1

1

2

3 RE-ASSEMBLING

4 Water Supply Implications and Costs of Removing O'Shaughnessy Dam

5 6 7 8 Sarah Null and Jay R. Lund2 9

10

11 ABSTRACT: The Hetch Hetchy System provides with most of its water supply. 12 O’Shaughnessy Dam is one component of this system, providing approximately 25% of water 13 storage for the Hetch Hetchy System and none of its conveyance. Removing O’Shaughnessy 14 Dam has gained interest to restore Hetch Hetchy Valley. The water supply feasibility of 15 removing O’Shaughnessy Dam is analyzed by examining alternative water storage and delivery 16 operations for San Francisco using an economic-engineering optimization model. This model 17 ignores institutional and political constraints and has perfect hydrologic foresight to explore 18 water supply possibilities through re-operation of other existing reservoirs. The economic 19 benefits of O’Shaughnessy Dam and its alternatives are measured in terms of the quantity of 20 water supplied to San Francisco and agricultural water users, water treatment costs, and 21 hydropower generation. Results suggest there would be little water scarcity if O’Shaughnessy 22 Dam were to be removed, although removal would be costly due to additional water treatment 23 costs and lost hydropower generation. 24 (KEY TERMS: water supply, dam removal, Hetch Hetchy, optimization, filtration avoidance, 25 restoration.)

2 Respectively, Doctoral Student, Geography Graduate Group, University of California, Davis; Professor, Department of Civil and Environmental Engineering, University of California, Davis, 3109 Engineering Unit III, Davis, California 95616

2

1 INTRODUCTION

2 O’Shaughnessy Dam, located in the Hetch Hetchy Valley of ,

3 was built by the city of San Francisco in 1923, as a component of the Hetch Hetchy water

4 system. The Hetch Hetchy System has ten other reservoirs, numerous water conveyance

5 pipelines, and water treatment facilities. This system provides water to 2.4 million people in the

6 San Francisco Bay Area, including the city and county of San Francisco and 29 wholesale water

7 agencies in San Mateo, Alameda, and Santa Clara Counties (USBR, 1987).

8 O’Shaughnessy Dam was controversial when proposed and built in the early 1900s.

9 Throughout the past century, the idea of removing O’Shaughnessy Dam to restore Hetch Hetchy

10 Valley has never entirely gone away, although today California is much different. Yosemite

11 National Park is now one of the most loved and visited parks in the United States. California’s

12 Bay Area is a major urban region, with millions of residents requiring water delivery, and the

13 now has several times more storage capacity with the construction of New Don

14 Pedro Reservoir. For Hetch Hetchy Valley restoration to be considered, the remainder of the

15 Hetch Hetchy System or feasible alternative sources must be able to supply water without

16 O’Shaughnessy Dam. The importance of O’Shaughnessy Dam must be evaluated in the context

17 of the greater Hetch Hetchy System and other water supply alternatives.

18 This study provides quantitative estimates for the water supply feasibility of removing

19 O’Shaughnessy Dam using a spatially refined economic-engineering optimization model. Much

20 has been written and discussed about removing O’Shaughnessy Dam, but quantitative estimates

21 have not been released publicly. This project identifies cost minimizing alternatives for San

22 Francisco’s water supply and highlights how removing O’Shaughnessy Dam could change

23 current operations, water supply, deliveries, hydropower generation, water treatment, and

24 economic water supply costs. Removing O’Shaughnessy Dam could raise many additional

3

1 institutional, political, and economic questions. However, this study ignores most institutional

2 and political implications to focus on optimization of water supply and economic factors. It is

3 helpful to ignore political and institutional constraints to understand the physical limitations of a

4 water supply system. It then becomes clear whether human constraints or physical constraints

5 are limiting factors. Modeling can provide quantitative support for future discussion or actions.

6 This analysis indicates whether water scarcity would increase substantially without

7 O’Shaughnessy Dam assuming no additional water storage (Null, 2003). Such information also

8 might be useful as part of a more comprehensive benefit-cost analysis, not undertaken here.

9 This paper begins with a brief overview of Hetch Hetchy Valley, O’Shaughnessy Dam,

10 the Hetch Hetchy System, and potential O’Shaughnessy Dam removal decisions. CALVIN, the

11 computer model used to evaluate alternatives to O’Shaughnessy Dam, is explained. A summary

12 of model parameters, infrastructure modifications, benefits and limitations of CALVIN follow.

13 Model runs are described, with important assumptions noted, and results are discussed in terms

14 of operational and economic performance. The paper concludes with a short discussion on the

15 implications and water supply costs of removing O’Shaughnessy Dam, and some larger

16 institutional and economic factors affecting the feasibility of removing O’Shaughnessy Dam.

17 Background

18 Prior to construction of O’Shaughnessy Dam, Hetch Hetchy Valley was similar to

19 neighboring Yosemite Valley. Both valleys were formed from jointed granite bedrock. The

20 valleys were initially cut by the Tuolumne and Merced Rivers respectively; then glaciers scoured

21 them, widening and polishing the surrounding granite. Both valleys once had natural lakes that

22 filled with sediment, forming flat meadows (Huber, 2002).

23 In 1906, following the San Francisco earthquake, the shortcomings of San Francisco’s

24 water supply became obvious. City water planners targeted Hetch Hetchy as a site for a large

4

1 dam to ensure San Francisco’s water supply (Hundley, 1992). San Francisco’s voters approved

2 the construction of a dam in Hetch Hetchy Valley by an 86% majority vote in 1908 (,

3 2002). In 1913, the was passed in Congress, enabling a large reservoir to be built in a

4 national park (Hundley, 1992). O’Shaughnessy Dam was completed in 1923, raised in 1938, and

5 has been in use for the past eighty years.

6 O’Shaughnessy Dam has a storage capacity of 360,360 acre-feet (af). Its current uses

7 include water storage, hydropower generation, and some flood reduction (USBR, 1987). The

8 reservoir behind O’Shaughnessy Dam is not used for recreation. In terms of total water storage

9 in the Hetch Hetchy System, O’Shaughnessy Dam is not exceptionally large. It’s 360 thousand

10 acre-feet (taf) of surface water storage are approximately 25% of the surface storage in the Hetch

11 Hetchy System and 14% of storage on the Tuolumne River. For this study, removal of

12 O’Shaughnessy Dam applies only to the dam and its reservoir. No pipelines, diversion capacity,

13 or conveyance facilities would be removed from the Hetch Hetchy System.

