is
the cost of not having a quantity of water and is the inverse of
the demand function that measures the value of having water. The
continuous economic demand functions are divided into discrete
steps for linear solution. The agricultural and urban demand
functions must be calibrated so that the marginal value of the
observed deliveries is equal to the observed water price.
- Slide 33
- Data Base for the SWAP Model The Statewide Agricultural
Production Model ( SWAP ) models the regional adjustment of profit
maximizing farmers allocating water over the range of crops
currently grown in the region. The base year for calibrating SWAP
is 1992 Base costs, prices, yields and input quantities are largely
based on CVPM data. SWAP crop acreages were extrapolated to 2020
levels using forecasts from DWR Bulletin 160-98
- Slide 34
- Slide 35
- SWAP Model Regions
- Slide 36
- Slide 37
- Agricultural Crop Descriptions
- Slide 38
- Agricultural Response to Changes in the Price and Quantity of
Water Changes at the Extensive Margin Changes in the total Area of
Irrigated crops Adjustments in the regional cropping mix. Changes
at the Intensive Margin A change in crop input use per acre Changes
in Water application efficiency due to technology and
management.
- Slide 39
- Slide 40
- Efficiency-Cost Trade-offs. Orchards Sacramento Valley
- Slide 41
- Calibrating Regional Crop Production Functions Regional data
available- acres, average yields, input quantities, crop price,
input cash cost, resource constraints Optimize a constrained
problem to obtain shadow values on binding constraints. Define the
form of the production function- in this case, a quadratic function
of three inputs. Use maximum entropy methods, marginal and average
product conditions to estimate the production function coefficient
values. Use the ME coefficient values to define a production model
that calibrates to the base year, but is only constrained by the
resource constraints.
- Slide 42
- Slide 43
- Linking Annual Cropping Decisions to Monthly Water Use Assume
that crops require water in a predetermined monthly pattern based
on ET requirements. Farmers can make small water reallocations
between months. Any changes in the total applied water due to
technology or stress irrigation are allocated are allocated
proportionally across months
- Slide 44
- Slide 45
- 0 200 400 600 800 1000 0204060 Water Availability in KAF
Willingness to Pay in $/AF July Demand April Demand SWAP-Derived
Agricultural Water demands
- Slide 46
- Agricultural Water Use Values 0 10,000 20,000 30,000 40,000
50,000 60,000 70,000 050100150200250300350400 Deliveries (taf)
Benefits ($ 000 ) March August June July May April September
October 0 1,000 2,000 3,000 51015 October February January
- Slide 47
- Tomato production-Yolo county Water Land
- Slide 48
- Calibrated Production Surface for Grapes, Fresno
- Slide 49
- Urban Demands Residential demands are based on comprehensive
survey of estimated price elasticities. The demand function is
fitted through the observed 1995 price and quantity demanded with
the function slope determined by the elasticity. The 2020 demand
are obtained by scaling the 1995 quantities by population growth.
Commercial is assumed proportional to the population and is added
to the residential demand. Industrial are based on a 1991 survey of
shortage induced production losses in 12 counties, and scaled to
2020.
- Slide 50
- Residential Price Elasticities
- Slide 51
- Estimating Residential Demands Residential price elasticities
1995 retail water prices 1995 population x pcu 1+2+3 = 1995 Demand
Curve 1995 demand curve scaled by 2020 population 4 = 2020 Demand
Curve
- Slide 52
- Calibrating Residential Demands to the 2020 Population
- Slide 53
- Converting Residential demands to Urban Loss Functions
Integrate the 2020 residential demand function Add a constant level
of 2020 commercial use Find the quantity that drives the demand
price ( or loss of not having water) to zero Plot the loss function
against monthly delivery.
