Probabilistic Forecasting offers an API that predicts the weather using a special method called probabilistic forecasting. Instead of just giving one possible forecast, like a regular weather forecast typically does, it gives a range of potential outcomes with a level of uncertainty. This type of forecasting is useful for applications that require a range of potential outcomes, such as risk management or decision-making.

Instead of providing the full spread of all the potential outcomes, the API provides a list of seven different percentiles – 5%, 10%, 25%, 50%, 75%, 90%, and 95%. These percentiles allow users to paint a picture of the potential scenarios that can occur along with their likelihood of occurrence.

For instance, the 25th percentile (temperatureP25) is the value below which 25% of forecasts fall. In the context of probabilistic temperature forecasts, the 25th percentile temperature forecast is (approximately) the value in the probabilistic ensemble in which 25% of the forecasted temperatures are cooler.

Underneath the hood, we are running’s advanced forecast model, called 1 Forecast, multiple times, generating a range of different potential outcomes each time – i.e. if we ran the model fifty times, we would get fifty different potential forecast outcomes. After running the model so many times, a list of possible scenarios is generated which can be used to show how likely different scenarios are to occur based on how often they happened across all the forecast runs. This is different from traditional weather forecasting which gives a single forecast, often referred to as a “deterministic forecast”.

The Probabilistic forecasting can also be combined with our different aggregations.

This is useful for businesses and other organizations that need to plan for different weather possibilities and make decisions based on the likelihood of different outcomes to occur. In some cases, an operation may be more sensitive to others and thus the mere possibility of the temperature dipping below 32°F at this time may mean something different to some than to others.



the value in the probabilistic ensemble in which 90% of the forecasted temperatures are cooler.
humidityPXXAvg/ humidityPXXMin / humidityPXXMax

the value in the probabilistic ensemble in which 75% of the average forecasted humidity is higher.