14 As illustrated in Figure 1, the Hetch Hetchy System consists of several reservoirs, power

15 plants, and conveyance facilities. Downstream of O’Shaughnessy Dam, the Kirkwood Power

16 Plant and Moccasin Power Plant generate hydropower. Nearby, Cherry Reservoir and Eleanor

17 Reservoir are currently operated primarily for hydropower at Holm Powerhouse. Adequate

18 water supply storage is usually provided by O’Shaughnessy Dam. Another much larger dam,

19 New Don Pedro Reservoir, is downstream of O’Shaughnessy Dam, Cherry, and Eleanor

20 Reservoirs (SFPUC, 2002).

21 These four reservoirs, together with numerous Bay Area reservoirs and the connecting

22 pipelines make up the Hetch Hetchy System, operated by the San Francisco Public Utilities

23 Commission (SFPUC). Total surface storage in the Hetch Hetchy System is approximately

24 1,500 taf (Table 1). Of this quantity, 570 taf of storage is located in New Don Pedro Reservoir,

5

1 downstream of current Hetch Hetchy Aqueduct intakes. In addition to the Hetch Hetchy System

2 total, another 1,460 taf of storage is owned by Modesto Irrigation District and Turlock Irrigation

3 District in the 2,030 taf New Don Pedro Reservoir. The Hetch Hetchy System supplies water to

4 77% of the urban and industrial uses of the city and county of San Francisco, as well as parts of

5 San Mateo, Santa Clara, and Alameda Counties; supplying over 2 million water users (DOE,

6 1988). Three powerhouses on the upper Tuolumne River together provide approximately 2

7 billion KW hrs/yr of hydropower (USBR, 1987). This is a clean source of energy and an

8 important source of revenue for the SFPUC.

9 Although water storage is a priority for SFPUC and the Hetch Hetchy System, it is not

10 the 360 taf of water storage that makes O’Shaughnessy Dam valuable; rather water from

11 O’Shaughnessy Dam has filtration avoidance status (SFPUC, 2002). Typically, filtration of

12 urban surface water supplies is required by federal law. Filtration avoidance means

13 O’Shaughnessy Dam impounds extremely high quality water that meets water quality standards

14 under the federal Surface Water Treatment Rule. Only minimal water treatment is currently

15 necessary, such as addition of lime for corrosion control and chlorine or chloramine as a

16 disinfectant (Redwood City PWSD, 2003, SFPUC, 2003). The watershed above O’Shaughnessy

17 Dam is pristine; it lies within Yosemite National Park. O’Shaughnessy Dam is the only reservoir

18 in the Hetch Hetchy System that has filtration avoidance (and one of only about a half dozen

19 sources nationally). Even Cherry and Eleanor Lakes, less than ten miles from O’Shaughnessy

20 Dam do not qualify for filtration avoidance.

21 Recently the SFPUC approved seismic retrofits for the Hetch Hetchy System as part of its

22 Capital Improvement Program (SFPUC, 2002). Increased local San Francisco water storage

23 would be valuable in case of an earthquake, terrorist act, or other emergency repair. For this

24 study, all water storage is valued equally, regardless of location. However, because

6

1 O’Shaughnessy Dam is at the upstream end of the system, it is of little value in the event of

2 emergency interruptions in the conveyance system.

3 Potential for Removing O’Shaughnessy Dam

4 There are valid arguments for both keeping and removing O’Shaughnessy Dam. Many

5 arguments on both sides are economic. Hydropower is generated when water is released from

6 O’Shaughnessy Dam. In addition, loss of filtration avoidance determination would incur

7 considerable costs to the Hetch Hetchy System, and water users. Finally, some

8 environmentalists believe O’Shaughnessy Dam is a poor choice for removal because it benefits

9 no threatened or endangered species, and would make only minor improvements for ecological

10 connectivity on the Tuolumne River system. The land under the reservoir could be restored, but

11 this is a small land area to justify removal on environmental grounds.

12 The arguments for removing O’Shaughnessy Dam deal primarily with ethical concerns

13 and increasing open space in Yosemite National Park for tourism and recreation. It has been

14 argued, most famously by , that a reservoir for San Francisco residents simply does not

15 belong in Yosemite National Park, land that in theory belongs to all Americans (Muir, 1912).

16 Furthermore, Yosemite National Park is one of the most heavily visited national parks in the

17 nation. Within the park, Yosemite Valley is grossly congested. Restoring Hetch Hetchy Valley

18 could open a valley nearly identical to Yosemite Valley in terms of beauty and size to the public

19 and wildlife. Revenue and increased local spending from tourism could offset some or all of the

20 losses from removing O’Shaughnessy Dam. This problem can be thought of as weighing two

21 scarce resources, water and space in Yosemite National Park.

22 In recent years, construction of new dams has declined due to economic factors,

23 environmental considerations, and the best locations being already in use. This has led to the

24 idea of increasing water efficiency without new infrastructure (Poff and Hart 2002). Options for

7

1 increasing the efficiency of our water supply are wide reaching, including: coordinated use of

2 existing water infrastructure, conjunctive use between surface water and groundwater, water

3 conservation, water transfers, and generally more sophisticated operations (CDWR, 1998). With

4 this shift of ideology, the popularity of dam removal has risen dramatically. At least 467 dams

5 were removed in the latter part of the 20th century (Poff and Hart, 2002).

6 METHODS

7 Although simulation is the most common modeling approach, optimization models offer

8 another method for exploring solutions to water resource problems. Optimization methods can

9 suggest promising solutions with the ‘best’ performance from among many alternatives.

10 Optimization models require explicit objectives to be maximized or minimized by modifying

11 decisions within system constraints. Until recently, optimization models were too

12 computationally burdensome to be practical for large systems or problems. Now, more powerful

13 computers make optimization of large systems feasible (Labadie, 2004).

14 CALVIN (CALifornia Value Integrated Network) is a network flow based economic-

15 engineering optimization model of California’s inter-tied water management system. It was

16 developed at the University of California, Davis based on the US Army Corps of Engineers’

17 HEC-PRM reservoir optimization software (Jenkins et al., 2001; Draper et al., 2003).

18 The objective function for CALVIN is to minimize total economic costs, mathematically

19 represented as

20

21 Minimize Z = c (1) ∑∑ ij X ij ij

22

8

1 where Z is total cost, cij is the cost coefficient on arc ij, and X ij is flow from node i to node j, in

2 space and time. Convex non-linear costs can be represented as piece-wise linear. CALVIN also

3 has constraints representing physical or operational bounds. These include constraints for

4 conservation of mass, upper bounds, and lower bounds, written mathematically as

5

6 ∑∑X ji =aXij ij +bj for all nodes j (2) ii

7 X ij ≤ uij for all arcs (3)

8 X ij ≥ lij for all arcs (4)

9

10 where X ij is the flow from node i to node j, aij is gains or losses on flows in arc, bj is external

11 inflows to node j, uij is upper bound on arc, and lij is lower bound on arc (Jenkins et al., 2001).

