Fatal errors in DWR’s runoff forecasting: comments to the State Water Resources Control Board

The State Water Resources Control Board held a workshop today (March 16, 2022) on Sacramento River Temperature Management. Among the topics considered were “hydrologic conditions, including operations, forecasting, and accounting.” These were my comments.

This is Deirdre Des Jardins with California Water Research. I’m the physicist who sent comments to DWR in 2012 predicting the 2013 to 2016 drought. There had been a once in a millennium drought in Australia, a once in a millennium drought in the Southwest, and exceptional droughts around the world. It was clear that the climate was changing.

We filed a protest in 2021 asking that the [Water] Board mandate that the projects produce a report on the reasons for the errors in the runoff forecasts, and hold a workshop on the forecasting methodology. It was clear then that DWR’s Bulletin 120 forecast methodology had fatal errors. It’s even more clear now.

In 2009, DWR’s own water resources engineers said that stationarity is dead. DWR’s operators knew that their runoff forecast model failed last year, yet they took huge risks in making a 15% allocation based on untested revisions to the models. I’m not doing Monday morning quarterbacking on this, I wrote a blog post about it on January 20.

The Board needs to remember that neither the Department of Water Resources nor the US Bureau of Reclamation have the right to a minimum 1500 cfs diversion from the Delta. The projects need to manage risk so that they have enough water in storage to meet their minimum export needs, as well as their in-basin obligations. They are not doing so.

As far as the Bulletin 120 forecast methodology, my understanding is that the forecaster picks out surrogate years from the historic record, and uses a linear regression model based on those years. Changing to a 30 year record from a 50 year record simply isn’t going to work. The climate is changing to rapidly.  Similarly, no machine learning algorithm based on these inputs is going to work.

DWR needs to move to a physics-based model, such as the one being used by NOAA’s California Nevada River Forecast Center.

Finally, I will note that the Coordinated Operations Agreement for the projects has no sharing formula for the obligation of the projects for storage release to meet in-basin obligations in a third or fourth dry and critically dry year. The Board reserved jurisdiction over the coordinated operations of the projects, and needs to exercise it.

Further reading

Des Jardins, D. 2022. Dear DWR, in runoff forecasting, stationarity is dead. California Water Research blog. February 17, 2022.

Des Jardins, D.  2022. In increasing State Water Project allocations, DWR is taking huge risks. California Water Research blog. January 20, 2022.

Des Jardins, D. 2021. TUCP: California Water Research asks Water Board to require report on 2021 runoff forecast errors. California Water Research blog. June 5, 2021.

Des Jardins, D. 2012. Incorporating Drought Risk from Climate Change into California Water Planning. August 2012.

National Oceanic and Atmospheric Association, National Weather Service: California Nevada River Forecast Center. [accessed March 17, 2022.] https://www.cnrfc.noaa.gov/

Obegi, D. 2021. Get Ready for Climate Whiplash from DWR’s February Forecast, Natural Resources Defense Council expert blog. January 31, 2022.

United States of America and California Department of Water Resources. 2018. Addendum to the Agreement for Coordinated Operation of the Central Valley Project and the State Water Project. December 2018.