Horizon AI Point

Horizon AI Point is Climavision's automated point-forecasting system for organisations whose operations depend on the weather at specific sites, such as electric utilities, energy traders, insurers, airports, farms and logistics hubs. Forecasts built only from government models often miss local conditions, which leads to unplanned outages, costly energy waste and less reliable demand forecasting.

The system starts from an ensemble of more than 100 individual forecast models, including GFS, NAM, HRRR and ECMWF, and adds Climavision's own Horizon AI Global and HIRES models. Machine-learning bias correction, trained on historical data and improved continuously, tunes the blend for each location. Its inputs include local sensors, satellite data and Climavision's proprietary numerical weather models and observations.

• Coverage: global, at any point the customer adds, from 0 to 15 days ahead, updated hourly.
• Outputs: more than 40 parameters, including temperature, wind and hourly power-production forecasts for energy assets.
• Delivery: an interactive map (DASH), GRIB2 files and a REST API that returns JSON.
• Integrations: Arcobi (formerly Arcus Power) offers Horizon AI Point forecasts in its Nrgstream power-market data platform for energy traders.

In Climavision's own analysis of the January 2024 ERCOT cold snap, Horizon AI Point predicted the extreme cold 6 to 7 days in advance. Climavision offers a four-week free trial.