Sub-Hourly Weather Forecasts: Enhancing Short-Term Accuracy
15-Minutely Weather Forecasts Now Available!
Tl;dr: Weather forecasts now come in 15-minutely resolution for North America and Central Europe. If you're curious about this, you can check it out in the weather API documentation!
Fifteen-minute weather data within forecast models plays a pivotal role in enhancing short-term weather predictions. The ability to observe and analyze changes in cloud cover with sub-hourly resolution is particularly valuable for accurate solar irradiance forecasts, especially for photovoltaic installations.
One of the primary advantages of these rapid updates in forecast models is their ability to provide better predictions for short-term precipitation. This includes the ability to forecast the onset of rain within a 15-minute window, which can be vital for planning and decision-making.
Our 15-minutely forecasts are generated using high-resolution weather models from two renowned sources: the German Weather Service (DWD) and the United States' National Oceanic and Atmospheric Administration (NOAA).
The DWD ICON-D2 model offers 15-minutely data for Central Europe, boasting a spatial resolution of 2 km. It is recognized as one of the best short-term weather models, delivering exceptional forecast quality even in complex terrains like the Swiss Alps. The model has been optimized to predict the rapid development of thunderstorms on a small scale.
For North America, NOAA's HRRR model provides 15-minutely data at a spatial resolution of 1 km. This model receives updates every hour, allowing it to swiftly adapt its forecasts to evolving weather patterns and developing storms.
These models are rooted in numerical weather predictions (NWP), distinguishing them from other nowcasting approaches like cloud-motion-vectors (CMV). NWP models factor in a comprehensive array of atmospheric variables, including temperature, pressure, humidity, wind speed, and direction, at various altitudes and locations. Running NWP models demands substantial computational resources and advanced data assimilation techniques to incorporate real-time observational data for accurate initial conditions. We appreciate the open-data initiative of the German and American Weather services in making this valuable data accessible.
Presently, 15-minutely data is available from two weather models. However, we are looking forward to integrate more sub-hourly weather data from various national weather services as soon as they become accessible.
15-Minutely Data in the Weather API Documentation
Our Weather API now seamlessly incorporates 15-minutely data, accessible in the Weather API Documentation. Users can select "15-Minutely Weather Variables" to access forecasts for up to 3 days. In instances where 15-minutely data is unavailable, we use data interpolation to offer 15-minutely data coverage worldwide.
Current Weather Conditions
The updated 15-minute data is additionally employed to enhance the precision of current weather conditions.
Furthermore, you now have the flexibility to selectively choose specific current weather variables. In the past, the options were confined to temperature, weather code, wind speed and direction. With this latest enhancement, you can get current conditions for any desired weather variable.
Displayed below is a screenshot of the updated weather API documentation, showing the ability to individually select weather variables for current conditions.
Outlook
Looking ahead, we have exciting developments on the horizon, including:
Support for API calls covering multiple locations simultaneously
Seasonal weather forecasts extending up to 9 months into the future
Additional high-resolution data for the historical weather API
Stay tuned!
The HRRR 15 minute data I'm getting using OpenMeteo has drastically improved my local forecasts' accuracy, especially for overnight lows and surface level relative humidity (2 meter) within 2-4km of a mountain range. Whether it's the improved spatial resolution or the temporal resolution, it's fantastic to see such close agreement between the HRRR and observation from my local weather station. In charting these 15-min data, it's also really nice to see such smooth lines, with sub-hourly variations revealed. This is an incredible improvement for my use-case. Thanks so much!