Wind Forecast Improvement Project Saves Millions for Utilities and Customers

The wind doesn’t always blow where it’s needed—that’s the biggest hurdle in fitting wind energy to the nation’s renewable energy needs. When the wind isn’t blowing, utility companies must turn to other clean electricity generators, such as solar power or hydropower, or even to fossil fuels. The key to clearing this hurdle is accurate weather forecasts, but weather forecasting isn’t a perfect science. 

To help make weather forecasting more accurate, scientists at the Pacific Northwest National Laboratory (PNNL) have teamed up with the National Oceanic and Atmospheric Association (NOAA), as well as universities and private industry organizations, to improve weather forecasts. 

Through their work on the Wind Forecast Improvement Project (WFIP), the multiagency research has already helped save utility companies, and their customers, millions of dollars.

“Wind energy is clean and low cost, but it's one drawback is that it's dependent on the fuel, which, in this case, is wind. And wind is not constant,” said Raghavendra Krishnamurthy, an Earth scientist at PNNL and principal investigator for WFIP. “With more accurate wind forecasts at turbine heights, utility companies can more efficiently balance their power generation from various sources, like wind, hydropower, or fossil fuels, and save money.”

Forecasting Complications

Utility companies depend on weather forecasts to prepare for the next day’s electricity generation, and inaccuracies in weather forecasts can cost millions. 

If wind is overpredicted (meaning there was less wind than forecast), utilities must quickly pivot to other types of energy, which is costly and inefficient. If wind is underpredicted (meaning there was more wind than forecast), it can mean that utility companies are unnecessarily depending on other energy sources, such as natural gas, often at a higher cost.  

Forecasts come from the National Weather Service, which uses a model called the high-resolution rapid refresh model (HRRR). The model incorporates data from weather sensors all over the United States about variables, like humidity, air pressure, and air temperature, and uses them to predict wind for the next 48 hours.

But those variables can change based on where wind farms are located in the United States, which affects what kinds of weather patterns they experience. Some are located in areas that are dry, flat, and hot, whereas some are in cold, wet, and mountainous. Other wind farms are even placed in the ocean, which comes with a completely different set of temperature and humidity ranges than land-based wind farms experience.

WFIP helps model builders incorporate these regional nuances.

Wind Forecast Improvements

The team realized they had to study the weather across different regions and incorporate those findings to improve the model. 

“If you think of the model as a fishing net, and think of weather phenomena—like clouds and storms—as the fish, the only fish you don’t catch are the ones getting through the holes in the net. The denser the net, the more fish you catch,” said Larry Berg, division director for the Atmospheric Sciences and Global Change Division at PNNL and former investigator on the WFIP team. Studying regional data helps researchers understand what is making it through the “net,” and improve the model, which produces more accurate forecasts.

In the project’s first phase, PNNL scientists, along with partners at other U.S. Department of Energy national laboratories, NOAA, universities, and private industry organizations, took data from wind farms in northern Texas and the Great Plains in 2011–2012. In the project’s second phase, the WFIP2 team collected data in 2015–2017 from the Pacific Northwest’s Columbia River Gorge and basin. There, mountains tower over near-sea-level basins and the Columbia River has cut a canyon between rocky cliffs. 

Researchers at NOAA used these data to improve the HRRR model, releasing the first updated version (called HRRR2) in 2016 and another (HRRR3) in 2018. With WFIP’s contributions, HRRR’s updates have improved weather modeling and led to significant savings. According to 2022 a publication in the Bulletin of the American Meteorological Society, utility companies likely saved more than $95 million per year after NOAA launched HRRR2 and $32 million after launching HRRR3. 

An additional paper published in 2022 in the Journal of Renewable and Sustainable Energy found that the improved models had the potential to save U.S. consumers more than $380 million. 

“The WFIP campaigns, and in particular WFIP2, provided a unique dataset that enabled us to markedly improve our wind forecasts in the lower atmosphere,” said David Turner, an atmospheric scientist at NOAA and manager of the agency’s Atmospheric Science for Renewable Energy program. “We have demonstrated that, if the energy community only used the HRRR for their day-ahead decisions on energy generation, then they would have saved hundreds of millions of dollars per year using more updated versions of HRRR.”

The Future of Wind Forecasting

The WFIP team is already planning for the future of the project, with WFIP3 starting in the last few months of 2023 gathering data from the two wind farms off the northeastern coast of the United States. 

“Offshore wind data is very sparse, and therefore, we are not sure on the accuracy of the wind forecasts offshore,” Krishnamurthy said. “The next phase of WFIP will provide this necessary data, which will be made freely available to the research community and support the development of more accurate forecasts.”

WFIP is supported by the U.S. Department of Energy’s Wind Energy Technologies Office and NOAA’s Atmosphere Science for Renewable Energy Program.

A previous version of this story was published by PNNL.