- Slide 54
- Residential Loss Functions for 2020
- Slide 55
- Industrial Values 1991 production losses due to water shortages
12 counties 1995 industrial use scaled by population to 2020 1+2 =
2020 Industrial Loss Function
- Slide 56
- 1995 Urban Residential Water Prices in California
- Slide 57
- Conclusions Water demand functions must reflect rational
adjustments to changing scarcity by users Demand functions need to
be specified by use, region and month Urban demands can be
estimated on a regional basis using published elasticities and base
year data Agricultural demand functions can be estimated using
regional production data and ET requirements.
- Slide 58
- Managing Data in CALVIN Model, Database, Documentation, and
Post-processing SOFTWARE
- Slide 59
- CALVIN Physical Outputs Flows across all links EOM Storage
Levels at all storage nodes Evaporation at all storage nodes
Deliveries to all demand nodes Flow Storage
- Slide 60
- CALVIN Economic Outputs Scarcities Cost of scarcities Marginal
WTP for more water Shadow values on constraints (Lagrange
Multiplier) Marginal value of water at each node Post-processed by
intersecting deliveries with economic water loss functions for Ag
& Urban Produced directly by HEC-PRM Optimization
- Slide 61
- SWAP Post-Processed Outputs Irrigation efficiencies Crop
acreages Crop yields Gross revenue Net revenue Ag annual deliveries
by water source from CALVIN SWAP Model
- Slide 62
- Transparent model assumptions Easily modified model inputs
Object oriented data management Documentation of input values =
metadata Metadata attached to each piece of data RELATIONAL
DATABASE Data Management Design Principles Time series stored in
DSS Paired data stored in DSS Model definition file in ASCII Output
in DSS HEC-PRM Requirements
- Slide 63
- CALVIN Data Storage & Software Data Storage Software
- Slide 64
- Region 3 Schematic
- Slide 65
- Network Component Listing
- Slide 66
- Node Properties
- Slide 67
- Metadata
- Slide 68
- Summary Metadata is essential Relational databases and software
allow modeling of complex systems in more detail than otherwise
possible Data management eliminates majority of input file errors
(especially related to syntax) More work to be done
- Slide 69
- Hydrologic and Agricultural Demand Calibration Integrate
DWRSIMs surface hydrology with CVGSMs groundwater hydrology
Reconcile DWR agricultural water demand assumptions with deliveries
in the CVPIA PEIS Produce a model consistent with established
representations of Californias hydrology and demands Identify data
problems and regions than cannot be fully reconciled
- Slide 70
- Calibration Procedures 1)Impose Base Case diversions,
deliveries, and operations on CALVIN 2)Adjust agricultural demands
to match BC deliveries 3)Adjust agricultural return flows and reuse
rates to calibrate groundwater to CVGSM NAA 4)Run CALVIN Base Case
and adjust streamflows to match DWRSIM 514a at 15 matching control
points 5)Verify scarcity results with similar estimates
- Slide 71
- Configuration of Physical and Calibration Links
- Slide 72
- Agricultural demands increased by about 10% (1.9 MAF) Reuse
rates reduced in a few regions SW return flows eliminated in much
of Tulare Basin GW calibration successful in Sac Valley (top chart)
Problems with GW calibration in Tulare Basin, esp. CVPM region 14,
18, 19, and 21 GW Calibration Results Sacramento Valley GW SJ
Valley & Tulare Basin GW
- Slide 73
- Net addition of 38 taf/year calibration flows Biggest monthly
imbalances on Sac R. (below Colusa BD) & Feather R. (incl.