12 CALVIN uses 72 years of monthly unimpaired historical data (1922-1993) to represent

13 hydrologic variability. The three worst droughts on record, 1929-1934, 1976-1977, and 1987-

14 1992, occurred during this period. California’s entire inter-connected water system has been

15 modeled with CALVIN (Jenkins et al, 2001; Draper et al., 2003). Additional CALVIN models

16 of California have been used for various regional and climate change water management studies

17 (Lund et al., 2003).

18 Physical parameters in the model include infrastructure capacities, environmental

19 requirements, and hydrology. Surface reservoirs and groundwater basins have upper and lower

20 bounds. For surface reservoirs, the maximum capacity is the bottom of the seasonally-varying

21 flood storage level and minimum capacity is the top of dead storage. For groundwater basins, the

22 maximum capacity is the total amount of water that can be stored in the aquifer, and the lower

9

1 bound is the lowest level that groundwater has historically been pumped. Capacity limits also

2 exist on pumping and conveyance facilities. Environmental requirements include minimum

3 instream flows and flows to refuges. Due to the controversy inherent in economic valuation of

4 environmental uses of water, environmental water requirements are modeled as constraints.

5 Economic parameters include economic penalty / demand functions and operating costs.

6 The objective function of CALVIN (which drives the model) is to minimize the sum of water

7 scarcity costs to urban and agricultural water users, lost hydropower benefits (compared with

8 some ideal condition), pumping costs, water treatment operating costs, and water quality costs.

9 Water scarcity costs are economic penalties imposed if agricultural and urban demands are not

10 fully met. The economic value of water for agriculture is derived from the Statewide

11 Agricultural Production Model (SWAP) (Howitt et al., 2001). SWAP is a separate model of

12 agricultural water and land use decisions that maximizes farm profits given a production function

13 and constraints on water, land, technology, and capital inputs. SWAP is similar to the Central

14 Valley Production Model (CVPM), except values of water delivery can vary between months.

15 For each agricultural region, SWAP is used to generate an economic loss function which

16 decreases as water delivery increases. The economic loss represents the reduction in agricultural

17 net revenues from limited water deliveries (higher marginal value of water), compared to ideal

18 profits if water were not a limiting factor (marginal value equals zero).

19 Urban economic cost functions are based on demand curves for urban water use. These

20 cost functions vary by month, and are assumed to have constant seasonal elasticity (Jenkins et

21 al., 2001). CALVIN uses projected demands for the year 2020 for both agricultural and urban

22 demand areas.

23 Operating costs correspond to variable costs, primarily for groundwater pumping, surface

24 pumping, and water treatment. Hydropower penalty curves often are non-linear, and are thus

10

1 difficult to model. For this study, separate hydropower storage and release penalties were chosen

2 to represent variable head facilities. These independent piece-wise linear penalty functions were

3 fit to the full multivariate non-linear surface using a least squares approach (Ritzema, 2002).

4 CALVIN produces a monthly time-series of water deliveries for each demand location,

5 which can be translated into scarcity costs based on economic value functions. For this paper,

6 scarcity is defined as the difference between the quantity of water delivered and the maximum

7 quantity demanded. Maximum demand delivery is derived from economic value functions, at

8 the point where marginal benefit of additional water is zero. Scarcity cost is the economic value

9 of maximum water deliveries minus the economic value of water actually delivered. The

10 marginal willingness-to-pay, or shadow value, for additional water is also output for each

11 demand node and time-step. This also is the slope of the economic value function at the

12 delivered quantity of water.

13 Model Area and Assumptions

14 To examine the effects of removing O’Shaughnessy Dam, this study focuses on the San

15 Joaquin and San Francisco Bay Regional CALVIN model (Figure 1). The model area includes

16 13 surface reservoirs, excluding O’Shaughnessy Dam, and five groundwater basins. Major

17 conveyance facilities include: the Hetch Hetchy Aqueduct, the California Aqueduct, the Delta

18 Mendota Canal, the South Bay Aqueduct, and the Pacheco Tunnel. Seven hydropower plants are

19 included. Minimum instream flows are imposed on the Tuolumne River below New Don Pedro

20 Reservoir, on the below the confluence with the Stanislaus River at Vernalis,

21 and on the Stanislaus River below Goodwin Reservoir (Ritzema and Jenkins, 2001).

22 Six urban demand regions and four agricultural demand areas are included in the model

23 area. Projected year 2020 demand data was obtained from DWR’s Bulletin 160-98 on per capita

11

1 urban water use by county and detailed analysis unit (DAU). Demand for several smaller cities

2 are represented as fixed 2020 demands (Ritzema and Jenkins, 2001).

3 The San Francisco Public Utilities Commission demand area combines the city and

4 county of San Francisco with most of San Mateo County. The Santa Clara Valley (SCV)

5 demand area includes Santa Clara Valley Water District, Alameda County Water District, and

6 Alameda County Zone 7. The SFPUC and SCV water urban demand areas and all agricultural

7 demand areas are represented using economic value functions (Ritzema and Jenkins, 2001).

8 Although the regional model is large enough to allow for coordinated use between many

9 different storage and conveyance facilities, the focus of this study is the Hetch Hetchy System

10 (Figure 2). In CALVIN, the local San Francisco area reservoirs (Calaveras, Lower Crystal

11 Springs, San Andreas, and San Antonio) have been represented as a single, aggregated service

12 area reservoir. Pilarcitos Reservoir was omitted because of its negligible storage (3 taf). Cherry

13 and Eleanor Reservoirs are represented as a single reservoir because of the inability to

14 disaggregate inflows into these reservoirs and the existence of an interconnecting tunnel. Non-

15 storage reservoirs in the Hetch Hetchy System, such as Early Intake and Moccasin Reservoir are

16 represented as junction nodes instead of reservoirs. Hydropower is included in the Hetch Hetchy

17 portion of the model for Kirkwood, Holm, Moccasin, and New Don Pedro power plants.

18 Minimum instream flows of 50-125 cfs are imposed on the Tuolumne River downstream of New

19 Don Pedro Reservoir (USBR, 1987).

20 Although raising the dam at Calaveras Reservoir has been discussed to increase storage

21 in the Hetch Hetchy System, Calaveras Reservoir is given a maximum capacity of 91 taf.

22 Likewise, storage at this site has not been lowered to the current restriction of 28% of total

23 maximum storage (SFPUC, 2003). Model runs indicate surplus storage (greater than the amount

24 reduced) already exists in the local San Francisco area. While additional service area water

12

1 storage is a priority for SFPUC and the Hetch Hetchy System in the event of a seismic event or

2 terrorist act, this study assumes no new storage.

