·
2 minute read
We updated the Foreca Weather API with ensemble forecast statistics for daily temperature extremes and precipitation. This allows developers and product owners to display forecast uncertainty and confidence levels directly inside consumer-facing apps. Here is what ensemble statistics mean in practice and how to start using them.
What are ensemble forecasts?
Weather forecasts always come with uncertainty, and ensemble forecasts are a way to estimate that uncertainty. In ensemble forecasting, the same numerical weather prediction (NWP) model, which simulates the behaviour of the global atmosphere, is run multiple times with slightly different initial atmospheric conditions. These runs are called ensemble members. Sometimes the members behave similarly for many days of simulation, which gives us great confidence in the forecast. Other times, they spread more, which indicates more uncertainty in the forecast.
What data is available?
The Daily endpoint in the Foreca API now provides ensemble statistics for high and low temperatures and accumulated precipitation. The data comes as a set of percentiles, which describe the spread of the ensemble members:
|
Weather Parameter |
Available Percentiles |
|
High & Low Temperature |
0th, 25th, 50th (median), 75th, 100th |
|
Accumulated Precipitation |
10th, 25th, 50th (median), 75th, 90th |
To give an example, if the 25th percentile value for a day’s low temperature is 20 degrees, this means that only 25% of ensemble members predict a temperature lower than that. In other words, we say that the actual low will be less than 20 degrees with a 25% probability. And if the value of the 75th percentile is 22 degrees, then the temperature range of 20 to 22 degrees estimates a range with a 50% likelihood.
The 0th and 100th percentiles are the most extreme ensemble members. For precipitation forecasts, these values are not provided because they are not very meaningful: the lower extreme would almost always be zero.
Sometimes the deterministic values (minTemp, maxTemp and precipAccum) are different from the ensemble median, and occasionally, they even fall outside the ensemble range. This happens because the single values are produced a bit differently than the probabilistic ensemble forecasts. The deterministic values are affected by different data sources and the most recent data, and even the analysis of a meteorologist.
How can I make the most out of this new feature?
Here is a practical application from our Foreca Weather app. The 15-day meteogram uses ensemble statistics to display temperature and precipitation trends and uncertainty in a very informative way. For temperatures, the curve in the middle is the ensemble median, and the dark and light shades correspond to the 50% and 80% probability ranges. Precipitation has the median and 80% bars. With just a quick glance, it’s possible to see the overall trend and observe how confidence decreases after four days in this particular situation.
Start Using Ensemble Statistics Today
Ensemble data is included in all active Foreca Weather API subscription plans at no extra cost.
-
Building right now? Test the data structure directly in our developer documentation or grab your free trial key on the Foreca Developer Portal.
-
Evaluating for a commercial project? Our team is here to help you choose the right parameters and data models for your app. Contact our team for a quick consultation.
