California isn’t running out of water; it’s running out of cheap water

by Wyatt Arnold

A California water myth which becomes especially pernicious in droughts is that California is “running out of water” (Hanak et al. 2009). Viewing California’s supply and demand pressures in terms of fixed water requirements perpetuates this myth and invariably places undue attention on building additional supply infrastructure. Instead, managing water as a scarce resource suggests a balanced portfolio of water trading, investments in conveyance, smart groundwater replenishment, and demand management. With such a balanced portfolio, 1) California’s water supply situation is not broadly dire, and 2) California’s vast and interconnected water infrastructure and groundwater resources can minimize most problems from the state’s highly variable climate.

An economics-driven model of California’s water system, the California Value Integrated Network (CALVIN), has provided such insight from several perspectives, including climate change, groundwater, water markets, and reservoir operations. But in many of these studies, authors lamented an unrealized potential to capture the impact of hydrologic variability more realistically. With perfect foresight, CALVIN was run with complete foreknowledge of 82-years of hydrology – giving exactly optimal solutions to managing reservoir over-year (“carryover”) storage through multi-year droughts, for example. Now, with access to high performance computing resources, a limited foresight carryover storage value function (COSVF) method (Draper 2001) has been applied to California’s entire system – more than 26 surface reservoirs and over 30 groundwater basins (Arnold 2021, Khadem 2018). These model runs are the most comprehensive and realistic analyses of the potential for broad integrated portfolios of actions across water agencies to address California’s water supply problems.

So, what do these new limited foresight CALVIN results tell us about California’s water supply? Here are three things to get started:

From the standpoint of long-term average marginal economic value of water, perfect and limited foresight closely agree; however, limited foresight is more relevant for the risk averse, who prefer to minimize larger but rarer shortages at the cost of average performance.

Limited foresight results also suggest that, in general, increasing the storage capacity of reservoirs in California has a very low marginal economic benefit relative to other infrastructure investments like conveyance and groundwater pumping capacities.

A large range of carryover storage and conjunctive use operations yield similar statewide economic performance (summing water operation and scarcity costs statewide over 82 years of wet and dry conditions). Consideration of a broad portfolio of conjunctive use, trading, water conservation, and local infrastructure options may not significantly change major surface reservoir operations.

Economics of Carryover Storage

Many reservoirs in California have been drawn down to near record low levels in the current drought. People are alarmed that reservoir storages are so low after only two dry years and are speculating whether prudent decisions were made about storage management. Climate change is making these decisions riskier, yet modeling the historical record remains important as a frame of reference.

Here, I focus on Shasta, Oroville, and off-stream San Luis carryover operations suggested by the new limited foresight optimization results and compare with carryover storage simulated by two versions of the State’s reservoir system model, CalSim-II, to shed light on how and why the system’s carryover storage is so volatile.

Figure 1 shows the time series of carryover operations modeled through two of California’s notorious droughts. The first thing to notice is that total carryover storage varies a lot from year to year for all models. Second, all models tend to quickly draw down carryover storage in dry years – only perfect foresight CALVIN, knowing the exact length and depth of drought in advance, maintains and draws down carryover storage in the final year of drought. Third, CalSim-II simulated carryover storage – both the Water Storage Investment Program run and the recent Delivery Capability Report 2019 run – lie just above the sampled range of limited foresight runs, suggesting near-optimal operations (blue shaded region) during dry years and droughts.

Without a crystal ball, it is not economical to maintain too much carryover or drought surface water storage. The probability of refilling the following year is high, both due to the volume of carryover storage capacity relative to annual runoff and the low year-to-year correlation of annual runoff. Lower carryover storage raises groundwater pumping in the latter year(s) of drought, but also reduces average groundwater use and pumping costs and helps reduce groundwater overdraft (see Figure 2). Also, higher reservoir releases tend to reduce shortages where access to groundwater is limited, which lowers average shortage costs. Sustaining higher carryover volumes (upper end of the limited foresight range and CalSim-II alike) provides more surface supply in the latter year(s) of a drought, which reduces maximum shortage costs; however, the more risk-averse operation raises long-term average costs and the marginal value of groundwater that would eliminate overdraft. Limited foresight modeling (both optimization and CalSim-II) tends to use more groundwater in drier years (Figure 2), which points to groundwater’s importance as a buffer against hydrologic uncertainty.

**Figure 1.** Total carryover storage (ending September) of Shasta, Oroville, and San Luis as modeled by perfect and limited foresight CALVIN and CalSim-II Delivery Capability Report 2019 (CS-II DCR 2019) and Water Storage Investment Program (CS-II WSIP) historical runs. Drought periods of 1976-77 and 1987-92 are shaded in light tan. The limited foresight range is based on 26 near-optimal statewide solutions.

**Figure 2.** Total groundwater pumping volume in the Central Valley as modeled by CALVIN for perfect foresight, limited foresight, and with reservoir carryover storage fixed to CalSim-II Water Storage Investment Program historical run outputs.

Other Considerations for Carryover Storage

Water supply is not the sole objective of carryover storage operations. Federal and State operators of Shasta and Oroville reservoirs seek to maintain storage reserves for environmental requirements. For example, Shasta’s carryover storage objectives include maintenance of a cold-water pool to support Salmon habitat in the Sacramento River. Other economic objectives include recreation and hydropower. CalSim-II’s higher carryover storage relative to limited foresight CALVIN are partially attributable to these objectives in addition to Federal and State contractual water supply obligations to Sacramento and Feather River water rights holders. While CALVIN incorporates minimum environmental flow constraints, more complex environmental requirements such as cold-water pool management and some Delta operational constraints are less well represented. Nevertheless, the limited foresight CALVIN results provide a more realistic representation of the economic value of carryover storage in California’s multi-reservoir conjunctive use system.

Concluding Thoughts

Aggressive use of carryover water storage in California’s major reservoirs is economically prudent and reduces overall groundwater reliance. Water supply risks of lower carryover storage are further mitigated through greater system integration such as increased water trading, groundwater banking, and drought water use reductions. The higher risks of having low carryover storage, although not quantified here, appear to fall on California’s stressed ecosystems. A warming climate, expected to continue through at least mid-century even with aggressive global greenhouse gas mitigation, is changing runoff timing, magnitude, and frequency in ways that will make managing carryover storage more challenging. Future work should focus on this aspect and incorporate alternative hydrologic traces reflecting expected climate changes.