With help from a supercomputer in Cheyenne, researchers have developed a new solar energy forecasting system that could help utilities integrate more renewables and save money.
Renewable energies like wind and solar are inherently variable, which makes it difficult for utilities to predict when they will be available. The new forecasting system from the National Center for Atmospheric Research combines ground-level observations, satellite data and artificial intelligence to make predictions about when solar energy will be available from 15 minutes to several days in advance, and is up to 50 percent more accurate than current forecasting models.
“Utilities need to have these forecasts so that they have confidence in deploying solar,” said Sue Haupt, the principal researcher on the project. “In combination with the cost of materials coming down, the cost of permitting is coming down, all of this works together so that more solar power can be deployed and that could make a big difference in the type of energy our country uses in future.”
Several utilities and private weather forecasting companies are already using the system, including Xcel Energy in Colorado.