Weather Methodology

Amperon developed a unique weather model that incorporates tens of thousands of hyper-local data points into our award-winning energy demand forecasts. This detailed weather data enhances the accuracy of our predictions, especially during critical outlier events, supporting your risk management strategy when it matters most.

Trusted by leaders in power & utility markets

How we approach weather

Weather data stands as one of the most important inputs for energy forecasting. Our models integrate historical weather observations and forecasts, employing diverse techniques to gauge their accuracy and sensitivity to local weather patterns and power consumption.

Tailored to meet specific customer requirements, our process involves running multiple machine learning models, assessing each model's performance, assigning weights, and aggregating them into a single demand forecast.

Weather sources: Short-term forecasts

Our short-term forecast models ingest historical weather observations and weather forecasts for 50+ parameters. The combined 40,000+ data points are preprocessed and weighted by population density before being integrated into Amperon's historical load models to generate a hyper-accurate demand forecast.

Weather sources: Long-term forecasts

The demand forecasting industry is constrained to a 15-day forecast due to availability of weather forecasts by vendors. To expand our forecast range up to 5-years, Amperon runs Markov simulations based on long-term historical weather observations of daily temperature and dew point from NOAA's Integrated Surface Database (ISD).

Collecting data from +40,000 weather points

Amperon's weather points are based on population density. While most forecasts use ISD weather data that usually correlates to a city's closest airport, Amperon's weather points are every 1km for urban areas and 5km for rural. If you need a forecast in a less populated area, we will work with your team to identify the optimal weather points for your organization. Contact us to see a map of our weather points for your area.

Our weather vendors

AG2

Atmospheric G2

DTN

Data TransmissionNetwork and Dataline

NBM

NOAA's National Blend of Models

ECMWF

European Centre for Medium-Range Weather Forecasts

In-House Meteorologist

Dr. Shipham, Amperon’s Chief Meteorologist, is a pivotal asset to our team. With nearly forty years of experience and a PhD in atmospheric chemistry and climatology, he has honed his predictive abilities through roles at esteemed organizations such as NASA and Dominion Energy. Known for his exceptional accuracy in forecasting extreme weather events, Mark is recognized as one of the foremost experts in America. As weather events become more frequent and severe, they increasingly influence the strategic grid adaptations that energy companies must prioritize.

Learn more

Ensemble Model

Amperon's Ensemble Forecasts optimally blends together the forecast models of four leading weather vendors (AG2, NBM, DTN and Spire) to mitigate the risk of relying solely on one weather forecast. The enhanced overall accuracy delivers a single demand forecast that is better than the sum of its parts.

Learn more

Advanced Weather

Amperon's Advanced Weather enables our customers to delve deeper than our base Weather Scenarios feature. The base scenario enables customers to conduct stress tests and scenario analysis with AG2 weather data up to one week out with temperature off-sets of up to +/- 5 degree F. Advanced Weather unlocks more extensive weather scenario analysis, aiding in trading decisions. You'll get access to individual demand forecasts based on weather forecasts from DTN, NWS’ NBM, and SPIRE (ERCOT only), the ability to modify weather scenarios up to two weeks out, and the ability to adjust temperatures by up to +/- 15 degrees F.

Learn more

Weather Methodology

How we approach weather

Collecting data from +40,000 weather points

More data points. More weather vendors. More accurate forecasts.