Yuba+Bear), > +/- 1 MAF Largest annual imbalances on Sac R. +620
taf/yr, Lower SJR -434 taf/yr, and in-Delta CU (- 380 taf/yr)
Largest net additions of SW in CALVIN Region 1 and 4; net removals
in CALVIN Region 2 and 3 SW Calibration Results Sac R. Colusa BD
Sac R. Hood In-Delta CU
- Slide 74
- What we learned Calibration is necessary and long Hydrologic
data needs improvement: more explicit and separate GW & SW
hydrologic data better estimates (methods) for local
accretions/depletions Agricultural water use uncertainty is
significant Very weak data for modeling Tulare Basin: conjunctive
use operations groundwater-surface water interactions Large
discrepancy in in-Delta CU estimates
- Slide 75
- A Limited Foresight Model and the Value of Carryover
Storage
- Slide 76
- Perfect Foresight or the Model that Knew too Much Multi-year
optimization too omniscient Over valuation of existing facilities
Under valuation of new facilities Excessive carryover storage prior
to droughts Single-year optimization too short sighted How to
construct model with limited foresight to: Reflect imperfect
knowledge of probability distribution of inflows Balance present
and future water needs
- Slide 77
- Sequential Annual Runs Use one year time period (Oct Sep) 72
consecutive model runs for period-of-analysis 1922-1993 Model runs
linked by ending/carryover storage Carryover storage value
functions limit drawdown System performance sum of 72 year costs
excluding carryover storage penalties
- Slide 78
- Iterative Solution Run HEC-PRM for one year yr=72? BOP Oct =
EOP Sep Set yr=1, read initial conditions STOP Define initial
carryover storage penalty function START Significant improvement
Yes No yr = yr +1 Revise carryover storage penalty function
Calculate total penalties Iteration n=1 Iteration n=1? Yes No
Iteration n= n + 1 Yes
- Slide 79
- Model Run Times Run times remain an obstacle for analyzing
complex systems Solver times proportional to cube of number of
constraints Solver times proportional to number of variables Run
time as a function of years of analysis found to be approx.
quadratic Limited foresight model exploits rapid increase in run
times
- Slide 80
- Case Studies Three case studies used as proof of concept Single
reservoir operation Integrated two reservoir operation Single
reservoir operated conjunctively with groundwater Reservoirs
operated for water conservation and flood control Objective
function to minimize economic cost of shortage associated with d/s
agricultural deliveries Network flow solver (HEC-PRM)
- Slide 81
- Case (a): Application to a Single Reservoir Four separate
models developed: New Don Pedro Reservoir, Tuolumne River Lake
McClure, Merced River Pine Flat Reservoir, Kings River Lake
Berryesssa, Putah Creek Case studies to provide a framework for the
presentation of ideas rather than a realistic operation of each
stream-reservoir system. Value functions developed for agricultural
deliveries
- Slide 82
- Valuing Carryover Storage Assume quadratic penalty P = a + bS +
cS2 P|S=K = 0 dP/dS is -ve, d2P/dS2 is +ve dP/dS between reasonable
limits
- Slide 83
- Grid Search Simple to implement Ensures global optimum Response
surface mapped-out Computationally inefficient Accuracy limited by
grid spacing Coarse grid search starting point for more efficient
methods
- Slide 84
- Nelder-Mead Simplex Method Unconstrained minimization of
several variables Zero-order search method Global optimum not
guaranteed 1: initial simplex 2: simplex expansion 3: simplex
reflection 4: simplex reflection (reflection adjusted to stay
within feasible region) 5: simplex reflection (reflection adjusted
to stay within feasible region) 6: simplex reflection 7: simplex
contraction 8: simplex contraction 9: simplex contraction, solution
tolerance satisfied optimal penalties Pmin = -9 $/af, Pmax = -148
$/af 1 5 4 3 2 6 7 8 9
- Slide 85
- New Don Pedro Reservoir Average Annual Shortage Cost as a
Function of P min and P max
- Slide 86
- Lake McClure Average Annual Shortage Cost as a Function of P
min and P max
- Slide 87
- Pine Flat Reservoir Average Annual Shortage Cost as a Function
of P min and P max
- Slide 88
- Lake Berryessa Average Annual Shortage Cost as a Function of P
min and P max
- Slide 89
- New Don Pedro Reservoir Operation under Perfect Foresight
