Improving Solar Forecasting

The development of an enhanced version of WRF-Solar

By improving the Weather Research and Forecasting (WRF)-Solar model, this project aims to reduce forecast errors of Global Horizontal Irradiance (GHI) and Direct Normal Irradiance (DNI) by 25%, yield better forecasts of irradiance ramps, improve estimates of sub-grid scale variability, and more accurately estimate forecast uncertainty. This enables solar power system operators to know how much solar power will be generated over the coming hours and days, ensuring economic and reliable delivery of renewable energy to American households and businesses.

Solar panels and open sky

Utilities, grid operators, solar power plant owners, and other stakeholders would like to better forecast when, where, and how much solar power will be produced at the desired locations in the United States.

Photo by American Public Power Association on Unsplash

This new system builds on the first version of the WRF-Solar model developed by the National Center for Atmospheric Research (NCAR).

Workflow diagram of WRF-Solar project aims
New model development in the WRF-Solar project aims to improve the representation of boundary-layer clouds, fine scale variability, cloud microphysics, and absorbing particles in WRF-Solar.

Anticipated improvements in Version 2 include the following:

  • New representation of boundary-layer clouds (both shallow cumuli and the breakup of stratocumulus) including the impact of entrainment
  • Improved treatment of cloud microphysics, and impacts of aerosol (including absorbing aerosol)
  • New parameterizations to account for the sub-grid temporal variability of solar irradiance during periods with broken clouds
  • Detailed analysis to better quantify model uncertainty and improved calibration of WRF-Solar v2 using Uncertainty Quantification (UQ) techniques