3 New Don Pedro Reservoir has a maximum capacity of 2,030 taf. Of this, SFPUC and the

4 Hetch Hetchy System own 570 taf. Modesto Irrigation District and Turlock Irrigation District

5 own the remaining storage. In CALVIN, storage space in New Don Pedro Reservoir is not

6 divided between different owners. Rather, the maximum capacity of New Don Pedro Reservoir

7 was set to 2,030 taf and water is allocated to different urban and agricultural demands as needed.

8 Because CALVIN is economically driven, any water scarcity tends to be allocated to demand

9 areas with lower economic values for water, usually agricultural areas. Here, results should be

10 interpreted to indicate the extent of water scarcity. However, in the real world water scarcity

11 often occurs to demand areas based on water rights and contracts. In the Tuolumne River area,

12 agricultural users with senior water rights would be unlikely to face scarcity, unless they

13 marketed water to other users.

14 For model runs in which O’Shaughnessy Dam has been removed, an inter-tie between

15 New Don Pedro Reservoir and the Hetch Hetchy Aqueduct has been added. Physically, the

16 Hetch Hetchy Aqueduct crosses New Don Pedro Reservoir. As stated above, the Hetch Hetchy

17 System owns storage space in New Don Pedro Reservoir; however, there currently is no way to

18 route this water to Bay Area users, except by releasing it through the Tuolumne River to the San

19 Joaquin River, pumping from the Delta, then routing it to either the South Bay Aqueduct or the

20 Pacheco Tunnel via the California Aqueduct (entailing significant quality degradation). This

21 hypothetical New Don Pedro-Hetch Hetchy Aqueduct inter-tie increases flexibility in the

22 conveyance system, and ensures higher quality water to Bay Area customers than water pumped

23 from the Delta. The New Don Pedro inter-tie capacity is limited by the downstream Hetch

24 Hetchy Aqueduct capacity of 465 cfs.

13

1 Water Treatment

2 Water treatment costs are typically non-linear with high fixed costs and economies of

3 scale. In CALVIN, treatment costs can only be modeled with unit costs on treatment links.

4 Operating treatment costs from Owens Valley and the LA Aqueduct System (another high

5 quality system) were applied to the Hetch Hetchy System to estimate possible increased variable

6 treatment costs from loss of the filtration avoidance determination. Additional fixed costs for

7 constructing water treatment plants require a side calculation outside of the model.

8 Detailed economic analysis of increased water treatment costs are beyond the scope of

9 both CALVIN and this study. Economic analysis of increased variable operating water treatment

10 costs were included in the model. Removal of O’Shaughnessy Dam would prompt loss of

11 regulatory filtration avoidance status, incurring considerable construction costs for new treatment

12 facilities. This study could not include a detailed analysis of construction costs of new facilities.

13 However, a rough estimate of construction costs can be based on costs for similar systems. The

14 Croton Water System, about 10% of New York City’s supply, is currently facing filtration

15 facility construction costs of about $950 million. If the much larger Catskills/Delaware System,

16 also for New York City, lost its filtration avoidance status, construction of treatment facilities are

17 estimated at $4 billion to $8 billion (NYC Independent Budget Office, 2000). The

18 Catskills/Delaware Water System is much larger than the Hetch Hetchy System. Thus, a rough

19 estimate of costs for additional water treatment facilities for the Hetch Hetchy System could

20 reach $2 billion. Although expansion of water treatment facilities is a long-term goal for the

21 SFPUC, new treatment facilities are costly, and even deferral of such a large expense has

22 considerable financial benefits. If construction of new treatment facilities is $2 billion, the value

23 of delaying construction is approximately $100 million/year (using a discount rate of 5%). This

24 makes keeping O’Shaughnessy Dam a part of the Hetch Hetchy System a priority for the

14

1 SFPUC. More detailed economic studies of water treatment costs might be useful as part of a

2 more comprehensive benefit-cost analysis.

3 Year 2100 Demands

4 Two model runs are included with estimated urban and agricultural water demands for

5 the year 2100, with roughly three times the current urban population (Landis and Reilly, 2002;

6 Lund et al., 2003). These extreme model runs are used to examine the effects of removing

7 O’Shaughnessy Dam when water demand is much, much greater in the very distant future.

8 Historic hydrology was used for these runs, ignoring potential effects of climate change.

9 Some additional changes made for the ‘2100’ model runs include: coastal demand

10 regions given unlimited access to seawater desalination at a constant unit cost of $1000/acre-

11 foot; urban wastewater recycling is available for up to 50% of return flows, also at a cost of

12 $1000/acre-foot; some environmental demands increased, although no changes were made to

13 environmental demands for links on or from the Tuolumne River; and O&M water treatment

14 costs increased to represent the loss of filtration avoidance by the year 2100. (Variable treatment

15 costs were increased to the same level as 2020 model runs with higher treatment costs.)

16 Model Runs

17 Six model runs are compared for this study, one constrained to current operating policies

18 with 2020 demands, three unconstrained runs from the year 2020 modeling set, and two model

19 runs with year 2100 demands (Table 2). Of the unconstrained year 2020 runs, two runs include

20 O’Shaughnessy Dam. One has increased water treatment costs to represent loss of filtration

21 avoidance; the other has no change to water treatment costs to represent filtration avoidance

22 being maintained. For all runs without O’Shaughnessy Dam, an inter-tie from New Don Pedro

23 Reservoir to Hetch Hetchy Aqueduct is added and higher O&M water treatment costs are

24 imposed, reflecting the end of filtration avoidance. Both model runs with year 2100 water

15

1 demands have increased water treatment costs. In one run O’Shaughnessy Dam is removed and

2 an inter-tie linking New Don Pedro Reservoir with the Hetch Hetchy Aqueduct is added. Results

3 from a base case modeling set represent 2020 conditions with current operating and allocation

4 policies (Jenkins et al., 2001). Some of these results are included for comparison and are

5 referred to as base case results.

6 Limitations

7 Consequences of O’Shaughnessy Dam removal could be substantial. The complexity of

8 the system and management options require a quantitative approach, for which computer

9 modeling is well suited. However, several limitations apply to the approach taken here. As with

10 all modeling studies, management and river systems are simplified. Economic benefits from

11 recreation are currently neglected. Recreation and tourism in Hetch Hetchy Valley would likely

12 be substantial, providing revenue and benefits to Yosemite National Park and nearby towns, but

13 ours is not a comprehensive benefit-cost analysis. Another important limitation with CALVIN is

14 perfect hydrologic foresight. This allows the model to prepare for droughts, reducing water

15 scarcity and associated costs. This limitation tends to be of lesser importance when large

16 amounts of storage (including groundwater) are available (Draper, 2001). Urban and agricultural

17 economic demands are assumed to be fixed, and groundwater basins are extremely simplified.

