The Solar Forecast Arbiter is a first-of-its-kind platform that analyzes solar forecasts, comparing them against a standard so grid operators can better manage the amount of solar energy on the grid
Office of Energy Efficiency & Renewable Energy
May 23, 2022
minute read time
If a snowstorm is coming, we check the forecast to learn how much snow we will get and when it will start. Many of us even consult multiple weather apps in pursuit of the most accurate information, only to despair when we find varying predictions.
Utility companies and grid operators face the same uncertainty when trying to figure out how much solar power could be generated on a given day. Forecasts vary, and some can predict only one to three hours ahead with accuracy. The industry needs a better way to look at forecasts to plan for available solar power.
In 2018, the U.S. Department of Energy (DOE) Solar Energy Technologies Office (SETO) awarded $1 million to the University of Arizona (UArizona) to tackle this challenge. The resulting Solar Forecast Arbiter (SFA) is a first-of-its-kind platform that analyzes solar forecasts, comparing them against a standard so grid operators can better manage the amount of solar energy on the grid.
UArizona and its partners developed the SFA to help utilities and grid operators compare and evaluate the accuracy of solar power production forecasts from different providers so they can make better decisions about their energy mix. UArizona has operated the SFA on behalf of utilities and grid operators; now the nonprofit Electric Power Research Institute (EPRI) will manage its operations.
While the energy reaching solar panels is predictable under clear skies, it is difficult to calculate how much energy will make it to the ground surface when there are clouds. When a cloud passes over a solar panel, the amount of electricity the panel generates fluctuates. How much it changes depends on the cloud’s size, thickness, and shape, and other atmospheric factors. The SFA enables users to assess the reliability of solar generation estimates over the next 24–48 hours and the likelihoods of the predicted power outputs.
Before DOE funded this project, there was no transparent and uniform way to compare forecasting methods or tools. UArizona designed the SFA as an open-source platform so anyone in the field can access benchmark data and unbiased metrics to evaluate forecast models. The software can help forecast vendors improve the accuracy of their forecasts, too.
Recently, teams competing in the American-Made Solar Forecasting Prize submitted day-ahead solar forecasts to the SFA every day for four weeks. The SFA calculated the models’ performance against a benchmark forecast that the platform generated. The prize administrator, the National Renewable Energy Laboratory, used the SFA results to evaluate each team. SETO expects the competition will raise awareness and increase adoption of the SFA.
As UArizona concludes its DOE-funded project in the next few weeks, EPRI will take over stewardship of the SFA, offering a tiered user subscription. While anyone can access the platform and its code at no cost (after signing a user agreement), premium SFA subscribers will have access to hands-on support and training, the ability to create reference forecasts, and a professional network.
EPRI researchers will work to expand the SFA to support wind energy and power demand forecast evaluation, and plan to use the tool to support forecasting trials. When this feature is ready, the renamed Forecast Arbiter will have a comprehensive multi-technology evaluation platform that can better predict and integrate clean energy into the U.S. electricity grid.