A Utility ‘Crystal Ball’ for Weather

Share this article:

With Strategically Placed ‘Mesonet,’ Ameren Missouri Reduces Forecast Error by 40%

EPRI, Ameren Missouri, and Saint Louis University demonstrated that carefully designed mesonets coupled with customized models can provide utilities with weather forecasts superior to traditional sources alone.

Forecasts from those sources, such as the National Weather Service, The Weather Channel, and AccuWeather, do not provide local detail necessary to guide crew positioning ahead of severe weather. Regionally configured mesonets can potentially provide more precise data to weather- and damage-forecasting models, enabling better utility response.

Using Saint Louis University’s QuantumWeather® algorithm, researchers determined the optimal placement of more than 100 weather stations for a mesonet in Ameren Missouri’s service territory, with 1.2 million electric customers. Over the past several years, weather data from the mesonet enabled a 40% reduction in average forecasting error relative to the National Weather Service. More than 700 storm simulations validated this improvement. Other findings:

  • Forecasts derived from properly configured weather sensors are more precise with respect to storm intensity and location.
  • Areas with the greatest population density and load generally require more sensors than rural areas.
A weather station deployed in the service territory of Ameren Missouri. Photo courtesy of Ameren Missouri. [Click to enlarge]

Recently, researchers used the algorithm to propose optimal mesonet configurations in portions of the service territory of American Electric Power. The utility desired 24-hour notice for conditions producing more outages than can be handled by their regular crews. This research pointed to the importance of considering a region’s weather history and topography along with a utility’s unique response needs.

Key EPRI Technical Experts:

Doug Dorr
For more information, contact techexpert@eprijournal.com.

Artwork by Kirk Anderson