18 Jenkins et al. (2001) provides more on the limitations of CALVIN.

19 All model runs, except the base case, are unconstrained by current institutional and legal

20 allocation policies. This reduces water scarcity, since water operations and allocations have

21 perfect institutional flexibility. However, removing current policy constraints does show what

22 could be possible with existing infrastructure. In these model runs, operations and water

23 allocation are only economically driven. Absence of institutional implications, and public or

16

1 political support for the idea is perhaps the greatest limitation. These are not part of this model,

2 but are nevertheless driving factors. Additional examination of these factors would be useful.

3 RESULTS

4 Overall, year 2020 model runs show little water scarcity when O’Shaughnessy Dam is

5 removed and an inter-tie added between New Don Pedro Reservoir and the Hetch Hetchy

6 Aqueduct. Although storage at the O’Shaughnessy damsite is eliminated, flow in the Tuolumne

7 River does not change. Capture of significant quantities of water into the upper Hetch Hetchy

8 Aqueduct remains possible. The addition of an inter-tie between New Don Pedro Reservoir and

9 the Hetch Hetchy Aqueduct allows capture of Tuolumne River water from New Don Pedro

10 Reservoir. Thus, the lower Hetch Hetchy Aqueduct can remain full at all times regardless of the

11 existence of O’Shaughnessy Dam.

12 Although water deliveries are not greatly affected by removing O’Shaughnessy Dam,

13 removal could be costly. Hydropower generation suffers without water storage at

14 O’Shaughnessy Dam. Also, removing this reservoir ends SFPUC’s filtration avoidance and

15 creates a need for additional water treatment facilities. Filtration avoidance is important and

16 valuable enough to drive decisions regarding potential removal of O’Shaughnessy Dam.

17 Water Storage

18 Total water storage in the Hetch Hetchy System falls in model runs without

19 O’Shaughnessy Dam. Without O’Shaughnessy Dam, storage drops by approximately 350,000

20 af, the capacity of O’Shaughnessy Dam, in all water years. To assess whether the storage space

21 lost from removing O’Shaughnessy Dam is critical to meet water deliveries, storage in the other

22 Hetch Hetchy System reservoirs is evaluated. Water storage remains about the same in

23 Cherry/Eleanor Reservoir, New Don Pedro Reservoir, and local San Francisco reservoirs when

24 O’Shaughnessy Dam is removed from the model. Model runs indicate much storage is never

17

1 used in local San Francisco reservoirs. Thus, considerable storage remains in the overall Hetch

2 Hetchy System without O’Shaughnessy Dam.

3 Water Deliveries and Scarcity

4 With and without O’Shaughnessy Dam, full deliveries are made to urban demand areas

5 for all optimized 2020 model runs (Table 3). Although it is possible to deliver water to San

6 Francisco and Santa Clara Valley urban areas via the Hetch Hetchy Aqueduct, the Pacheco

7 Tunnel, or the South Bay Aqueduct, deliveries via all these pipelines are unaffected by removing

8 O’Shaughnessy Dam. This indicates that removing O’Shaughnessy Dam would change

9 operation of the Hetch Hetchy System, but need not affect surrounding water resources. When

10 model runs are constrained to current operational constraints (base case), a small amount of

11 scarcity occurs to urban water users (Jenkins et al., 2001; Ritzema and Jenkins, 2001).

12 In all model runs, full deliveries are made for environmental uses. This includes

13 minimum instream flows on the lower Tuolumne River and flows to wildlife refuges such as the

14 San Joaquin and Mendota Refuges.

15 A slight decrease in deliveries to agricultural demand areas occurs in model runs without

16 O’Shaughnessy Dam. Total annual average deliveries to agricultural regions is 5,257,983 af/yr

17 with O’Shaughnessy Dam, but average 575 af/yr less without O’Shaughnessy Dam. In the driest

18 year this increases scarcity to 72.5 taf. It slightly reduces water delivered during dry years to the

19 eastern north of the Tuolumne River (CVPM 11) and the Eastern San

20 Joaquin Valley floor between the Merced River and the Tuolumne River agricultural area

21 (CVPM 12) (Figure 3). In the runs without O’Shaughnessy Dam, inclusion of a New Don Pedro

22 inter-tie routes some water from the Tuolumne River into the Hetch Hetchy Aqueduct. Some

23 water is diverted to CVPM 11 and CVPM 12 from the Tuolumne River in CALVIN, so there is a

24 small transfer of water from agricultural uses to urban uses during a few very dry years.

18

1 Water rights and the allocation of storage space to distinct operating agencies are not

2 included in CALVIN. Essentially CALVIN assumes that SFPUC purchases a small amount of

3 water from irrigation districts during shortage events, and this amount of water purchased

4 increases slightly without O’Shaughnessy Dam. Therefore, these results should be interpreted as

5 indicating the water scarcity that could be anticipated from removing O’Shaughnessy Dam.

6 Using results from the base case modeling set, constrained by 1997 operational policies

7 (including O’Shaughnessy Dam), water scarcity is observed in SFPUC and Santa Clara Valley

8 residential demand area in 1921-1934, 1977, and 1986-1993. Over the entire 72 year time span,

9 average annual water scarcity is 6 taf for SFPUC and 10 taf for Santa Clara Valley (Jenkins et

10 al., 2001).

11 Conveyance

12 Without O’Shaughnessy Dam, flows through the upper Hetch Hetchy Aqueduct (above

13 New Don Pedro Reservoir) are less, but remain considerable. Flows in the Tuolumne River

14 above the O’Shaughnessy damsite do not change with removal of the reservoir. Only storage is

15 eliminated. Thus, diversion of considerable quantities of runoff could be possible at the damsite

16 for much of most years. When flows in the upper Hetch Hetchy Aqueduct are examined

17 seasonally, the importance of spring snowmelt can be seen (Figure 4). The upper aqueduct is

18 always at capacity in April and May, from spring runoff. During other months, flows through

19 the upper aqueduct vary considerably with streamflow.

20 Flows through a New Don Pedro inter-tie to the lower Hetch Hetchy aqueduct are the

21 inverse of flows through the upper Hetch Hetchy Aqueduct (Figure 5). During April and May,

22 flows through the New Don Pedro inter-tie are zero, because the aqueduct is already at capacity

23 with diversions at the O’Shaughnessy damsite. During other months, water is diverted from

24 New Don Pedro into the inter-tie as is needed to bring the lower portion of the Hetch Hetchy

19

1 Aqueduct to capacity. Thus, the lower Hetch Hetchy Aqueduct (downstream of New Don Pedro

2 Reservoir) is always at capacity when flows through a hypothetical New Don Pedro inter-tie are

3 incorporated. The New Don Pedro inter-tie adds flexibility to the Hetch Hetchy System. Were

4 O’Shaughnessy Dam to be removed, additional flexibility and conveyance from New Don Pedro

5 would remain of value.

