Advanced Statistical-Dynamical Downscaling Methods and Products for California Electricity System Climate Planning

Optimizing physical models of climate to inform climate impacts across California.

University of California, San Diego Scripps Institution of Oceanography 0955


La Jolla, CA

Recipient Location


Senate District


Assembly District



Amount Spent



Project Status

Project Result

The project has ended and the Final Report is in review. The research team has run dynamic regional climate models and exploring the use of statistical models for hourly simulations. In the past, climate scenarios for CA only included projections with daily resolution. The research team has made significant progress on key areas such as: development of dataset variables that focus on low clouds, fire weather, and wind generation, with input from other CEC project(s); comprehensive verification of downscaled clouds, wind, and near-surface temperature; integrated hydrologic modeling using machine learning for building statistical models of hydrologic quantities through watersheds; and, merging multiple precipitation products to improve simulated hydrologic fluxes. The project also completed model simulations to replicate historical conditions such as coastal clouds, wind, and humidity.

The Issue

There are two basic ways to produce climate scenarios for California. One of them involves the use of dynamic regional climate models. These "weather forecast models" are very expensive to run. The second option is to use statistical methods that use historical relationships with outputs from global climate models to create high resolution climate scenarios for California. This approach is far less expensive than running an entire weather forecast model, but it is unclear if the historical statistical relationships will be valid under future conditions. The researchers are developing and testing a hybrid downscaling technique that merge the benefits of statistical and dynamic models.

Project Innovation

This project develops new and better ways of merging two modeling approaches, using both weather forecast models (more generally called dynamical models) and inferences from past history (statistical models). The combined method is called a hybrid dynamical-statistical approach for inferring fine-resolution climate information from the coarse-resolution global climate models. Ideally, the hybrid approach will be able to capture many of the physical processes simulated by the costly weather forecast models, but with the reduced expense of statistical models. The hybrid approach will be applied to three key areas of California's climate that have important implications for the state's ratepayers: wind, clouds, and hydrology.

Project Goals

Develop hybrid downscaling using a statistical approach employing results from regional dynamical modeling.
Develop statistical downscaling to provide hourly estimates of temperature, wind and humidity.
Explore dynamical model skill in simulating coastal low cloudiness and related properties.
Investigate "Simulator for Hydrologic Unstructured Domains" simulations of Sacramento Valley hydrology.
Investigate performance of ParFlow.CLM model in simulating surface and groundwater hydrology in California.

Project Benefits

The project includes an extensive quantification (model validation) effort based on data from observed meteorological stations, satellite records of cloudiness compiled by project members, and USGS streamflow and groundwater observations (for the hydrologic modeling). The method under development could be used for the California's Fifth Climate Change Assessment and future energy planning.

Lower Costs


Knowing how the climate is likely to change provides a sound scientific basis for minimizing economic impacts on the electricity system.

Greater Reliability


The methods to produce high-resolution projections of climate parameters are of great importance for managing the electricity system, in particular for managing peak demand and shifting to a grid reliant on renewable resources.

Increase Safety


This research supports predictive modeling, providing information on how the climate is likely to change, which can be used to limit impacts to residents, infrastructure, and the economy.

Key Project Members

Daniel Cayan

Daniel Cayan

Research Meteorologist
David Pierce

David Pierce

Data Systems Analyst
Scripps Institution of Oceanography, University of California San Diego



The Regents of the University of California, Davis


The Regents of the University of California, Riverside

Contact the Team