Following the preview and migration program announced earlier this year, Visual Crossing’s next-generation weather forecast engine is now fully available.
The new forecast platform represents one of the largest upgrades to our forecasting infrastructure to date. It expands the number and types of weather models available to the Visual Crossing Weather API, improves how those models are combined for different locations, reduces the delay between the release of new model data and its availability through the API, and introduces new tools for understanding forecast agreement and uncertainty.
These improvements are delivered through the existing Visual Crossing Timeline Weather API, so applications can benefit without migrating to a new API.. Existing applications can continue to request a single blended forecast without needing to understand or manage the individual forecasting models behind it.
At the same time, users who need greater meteorological detail can now inspect individual forecast models, compare their predictions, and retrieve information describing how strongly the available models agree.
A broader multi-model weather forecast
No single weather forecast model performs best everywhere or under every weather situation.
Global models provide broad and consistent coverage, while higher-resolution regional models can provide important additional detail for shorter-range forecasts. Ensemble systems add information about the range of possible outcomes, and newer AI forecasting systems provide another independent source of forecast information.
The new Visual Crossing forecast engine brings these sources together into a common forecasting platform.
Depending on location and forecast period, the system can incorporate global, regional, ensemble and AI forecasting systems including ECMWF, GFS, ICON, HRRR, UK Met Office models and other supported forecast datasets.
Rather than applying a single static model combination globally, the forecast engine evaluates the performance and availability of models for different regions and forecast horizons. This allows the Visual Crossing Blended Forecast to place greater emphasis on the forecast information most appropriate for a particular location.
The result remains what most applications need: one consistent weather forecast available worldwide through a single API.
Faster access to new forecast model runs
The new platform also changes how forecast data moves through our infrastructure.
Forecast models are continually being updated by national weather agencies and forecasting centers around the world. Our new ingestion architecture processes those model runs as they become available rather than relying on a small number of large scheduled processing cycles.
This substantially reduces the time between the publication of a new forecast model run and its inclusion in Visual Crossing forecast data.
It also allows the forecasting system to work with models that update at very different frequencies, from large global forecast systems to rapidly updating regional and high-resolution weather datasets.
Compare individual weather forecast models
While the Visual Crossing Blended Forecast remains the recommended forecast for most applications, the new forecasting infrastructure also makes the underlying models accessible.
Users can retrieve supported individual forecast models through the Timeline Weather API using the forecastDataset parameter.
This makes it possible to compare models such as ECMWF, GFS, ICON and regional forecasting systems against each other and against the standard Visual Crossing blended forecast.
Individual model access can be useful when:
- investigating an unusual forecast
- comparing global and regional forecast models
- evaluating emerging AI forecasting systems
- examining differences between successive model runs
- building custom forecasting or decision-support systems
Our Forecast Model Comparison Tool provides an interactive way to explore these differences without constructing API requests manually.
Understand forecast agreement with spread and confidence
Different forecast models, and different ensemble members within those models, can predict somewhat different outcomes. The amount of agreement between them provides valuable additional information about the stability of a forecast.
The new forecast engine therefore introduces forecast spread and confidence information alongside the blended forecast.
For supported weather elements, spread measures the weighted variation between the forecasts contributing to the blend. Lower spread indicates that the models are clustered relatively closely together, while higher spread indicates greater disagreement.
Spread is currently available for:
- temperature
- wind speed
- sea-level pressure
The accompanying confidence value provides a 0–100 summary of overall forecast-model consensus.
High model agreement does not guarantee that a forecast will ultimately be correct, but stronger agreement is generally associated with a more stable forecast. Applications can use this information to identify periods where the available forecast guidance is converging or where there is substantially greater uncertainty.
This makes it possible to move beyond simply asking “What is the forecast?” and also ask “How strongly do the available forecast models agree with it?”
A foundation for higher-resolution and specialized weather data
Beyond the forecast improvements in this release, the same infrastructure provides a foundation for integrating expanded weather information available alongside conventional forecast fields.
The platform can integrate high-frequency radar, satellite and specialized environmental datasets with numerical weather prediction models. These sources support improved short-range precipitation analysis and additional weather elements such as severe-weather indicators, air quality, wave conditions and other specialized datasets where coverage is available.
As new forecast models and observational sources become operationally useful, they can be incorporated into the platform without requiring customers to migrate to a new API.
Built for applications, not just meteorologists
One of the primary goals of the new architecture is to provide sophisticated forecasting capabilities without requiring every customer to become a numerical weather prediction specialist.
Developers can continue to request weather using a location and receive a normalized forecast through the Timeline Weather API.
Behind that response, the forecast engine can select and combine information from multiple forecasting systems, geographic resolutions, update schedules and forecast horizons.
Users who need more detail can progressively access it — comparing individual models, examining forecast spread, investigating model runs or working with specialized datasets — while simpler applications can continue using the standard blended forecast.
Existing API applications remain compatible
The next-generation forecasting platform has been designed around the existing Visual Crossing Weather API.
Standard Timeline Weather API queries and existing weather elements remain compatible, allowing customers to benefit from improvements to the forecasting system without rebuilding their integrations.
The same API can now support a spectrum of use cases ranging from a straightforward global weather forecast to detailed analysis of individual numerical weather prediction models and forecast consensus.
Explore the new forecast platform
The next-generation Visual Crossing forecast engine is now available through the Visual Crossing Weather API.
You can:
- use the Visual Crossing Blended Forecast for a regionally optimized multi-model forecast
- use the Forecast Model Comparison Tool to visually compare models
- retrieve individual weather forecast models through the Timeline Weather API
- request forecast spread and confidence to understand model consensus
- continue using existing Timeline Weather API integrations without changing your application architecture
This release completes the core forecast-engine migration we announced earlier this year, but it is also the foundation for continued expansion of the Visual Crossing forecasting platform.
Additional regional models, AI forecasting systems, ensemble information, higher-resolution datasets and forecast-analysis capabilities can now be incorporated into a common infrastructure and delivered through the same API.
For developers and businesses, that means access to increasingly sophisticated weather forecasting while retaining the simplicity of a single, consistent weather API.