6 Hydropower and Water Treatment

7 Hydropower generation is reduced substantially without O’Shaughnessy Dam, primarily

8 from eliminating hydropower generation at Kirkwood Power Plant, which requires head from

9 O’Shaughnessy Dam. Generation at Moccasin Power Plant is reduced significantly, and is

10 reduced slightly at Holm Power Plant. The loss of hydropower generation at Kirkwood and the

11 reduction at Moccasin and Holm average 457 GWhr/yr. The average annual hydropower

12 revenue loss is approximately $12 million/yr, assuming monthly varying wholesale electricity

13 prices (Ritzema, 2002).

14 Filtration treatment O&M costs of $17/af are used, based on O&M costs for other

15 California cities with high quality source water, like that of the Hetch Hetchy System. If

16 filtration avoidance were lost, treatment operating costs for the SFPUC could raise to $17/af, or

17 $6 million/year. Most likely, these costs would be passed on to urban water users, raising

18 monthly water bills to rates comparable to other California cities. Additionally, a slight decline

19 in water quality would occur from removing O’Shaughnessy Dam. Nevertheless, water quality

20 would remain high because reservoirs such as New Don Pedro (which do not have filtration

21 avoidance) also have exceptional water quality (Tuolumne Irrigation District, 2002). Given

22 current filtration avoidance, the total average annual water treatment cost of removing

23 O’Shaughnessy Dam would be the capital cost of new filtration (about $100 million/year) plus

24 the annual O&M cost of $6 million, totaling about $106 million/year.

20

1 Year 2100 Results

2 In model runs with projected “year 2100” demand, the entire region is short of water due

3 to population growth, but not short of storage. In these “year 2100” model runs, water scarcity is

4 extensive for both urban and agricultural demand regions. There is simply not enough water,

5 despite a surplus of surface reservoir storage space. This underscores an important distinction;

6 water and storage space are not the same. In year 2100 runs, water is generally not stored in

7 surface reservoirs for extended periods, it is used promptly to meet increased demands, reducing

8 evaporative losses to water supply. (These 2100 model runs ignore possible climate change

9 effects. Were precipitation patterns in California to change, most likely resulting in less snowfall

10 and more rainfall, some of the surplus storage seen in these results could be utilized seasonally,

11 but that would be another study.)

12 The year 2100 model run without O’Shaughnessy Dam stores slightly more water in New

13 Don Pedro Reservoir. Like year 2020 model runs, the Hetch Hetchy Aqueduct below New Don

14 Pedro Reservoir remains full despite the removal of O’Shaughnessy Dam with year 2100

15 demands. There is always less surface storage in the remaining Hetch Hetchy System reservoirs

16 in year 2100 models than in year 2020 models. This implies that despite considerable storage

17 space, there is not enough water to meet demands. Excess water rarely exists to be stored for

18 future years, rather it is usually sent to demand areas within a year. More groundwater is used

19 for drought storage in year 2100 models than in year 2020 models. As demand increases in the

20 future, conjunctive use (which reduces evaporative losses) will probably become more

21 widespread. Little difference in groundwater storage exists between the year 2100 results with

22 and without O’Shaughnessy Dam.

23 In both year 2100 model runs there is a small amount of water scarcity to urban water

24 users. Residential San Francisco water users have an average annual 5 taf of water scarcity, and

21

1 Santa Clara County water users have an average annual 1 taf of water scarcity (Table 4). Full

2 deliveries are made to all other urban demand areas. There is at least 100 taf/year of water

3 scarcity to all agricultural demand areas. Surprisingly, scarcity decreases slightly in the year

4 2100 demand model run without O’Shaughnessy Dam, attributable to greater evaporative losses.

5 Additional desalination and recycling facilities in CALVIN went unused. Model results

6 show the marginal willingness to pay for additional water by urban users varied between $650-

7 $800/af. The marginal willingness to pay for agricultural users was far less, between $100-

8 $200/af. The large difference in marginal willingness-to-pay between urban and agricultural

9 users arises because the lower Hetch Hetchy Aqueduct is at maximum capacity. Despite water

10 scarcity to urban regions, more water cannot be delivered without additional conveyance

11 infrastructure. Both desalination and recycled water was available at a price of $1000/af. Users

12 opt to face some scarcity (reducing use by conservation efforts) rather than pay for additional

13 water from these sources (unless costs of desalinization decrease significantly). Some scarcity

14 can be optimal.

15 DISCUSSION

16 A stable water supply for the San Francisco peninsula could be maintained after

17 removing O’Shaughnessy Dam. The numerous reservoirs and pipelines in the Hetch Hetchy

18 System provide considerable flexibility for delivering water. Adding an inter-tie linking New

19 Don Pedro Reservoir with the Hetch Hetchy Aqueduct makes it possible for little change to

20 occur to water deliveries when O’Shaughnessy Dam is removed. This inter-tie from New Don

21 Pedro Reservoir allows for the potential isolation of O’Shaughnessy Dam decisions from other

22 parts of the San Joaquin and Bay Area. Water storage space in New Don Pedro Reservoir is

23 shared between three entities: SFPUC, the Modesto Irrigation District, and the Turlock Irrigation

24 District. This inter-tie would also allow water transfers, exchanges, or other forms of flexible

22

1 operations among these agencies in the event of a long drought or operational interruptions to

2 facilities upstream.

3 When O’Shaughnessy Dam is removed and water demands are increased to represent

4 projected year 2100 demands, there are surprisingly few effects of dam removal on water

5 deliveries and operation of the Hetch Hetchy System. Some water scarcity occurs to urban

6 residential demand areas, and considerable water scarcity to agricultural demand areas,

7 regardless of the existence of O’Shaughnessy Dam. Scarcity occurs because insufficient water

8 exists in the system to meet demand, despite unused surface water storage and the lower Hetch

9 Hetchy Aqueduct remaining at capacity at all times. Although water desalination and water

10 recycling are available, some water scarcity costs (for water conservation) are preferable to

11 higher costs of acquiring additional water supplies. With increased future demands, water

12 storage in groundwater basins increases, suggesting greater utilization of conjunctive use

13 strategies in the future.

14 Removing O’Shaughnessy Dam carries considerable financial costs. These include lost

15 hydropower revenue, construction costs for additional water treatment facilities, increased

16 treatment costs, and dam removal costs. Expanded opportunities for tourism and recreation in

17 Hetch Hetchy Valley and resulting regional economic development would be needed to justify

18 dam removal and restoration economically. If urban, agricultural, and environmental water

19 demands can be met without O’Shaughnessy Dam, the decision to remove the reservoir and

20 restore Hetch Hetchy Valley becomes economic and political.