- Slide 90
- New Don Pedro Reservoir Operation under Limited Foresight
- Slide 91
- New Don Pedro Reservoir Carryover Storage for Different Levels
of Information
- Slide 92
- Reservoir Operating Rules
- Slide 93
- Case (b): Integrated Two- Reservoir System
- Slide 94
- Flood Control & Carryover Storage
- Slide 95
- Balancing Storage Between Reservoirs in Parallel 4 5 3 2 1 4 5
3 2 1 Objectives: Equalize refill potential Minimize EV(spills)
Avoid inefficient conditions
- Slide 96
- Carryover Storage Penalty Functions
- Slide 97
- Carryover Storage: New Don Pedro and New Exchequer
Reservoirs
- Slide 98
- Case (c): Conjunctive Use
- Slide 99
- Groundwater Mining For perfect foresight model groundwater
mining prevented by applying constraint to ending groundwater
storage For limited foresight model linear penalty attached to
end-of-year storage Initial value of penalty set equal to shadow
value on groundwater mining constraint from perfect foresight run
Penalty iteratively raised until no groundwater mining occurs
- Slide 100
- Comparison of Groundwater Storage under Perfect and Limited
Foresight Models
- Slide 101
- Conclusions: Perfect Foresight 1.Perfect foresight may
significantly distort reservoir operation where reservoirs are used
for over-year storage, and where multi-year droughts occur. 2.The
impacts of perfect foresight revealed by lack of hedging under
average hydrologic conditions aggressive hedging during the initial
years of an extended drought. 3.The perfect foresight models can
substantially under estimate shortages and shortage costs 4.As
system storage increases the effects of perfect foresight are
diminished.
- Slide 102
- Conclusions: Limited Foresight 1.The limited foresight model
Results in an economically derived value of carryover storage.
Prescribes more realistic reservoir operations. More likely to be
acceptable to stakeholders Facilitates the deduction of operating
rules Can quantify the over-achievement of perfect foresight
models. 2.There exists a wide range of near-optimal carryover
storage policies. 3.Differences between the limited and perfect
foresight model are minor except prior and during drought
conditions.
- Slide 103
- Conclusions: Conjunctive Use 1.Conjunctive use can
substantially improve overall system reliability and reduce total
costs 2.Considerable benefits may accrue by explicitly adjusting
surface reservoir operations to account for contingent groundwater
supplies. 3.The value of surface carryover storage rapidly
diminishes with increasing groundwater supplies. 4.Carryover
storage rules determined without explicitly accounting for the
presence of groundwater storage become economically very
inefficient as groundwater supplies increase. 5.Integrated
conjunctive use greatly reduces the impact of perfect
foresight.
- Slide 104
- Limitations Major sources of CALVINs limitations: Weak or
unavailable input data from other sources. Limitations of HEC-PRM
network solver. Lack of hydropower, flood control, and recreation
benefits.
- Slide 105
- Major Limitations 1)Surface Hydrology a)Valley floor inflows
b)Delta local supplies c)Southern California 2)Groundwater
Hydrology a)Weak Tulare Basin data b)Simplified stream-aquifer
interaction c)1991-1993 groundwater extensions d)CVGSM data
e)Pumping capacities, costs, and use
- Slide 106
- Major Limitations (continued) 3)Water Demands and Deliveries
a)Year-type variation b)Limited water quality c)Limited
understanding of agricultural flows d)Base case delivery data from
CVGSM e)Representation of local systems 4)Environmental Regulations
a)Delta representation b)Other pre-operated flow constraints
c)Water quality limitations
- Slide 107
- Major Limitations (continued) 5)Perfect Foresight a)Less
important with lots of groundwater storage b)Probably causes 5-10%
of S. California benefits c)Need to be careful d)Limited foresight
methods under development e)Value of simulation modeling;
6)Excluded Operating Benefits a)Hydropower b)Flood Control
c)Recreation
- Slide 108
- Limitations Implications A.Data problems apply to ANY regional
and statewide analysis. B.Use simulation models to refine and test
optimization-based solutions. C.All models are wrong, but some are
useful. G.E.P. Box (Like budget estimates?)
