Embedded Files
All data on this page sourced from ECMWF
Ensemble forecast charts
Ensemble forecast charts
Move to next or previous forecast lead time using the arrow keys or by swiping left or right.
About the ensemble charts
About the ensemble charts
Introduction
Introduction
The ECMWF (European Centre for Medium-Range Weather Forecasts) produces a fifty member ensemble, plus a control forecast. These ensemble members represent a range of possible future weather outcomes. An example of just two members is shown below.
Examples of two ensemble members from the 2025-05-01T06:00Z run, both valid for 2025-05-01T18:00Z.
An important consideration is how the weather forecaster should interpret this deluge of data coming from the ensemble forecasting system. They could look at all fifty possible scenarios, but this would quickly become overwhelming.
Ensemble mean valid at 2025-05-01T18:00Z
A common approach is to compute the average of the ensemble, as shown on the left. However, it is immediately apparent that the ensemble average smooths out the forecast, and removes extreme values.
Furthermore, different ensemble mean fields (e.g. wind, temperature and rainfall) are unlikely to be physically consistent with each other.
An alternative method is to try and select ensemble members which are representative of a handful of possible weather scenarios. This is the approach taken to produce this experimental product.
There are numerous ways to divide the ensemble members into different weather scenarios. Initially I tried classifying the ensemble members based on the average CAPE (convective available potential energy) in the Tonga region. The premise being that higher CAPE values might be associated with more severe weather (thunderstorms and so on). However, I was not convinced this was helping separate out different weather scenarios. So I have simply used the average rainfall in the Tongan region as a proxy for different weather scenarios - high rainfall in the Tonga region may be associated with different weather scenarios than low rainfall in the region. And in any case, rain is a parameter which is of interest to many forecast users.
Method
Method
For the experimental products on this page, the rainfall for the preceding six hours is averaged over a domain surrounding Tonga (indicated by the box in the plots) for each of the fifty ensemble members. The box extends from 25°S to 15°S and 180°E to 170°W. The fifty ensemble members are then ordered based on the average rainfall value within this box. The plots show the wind and rain fields for the individual members with the lowest rain, 10th, 25th, 50th, 75th and 90th percentiles of rain, and the maximum rain. Additionally, the ensemble mean and control forecast are also shown.
Feedback
Feedback
This product is experimental. I would be interested to get feedback on the method or suggestions for different ways of displaying the data.
Page updated
Report abuse