21 The importance of the filtration avoidance determination of O’Shaughnessy Dam cannot

22 be overemphasized. Filtration avoidance makes O’Shaughnessy Dam extremely valuable for

23 SFPUC and the Hetch Hetchy System, saving the SFPUC several tens of millions of dollars each

24 year in operating and deferred capital costs. It is very possible that this filtration avoidance

23

1 status will drive decisions regarding dam removal. However, if filtration avoidance status were

2 lost, O’Shaughnessy Dam would lose most of its value to the Hetch Hetchy System. In that case,

3 economic value and revenues from recreation and tourism in Hetch Hetchy Valley might offset

4 lost hydropower revenue and increased treatment facility operation costs. However, if Hetch

5 Hetchy Valley was opened to recreation, the economic benefits would go primarily to the

6 National Park Service, concessionaires, and the local economy, while SFPUC would incur most

7 of the costs due to lost hydropower generation and additional water treatment costs. Further

8 thought and research is needed to examine these possibilities and changes.

9 Finally, water managment in California is very dynamic. Changes in climate, water laws,

10 water markets, or technology could alter considerably the way water is moved and valued. It is

11 beyond the scope of this project to provide a comprehensive economic benefit-cost analysis or an

12 estimate of public support for removal of O’Shaughnessy Dam. A thorough benefit-cost analysis

13 of potential dam removal would be useful. Travel cost surveys and contingent valuation surveys

14 could be used to estimate the economic benefits and public support for expanded recreation

15 potential. Estimates of increased regional economic development also would be useful. Future

16 ecological studies include creating a restoration plan for Hetch Hetchy Valley, and measuring the

17 impacts of dam removal on the Tuolumne River and surrounding ecosystems. These benefits of

18 restoration could then be compared with losses from lower hydropower production and other

19 costs. Additional research on the institutional aspects of possible water transfers or exchanges

20 between SFPUC, the Modesto Irrigation District, and the Turlock Irrigation District would also

21 be useful. Potential effects of climate change also merit examination.

22 CONCLUSIONS

23 1) Removing O’Shaughnessy Dam need not substantially increase water scarcity. Without

24 O’Shaughnessy Dam, capture of considerable quantities of runoff could be possible at the

24

1 damsite for much of most years. Only storage is eliminated. Flow in the Tuolumne River above

2 the O’Shaughnessy damsite does not change with removal of the reservoir. If New Don Pedro

3 Reservoir is connected directly with the Hetch Hetchy Aqueduct, almost all demands can be

4 satisfied, without affecting the larger region outside the Tuolumne Basin. This substantially

5 reduces the need for difficult large-scale coordinated use agreements.

6 2) Conveyance sometimes can substitute for water storage. Connecting New Don Pedro

7 Reservoir with the Hetch Hetchy Aqueduct allows operators of the SFPUC flexibility to use

8 different reservoirs most effectively to meet full water deliveries to demand regions. Even with

9 O’Shaughnessy Dam remaining in the Hetch Hetchy System, the inter-tie is valuable to provide

10 flexibility in the system.

11 3) Removing O’Shaughnessy Dam substantially reduces hydropower generation and revenues.

12 Approximately $12 million/year would be lost from hydropower revenue, primarily from

13 decreased hydropower generation at Moccasin and Kirkwood power plants.

14 4) The loss of filtration avoidance, which would occur with removal of O’Shaughnessy Dam,

15 may be a predominant economic factor in the debate to remove O’Shaughnessy Dam.

16 Construction costs of additional filtration facilities would be huge ( roughly $2 billion). With

17 resulting capital costs of roughly $100 million/year, and O&M costs around $6 million/year,

18 removing O’Shaughnessy Dam could increase Bay Area drinking water costs significantly, to

19 levels common for most California cities.

20 5) Even with much greater future water demands, removing O’Shaughnessy Dam has little

21 effect on the Hetch Hetchy System and water deliveries to demand regions. Although there is

22 unused surface storage space with projected future demands, there are not enough water inflows.

23 Water and surface storage are not interchangeable.

25

1 6) Optimization modeling is useful in identifying effective re-operations for water resource

2 systems potentially undergoing restoration.

3 7) Water delivery to urban and agricultural users may be limited by political and institutional

4 constraints, rather than water storage or other physical components of the Hetch Hetchy System.

5 Hopefully, technical studies such as this will help enlighten deliberations and suggest

6 opportunities.

7 ACKNOWLEDGEMENTS

8 The authors thank Mimi Jenkins and Stacy Tanaka for guidance and modeling expertise

9 throughout this project. We also thank three reviewers for their helpful comments on this

10 manuscript. Ironically, this work was partially funded by a fellowship from the UC Davis John

11 Muir Institute for the Environment.

12 LITERATURE CITED

13 California Department of Water Resources. 1998. The California Water Plan Update. Bulletin 14 160-98. DWR. Sacramento, CA. 15 Draper, A. J. 2001. Implicit Stochastic Optimization with Limited Foresight for Reservoir 16 Systems. UC Davis PhD Dissertation. 17 Draper, A. J., M. W. Jenkins, K. W. Kirby, J. R. Lund, R. E. Howitt. 2003. Economic- 18 Engineering Optimization for California Water Management. Journal of Water Resources 19 Planning and Management. 129(3):155-164. 20 Howitt, R. E., K. B. Ward, S. M. Msangi. 2001. CALVIN Water Management Appendix A. 21 Statewide Water and Agriculture Production Model. Available at 22 http://cee.engr.ucdavis.edu/faculty/lund/CALVIN/Report1/Appendices/AppendixA.pdf. 23 Huber, N. K. 1987. The Geologic Story of Yosemite Valley. U.S. Geological Survey Bulletin 24 1595. 64 pp. 25 Hundley, N. 1992. The Great Thirst: Californians and Water, 1770s-1990s. University of 26 California Press. Berkeley and Los Angeles, CA. 27 Jenkins, M. W., et al. 2001. Improving California Water Management: Optimizing Value and 28 Flexibility. Center for Environmental and Water Resources Engineering. Available at 29 http://cee.engr.ucdavis.edu/faculty/lund/CALVIN. 30 Labadie, J. W. 2004. Optimal Operation of Multireservoir Systems: State-of-the-Art Review. 31 Journal of Water Resources Planning and Management. 130(2): 93-111. 32 Landis J.D., M. Reilly. 2002. How Will We Grow: Baselines Projections of California’s Urban 33 Footprint through the Year 2100. Department of City and Regional Planning, Institute of 34 Urban and Regional Development, University of California, Berkeley.