- Slide 109
- CALVIN Modeling Alternatives Randall Ritzema
- Slide 110
- How is water allocated now? Water rights and contracts +
Operating agreements + Environmental regulations + Water markets
(sometimes) = Complex institutional framework Typically analyzed
using simulation
- Slide 111
- Objective, in words: To quantify overall economic performance
under flexible operations and allocations, within environmental
requirements. Performed by comparing current operations (Base Case)
to flexible operations (Unconstrained Case).
- Slide 112
- Example Network Optimization SWGW URBAN ENVIRONMENTAL
AGRICULTURAL Inflow Link, no economic value Link, w/ economic
value
- Slide 113
- Base Case SW GW URBAN ENVIRONMENTAL AGRICULTURAL Constrained
flow Unconstrained flow
- Slide 114
- BC Delivery Constraints SW GW URBAN ENVIRONMENTAL AGRICULTURAL
Constrained flow Unconstrained flow DWRSIM Run 514, CVGSM NAA
1997
- Slide 115
- BC Operations Constraints SW GW URBAN ENVIRONMENTAL
AGRICULTURAL Constrained flow Unconstrained flow CVGSM NAA 1997
DWRSIM Run 514
- Slide 116
- Environmental Constraints SW GW URBAN ENVIRONMENTAL
AGRICULTURAL Constrained flow Unconstrained flow Minimum Instream
Flows (PROSIM NAA 1997) Refuges (Level 2)
- Slide 117
- Unconstrained Case SWGW URBAN ENVIRONMENTAL AGRICULTURAL
Constraints: Environmental flows Physical capacities Reservoir
flood control storage
- Slide 118
- Model Alternatives Base Case Basis for Comparison Goal: mimic
simulation planning models Current water allocations Current water
operations Current environmental flows
- Slide 119
- Model Alternatives Base Case Regional Unconstrained Statewide
model is divided into 5 regions Supplies re-operated and
re-allocated within regions. Base Case inter-regional boundary
flows
- Slide 120
- CALVIN Regions
- Slide 121
- Model Alternatives Base Case Regional Unconstrained Statewide
Unconstrained Inter-regional boundary constraints removed
- Slide 122
- Delivery and Economic Performance Comparison of Base Case,
Regional Unconstrained, and Statewide Unconstrained Alternatives
Randall Ritzema
- Slide 123
- Urban and Agricultural Deliveries (annual average, in taf)
- Slide 124
- Delivery Changes to Ag Sector (annual average, in taf)
- Slide 125
- Delivery Changes to Urban Sector (annual average, in taf)
- Slide 126
- Agricultural Scarcity Costs (annual average, in $
millions)
- Slide 127
- Urban Scarcity Costs (annual average, in $ millions)
- Slide 128
- Total Costs (annual average, in $ billions) 4.18 2.852.78
- Slide 129
- Total Cost by Region (annual average, in $ millions)
- Slide 130
- The importance of marginal values and spatial equilibrium
Marginal changes in water allocation and management must be
evaluated using marginal measures of value. Models of spatial
equilibrium have conditions which balance the marginal
profitability conditions of water between regions and uses In
CALVIN, the marginal values are measured in terms of the productive
uses of water at each of the demand nodes. Underlying marginal
values are production decisions which balance the marginal value
productivity of water across crops and months. Optimal CALVIN runs
are in spatial equilibrium since water cannot be reallocated
without violating a constraint or reducing the overall economic
value of Californias water.
- Slide 131
- The Spatial information in CALVIN CALVIN explicitly defines the
constraints and conveyance costs that constrain the movements of
water across space. The shadow values measure the marginal cost of
constraints. Calvin has the willingness-to-pay explicitly
represented for each node. The gainers and losers from trades are
clearly identified by location and sector
- Slide 132
- Slide 133
- Slide 134
- Practical Impediments to Water Trades Third party economic
impacts in the exporting regions Defining the tradeable quantities
of water by the consumptive use of applied water Avoiding negative
environmental impacts from wheeling traded water or changing the
type and location of use.
