DWR 2006 study: “too much risk” taken with reservoir operations

In 2006, for the first California Climate Assessment, DWR published a study of operations of the State Water Project and Central Valley Project with climate change. The report was titled, “Progress on Incorporating Climate Change Into Management of California’s Water Resources.“

DWR’s 2006 modeling predicted the current crisis, with Shasta, Folsom, and Oroville reservoirs being drawn down to dead pool. DWR’s modelers noted, “[t]oo much risk was taken in the delivery allocation decisions… and not enough storage was carried into the drought periods as a result.”   With respect to shortages, they stated:

The other type of shortage is usually unacceptable. This is when the first priority obligations – prior right contracts, minimum in-stream flow requirements, Delta requirements – are not met. The only way for this shortage to occur in CalSim-II is for one or more North-of-Delta reservoirs to be drawn down to dead storage. At this point, the model has lost control of meeting the watershed’s most basic needs not to mention the lawful obligations of the CVP and SWP. Such a simulation is broken.

This is exactly the situation we are facing in 2021.  It’s worth reading the entire section, since it explains the limitations of the CALSIM II model that has been used for water supply planning for the last 15 years.  DWR’s modelers testified in 2016 in the WaterFix Water Right Change Petition Hearing that the CALSIM II water use prioritization remains unchanged.

————–

4.6.1 Shortages

To discuss CalSim-II shortages, we must first discuss water use priorities. There are many competing demands for the water that flows into the Central Valley. They include farm irrigation, urban and industrial use, ecosystem protection and restoration, and reservoir storage for hydropower production, recreation or for later use in the next inevitable drought. In CalSim-II, distribution of water is prioritized as listed in Table 4.12.While CVP and SWP contractor deliveries take precedence over next year’s storage, a balance between the two is struck in the allocation decision. During the winter and spring, the SWP and CVP decide how much of contractor demand can be met for the year based on available storage and forecasted runoff. Part of the allocation decision is to ensure that enough water is left in storage at the end of the year in case of impending drought. Once the allocation decision is made though, deliveries to meet that allocation take priority over maintaining the storage carryover target.

Given this simple explanation of prioritization, there are two types of shortages in CalSim-II. One is an acceptable, though not desirable, result of making water allocations based on imperfect forecasts. In wetter years, the SWP and CVP sometimes allocate more south-of-Delta (SOD) deliveries than can be delivered through the pumps due to various export constraints. For the base and four climate change scenarios, this type of shortage is infrequent and, compared to total annual deliveries, insignificant. This type of shortage is also implicitly included in the delivery analysis; if it’s not delivered, we don’t count it.

The other type of shortage is usually unacceptable. This is when the first priority obligations – prior right contracts, minimum in-stream flow requirements, Delta requirements – are not met. The only way for this shortage to occur in CalSim-II is for one or more North-of-Delta reservoirs to be drawn down to dead storage. At this point, the model has lost control of meeting the watershed’s most basic needs not to mention the lawful obligations of the CVP and SWP. Such a simulation is broken. The lower priority metrics are questionable: Could the shortage of high priority water uses be avoided at the expense of lower priority uses through some simple changes in operating rules? And the results of a broken simulation can not be confidently compared to an unbroken simulation.

Table 4.13 shows that Shasta and Folsom reservoirs were at dead storage for a significant number of months in scenarios GFDL A2, PCM A2, and GFDL B1. These months are all concentrated in the critical year of 1924 and the droughts of 1929-1934, 1976-1977, and 1987- 1992. During these months, streamflow requirements were not met on the Sacramento and American rivers and the CVP was unable to contribute its Coordinated Operation Agreement  defined share of in-basin use. The base scenario had one month of shortage on the American and Sacramento rivers – October 1977. Due to the severity of the 1976-1977 drought, this is frequently unavoidable in CalSim-II simulations.

The length of shortages in GFDL A2, PCM A2, and GFDL B1 indicate that the delivery results presented for these scenarios in the next section are not always reliable. Too much risk was taken in the delivery allocation decisions of these three scenarios and not enough storage was carried into the drought periods as a result. In future climate change simulations, modifications to the rule that divides available water into delivery and carryover should be investigated as a means to prevent these shortages. Since CVP allocations are dependent on Shasta and Folsom storage, such modifications will likely alter the resulting delivery capability of the CVP as compared to the results presented in the next section.