How to use Ensemble models for weather forecast

How to use Ensemble models for weather forecast

What is a forecast model? 

The weather depends on hundreds of factors: wind speed, cloudiness, atmospheric pressure, humidity, precipitation, terrain, etc. All of these factors can be quantified: meters per second, millimeters of mercury, etc. If you put these variables into an equation, you can create a mathematical weather model.  

A forecast model may be wrong. It is impossible to take into account all of the atmospheric factors. Also, there could be an accidental error during data input. For instance, a wind speed of 5.6 m/sec was perceived as 5.7 m/sec by the sensor. 

As you can see, the forecast does not always predict the events correctly. This is why the meteorologists use a so-called probabilistic approach in forecasting-they examine the ensemble models.

 

How do know if the weather model for today can be trusted? 

Just imagine that we are using a forecast model, for instance, the European ECMWF. We input the data on humidity, temperature, etc and received the weather forecast for 10 days. Here it is on the chart (pic.1). Can we trust it?

Enseble model

Picture 1

Meteorologists have come up with a clever solution-they use the same model all over again but input slightly different data. For example, the wind speed was 3 m/sec, but they input it as 3.3 m/sec. This way the parameters of the models are slightly modified. 

Given these new parameters, they develop a new weather forecast. Then repeat this 40 times getting 40 different forecasts (pic.2).

Picture 2

If the resulting forecasts differ significantly, it means that we can’t trust our initial assessment, because a slight change of weather (which is entirely possible in nature) may lead to a completely different result. Hence, the model would not be helpful.

However, if the forecasts are more or less similar, it means that even if the atmospheric conditions change somewhat, the models are still trustworthy. 

How to use Ensemble model in Windy.app? 

The ensemble model in our application is called ECMWF+. You can take a look at the ECMWF+ model on any spot. Now iOS only, Android soon.

weather forecast

Ensemble model in Windy.app

If the lines on the graph are spaced widely apart, then you should not rely on the forecast too much. And if the lines are close together-the determined forecast (the main forecast according to the ECMWF model) is reliable. 

Do make sure to check the Ensembles, if you are looking for a more than a 3-day forecast. 

You could also use the ensemble models if you want to look at the potential range of temperatures or wind speeds. 

For instance, you work on a farm. Might the temperatures drop below freezing the following week? The Ensembles will provide you with the possible scenarios. 

The feature is available to all PRO-users. Try it now!