- Slide 135
- Using CALVIN to Measure Third party Economic Impacts CALVIN is
optimized for a base condition, and an optimal trade condition.
Urban benefits from trades are immediately shown by comparing the
net benefits at urban nodes. Agricultural effects are measured by
Post- optimality analysis using SWAP. CALVIN traded water
allocations are fed back into SWAP which then estimates the change
in regional crop production caused by the trade. Regional income
and employment multipliers are used to measure the community
effects of the change in production.
- Slide 136
- Defining Tradeable Water Quantities CALVIN tries to explicitly
measure consumptive use and return flows. The net change in
consumptive use by region and water use is calculated. CALVIN only
proposes trades in which the change in return flows do not violate
environmental constraints.
- Slide 137
- 00 -35 0 -15 -4 -18 39 -1,255 34 -1,323 -1,400 -1,200 -1,000
-800 -600 -400 -200 0 200 Upp Sac V. Ag Upp Sac V. Urb
L.Sac.V.& BD AgL.Sac.V.& BD Urb SJV.& SBay AgSJV.&
SBsy Urb Tulare B. Ag Tulare B. Urb SoCal Ag SoCal Urb TOTAL AG
TOTAL URB CHANGES IN SCARCITY COSTS ($ M/yr)
- Slide 138
- Slide 139
- Measuring Wheeling Costs and Constraints In calculating the
value of water trades, CALVIN accounts for wheeling costs and
constraints Constraints on wheeling water often limit trades,
CALVIN shows the marginal shadow values of such constraints. In
proposing trades, CALVIN not only identifies the buyer and seller,
but shows how the water can optimally be wheeled between the buyer
and seller in a particular month and for a water year type.
- Slide 140
- Conclusions on Marginal Value and Trade Water allocations and
adjustments occur on the margin accordingly, marginal values- not
average values- must be used to calculate efficient allocations.
CALVIN optimal results suggest potentially valuable water trades
that take into account all the costs and constraints on the system.
By post optimality analysis CALVIN can measure and adjust for the
main third party effects of water trades.
- Slide 141
- Conjunctive Use and Operational Changes Presented by Mimi
Jenkins
- Slide 142
- Overview Role of GW in California Conjunctive Use Operations
& Changes in CALVIN Operational Changes Sacramento Valley
Conjunctive Use Potential Impacts Some Limitations
Implications
- Slide 143
- GW in California 30-40% of Californias supplies in average
years More groundwater use in dry years Total storage capacity =
850 MAF Largely unmanaged
- Slide 144
- GW in CALVIN GW Resources: 28 GW basins Fixed inflows Economic
Drivers: Economic values for water use Operating costs of using
water sources Operating Constraints: Ending storage set to Base
Case (no additional mining) Pumping capacities Other
capacities
- Slide 145
- Statewide Reliance on GW and CU
- Slide 146
- Average Monthly % GW Supply
- Slide 147
- Statewide Groundwater Storage
- Slide 148
- Operational Changes in CALVIN Quality exchanges in Sac Valley,
& between SJ River & Bay Area *Sac Valley CU Operations
Regional operation of Bay Area resources Reduced Delta exports with
South-of-Delta re-operations (Region 3, 4 and 5) Increased Ag-Ag
& Ag-Urban transfers in Tulare Basin to increase CU and
eliminate scarcities Mojave River Basin GW banking operations
Southern California Ag-Urban & Urban-Urban transfers, increased
CU operations
- Slide 149
- Sacramento Valley Conjunctive Use UnconstrainedBase Case
Non-drought Year Drought Year Sacramento R. American R.