26

1 Lund, J. R. et al. 2003. Climate Warming and California’s Water Future. Available at 2 http://cee.engr.ucdavis.edu/faculty/lund/CALVIN/ReportCEC/CECReport2003.pdf. 3 Muir, J. 1912. The Yosemite. The Century Co. New York. 4 New York City Independent Budget Office. 2000. The Impact of Catskill/Delaware Filtration on 5 Residential Water and Sewer Charges in New York City. Available at 6 http://cdec.water.ca.gov/, and http://www.ibo.nyc.ny.us/iboreports/watershedlet.pdf. 7 Accessed on August 2003. 8 Null, S. 2003. Re-Assembling Hetch Hetchy: Water Supply Implications of Removing 9 O’Shaughnessy Dam. UC Davis Master’s Thesis. 10 Poff, L. N., D. D. Hart. 2002. How Dams Vary and Why It Matters for the Emerging Science of 11 Dam Removal. Bioscience. 52(8):659-668. 12 Redwood City Public Works Services Department. 2003. Urban Water Management Plan. 13 Available at http://www.ci.redwood- 14 city.ca.us/services/works/pdf/UWMP_June_2003_Chapter4.pdf. Accessed on August 2003. 15 Ritzema, R. S., M. W. Jenkins. 2001. CALVIN Water Management Appendix 2C. Region 3 16 Results: San Joaquin and South Bay Area. Available at 17 http://cee.engr.ucdavis.edu/faculty/lund/CALVIN/Report2/Appendix2C.pdf. 18 Ritzema, R. S. 2002. CALVIN Climate Warming Appendix D. Hydropower in the CALVIN 19 Model. Available at 20 http://cee.engr.ucdavis.edu/faculty/lund/CALVIN/ReportCEC/AppendixD.pdf. 21 San Francisco Public Utilities Commission. 2002. Available at http://sfwater.org. Accessed on 22 March and August 2002. 23 Sierra Club. 2002. Hetch Hetchy Timeline. Available at 24 http://california.sierraclub.org/hetchhetchy/hetch_hetchy_timeline.html. Accessed on March 25 2002. 26 Turlock Irrigation District. 2002. TID Water Quality. Available at www.tid.org. Accessed on 27 June 2003. 28 U.S. Bureau of Reclamation. Hetch Hetchy: A survey of Water and Power Replacement 29 Concepts. Sacramento, CA. 1987. 30 U.S. Department of Energy. Hetch Hetchy, Striking a Balance: A Review of the Department of 31 the Interior’s Survey of Water and Power Replacement Concepts for Hetch Hetchy. 32 Washington D.C. 1988 or 1989.

27

1 Table 1. Storage in the Hetch Hetchy System

Capacity (TAF)

O'Shaughnessy 360

Lake Eleanor 28

Cherry Lake 268

New Don Pedro 570*

San Antonio 50

Calaveras 97

Lower Crystal Springs 58

Pilarcitos 3

San Andreas 19

Total Storage 1,454

*Space owned by the City and County of San Francisco Total Storage in New Don Pedro Reservoir 2

28

1 Table 2. Model Runs by Filtration Condition, Dam Status, and Year

Keep Filtration Avoidance Lose Filtration Avoidance

2020 2020 O'Shaughnessy Dam Retained 2020 Base Case 2100

2020 Remove O'Shaughnessy Dam Scenario modeled, produced with New Don Pedro inter-tie no new results 2100

2

29

1 Table 3. Average 2020 Deliveries, Scarcity, and Scarcity Cost

Base Case with With Without O'Shaughnessy Urban Regions O'Shaughnessy* O'Shaughnessy Dam Dam**

Annual Average Deliveries (taf/yr) 1,424 1,440 1,440

Annual Average Scarcity (taf/yr) 16 0 0

Annual Average Scarcity Cost ($K/yr) 15,290 0 0

Agricultural Regions

Annual Average Deliveries (taf/yr) 5,259 5,258 5,257

Annual Average Scarcity (taf/yr) 0 1 1.5

Annual Average Scarcity Cost ($K/yr) 0 5 11

* Constrained to current operating policies

** Results do not change with loss of filtration avoidance

2

30

1 Table 4. Average 2100 Deliveries, Scarcity, and Scarcity Cost

With Without O'Shaughnessy Urban Regions O'Shaughnessy Dam Dam

Average Deliveries (taf/yr) 1,948 1,948

Average Scarcity (taf/yr) 6 6

Average Scarcity Cost ($K/yr) 4,086 4,076

Agricultural Regions

Average Deliveries (taf/yr) 4,506 4,509

Average Scarcity (taf/yr) 753 749

Average Scarcity Cost ($K/yr) 75,466 74,754

2

31

1 Figure Titles

2 Figure 1. Map of San Joaquin and San Francisco Bay Regional CALVIN Model

3 Figure 2. Hetch Hetchy System Model Schematic

4 Figure 3. Total Annual Agricultural Scarcity

5 Figure 4. Seasonal Flows in the Upper Hetch Hetchy Aqueduct

6 Figure 5. Monthly Flows Through a New Don Pedro Inter-tie

32

1 Figure 1. Map of San Joaquin and San Francisco Bay Regional CALVIN Model

33

Cherry Creek / HETCH HETCHY SYSTEM Tuolumne Eleanor Creek River O'SHAUGHNESSY LAKE LLOYD / DAM / HETCH ELEANOR LAKE Kirkwood HETCHY RESERVOIR Powerhouse

Canyon Tunnel Holm Mountain Tunnel Powerhouse Lower Cherry Aqueduct

Early Intake Tuolumne River

Cherry Creek Tuolumne River Mountain Tunnel NEW DON PEDRO Priest Reservoir Sullivan & RESERVOIR & Woods Creeks POWERHOUSE Foothill Tunnel

GW Moccasin La Grange Dam Reservoir & GW Powerhouse

CVPM 12 CVPM 11 CVPM 12 Ag Urban CVPM 11 Ag San Joaquin Urban Pipelines 1,2,3

San Joaquin River

LEGEND Pulgas W ater Temple Ind: SC V Reservoir

Res: SCV Powerhouse

LOCAL River SAN FRANCISCO

al Springs RESERVOIRS Pipeline yst r Bypass Tunnel C Non-storage Node

GW Urban / Ag Demand

Ind: Treatment Plant SFPUC GW Groundwater Basin Res: SFPUC Return Flow

1 Figure 2. Hetch Hetchy System Model Schematic

34

80 70 60 f) 50 ty (ta

i 40 c r

a 30 c

S 20 10 0 1921 1931 1941 1951 1961 1971 1981 1991 Oct 1921 - Oct 1993

Without O'Shaughnessy With O'Shaughnessy

1 Figure 3. Total Annual Agricultural Scarcity

2

35

30

h) 25

ont Maximum

m 20 /

F Mean A

T 15 Minimum on ( i

s 10 r e v i

D 5 0 Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep

1 Figure 4. Seasonal Flows in the Upper Hetch Hetchy Aqueduct

36

30

h) 25 ont

m 20 /

F Maximum A

T 15 Mean n (

o Minimum i

s 10 r e v i

D 5

0 Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep

1 Figure 5. Monthly Flows Through a New Don Pedro Inter-tie