Groundwater
- Slide 150
- Sacramento Valley Re-operations Principal Demands: Upper
Sacramento Valley Ag (CVPM 1-4) Lower Sacramento Valley Ag (CVPM
6-8) Lower Sacramento Valley Urban (Greater Sac Area)
- Slide 151
- Sacramento River Diversions
- Slide 152
- American River Diversions
- Slide 153
- Groundwater Pumping
- Slide 154
- Sac Valley CU Impacts & Outcomes Surplus Delta outflow up
in drought years, slightly down in non-drought years, seasonal
shift uncertain Drought year diversions down - 430 taf on Sac R.,
228 taf on Amer. R. More flexibility to manage instream flows on
Sac and American R. and in Delta Reduced opportunity costs of
environmental flows $10 M/yr reduced operating costs (CVPM 7, 8 and
Greater Sac Urban) $42 M/yr reduced scarcity costs
- Slide 155
- Some Limitations of Results Minimum GW pumping for Ag demands
Capacity for Folsom S. Canal supply to CVPM 8 Dynamic
stream-aquifer interactions Variable head pumping costs Year-type
variation in Ag & Urban demands
- Slide 156
- Implications GW can serve both seasonal & drought demands
Optimized GW doesnt necessarily drain basins Economics &
markets can help us better employ GW Optimization models can
suggest promising conjunctive use solutions Conjunctive use can
substantially reduce need dry year diversions from streams
Conjunctive use can ease conflicts with environmental requirements
and operations
- Slide 157
- Conclusions and Implications
- Slide 158
- Conclusions from Results Some qualitative policy conclusions:
a) Regional or statewide markets have great potential to reduce
water scarcity costs. b) Economically efficient local and regional
management reduces demands for imports. c) Environmental flows have
economic costs for agricultural, urban, & other activities. d)
Economic values exist for expanded facilities.
- Slide 159
- e) Some scarcity is optimal. f) Economically optimal water
reallocations are very limited, but reduce scarcity and scarcity
costs considerably. g) Integrated local, regional, and statewide
operation of water decreases competition with environmental
uses.
- Slide 160
- Policy and Planning Implications 1)Optimization works and shows
promise. 2)Significant new capabilities: a)Statewide and regional
analysis b)Economic and engineering analysis c)Explicitly
integrated operations d)Transparent operations e)Suggests new
management options f)Take the good, but remember the
limitations.
- Slide 161
- Implications (continued) 3)Any regional & statewide
analysis needs to: a)Improve current data Central Valley hydrology
and demands Agricultural and urban water use Tulare Basin
b)Modernize data management (software and institutions)
c)Coordinate & extend data
- Slide 162
- Implications (continued) 4)CALVIN must graduate from the
University. a)Most uses and data are outside the University b)Were
happy to help others use this capability c)Models shouldnt get
tenure. 5)CALVIN is just a tool to help: a)Make better sense of a
complex system b)Develop ideas for water management 6)Why do
modeling? Need analytical ability to provide convincing ideas.
- Slide 163
- Uses for CALVIN? 1)Integrated statewide supply and demand
accounting & data framework 2)Preliminary economic evaluation
3)Long-term statewide water planning 4)Planning & operations
studies: Facility expansion, Joint operations, Conjunctive use,
& Water transfers 5) Suggest new management options 6) No
panacea, but a step along the way.
- Slide 164
- Continuing Work: Current Climate Change Study Economic costs
and benefits of different Delta Export levels Add hydropower and
flood control benefits Faster, more flexible solver Reasonable
foresight Fixing little things
- Slide 165
- Improvements in economic water demands Improvements in
environmental demands Policy studies (catastrophe response, water
transfers, new facilities, conjunctive use, environmental flows,
etc....) More detailed regional studies (Bay Area) Continuing Work:
Envisioned
- Slide 166
- Themes 1.Economic scarcity should be a major indicator for
Californias water performance. 2.Water resources, facilities, and
demands can be more effective if managed together, especially at
regional scales. 3.The range of hydrologic events is important, not
just average and drought years. 4.Newer methods, data, and
software, including optimization, support more transparent and
efficient management.
- Slide 167
- More Information... Web site:
cee.engr.ucdavis.edu/faculty/lund/CALVIN/