Daily weather by email
14 day outlook
Monthly outlook
Seasonal outlook
Christmas weather
Charts and data
Weather updates from site founder Brian Gaze that keep you in the know.
The dashboard covers the UK, Ireland and much of Europe, and supports four ensemble systems:
GEFS, IFS and AIFS cover 40°N to 63.5°N and 22°W to 36.5°E on a 0.5° grid. MOGREPS-G covers 36°N to 66°N and 10°W to 26°E on a 1° grid. However, the grid coverage and resolution may changed in the future, depending on demand and resource availability.
A selected location is matched to the nearest grid point, so some coastal locations may use a point slightly offshore.
As well as the traditional ensemble charts, there are five new views: Spread, Spaghetti, Scatter, Cluster and Scenario. The result is a more powerful way of examining not just what the forecast says, but how confident we can be in it and where the members are heading.
Take a look at the Ensemble Dashboard now.
An ensemble is a set of model runs, each starting from slightly different initial conditions. Comparing the members gives a better indication of uncertainty than a single deterministic forecast. For more information see, Why one forecast is never enough.
The box spans the middle 50% of members (the 25th to 75th percentile), with the median shown as a line through it. The whiskers extend to either the 10th and 90th percentiles or the full minimum and maximum, depending on the selected Whisker Mode.
The same information shown as shaded bands, with the median as a line. It is probably the easiest chart for following how the forecast evolves over time.
Each percentile is plotted as a separate line, making it easy to read off values and see how the distribution changes through the forecast period.
Focuses on the width of the ensemble. Rising lines mean the members are diverging and confidence is falling, while a narrower spread means greater agreement.
Every member is plotted as a faint line. This can reveal what a plume may hide: rather than simply broadening, the members may be following two or more distinct paths.
Each member is shown as a dot at each forecast time. Gaps between groups of dots reveal the ensemble splitting into different possible outcomes.
Groups members with similar values at each forecast time and displays them as bubbles, sized by the number of members. It gives a quick indication of whether the ensemble is converging on one outcome or splitting into several.
Takes clustering a step further, grouping members by their whole forecast path over a period you choose, such as Days 4–7.
Tools like these are usually found only in specialist or expensive forecasting systems. Cluster analysis answers a simple question: are the members actually agreeing, or are there several competing outcomes?
Members are grouped when temperature is within ±2°C, pressure within ±4 hPa and wind within ±2 m/s. Precipitation and CAPE use fixed categories such as Dry, Light, Moderate and Heavy.
The largest group is identified as the most likely outcome, but don't over-interpret small differences. With up to 50 members, groups of 25 and 23 are effectively a tie, and the tooltips flag when there is no clear favourite. Hovering over a bubble shows every group present at that time.
Choose the period you want to analyse, such as Days 4–7, and members are grouped by how similar their whole path is over that period. For temperature, members must stay within an average difference of 2°C, while pressure uses 4hPa and wind 2m/s. Precipitation and CAPE are assessed relative to the ensemble's own spread.
Each scenario has its own colour, with Scenario A always the largest group. Its average line is solid within the selected period and dashed outside it, because members that form a coherent scenario can diverge again before or afterwards. A scenario is marked diverged when its members become more than twice as spread out as they were during the selected period. The tooltip also gives the range for each scenario.
Scenarios are calculated separately for each chart, so Scenario A for temperature is not necessarily made up of the same members as Scenario A for pressure.
This makes the tool particularly useful for asking whether a potentially unsettled spell is one broadly consistent outcome, or whether the members are split between several quite different possibilities.
The median splits the members into two equal halves, and its position within the box shows the shape of the distribution. A median near the bottom of the box means the upper half of the members is spread more widely towards higher values. Near the top, the distribution stretches more towards lower values.
On Spaghetti and Scatter charts the ensemble mean is shown as a dashed black line. In Min/Max mode it is calculated from all members. In P10/90 mode the most extreme 10% at each end are excluded (a trimmed mean), reducing the influence of outliers. A significant gap between mean and median can indicate that a minority of members are forecasting something quite different from the main body.
Show Steps displays just one time of day throughout the forecast. Selecting 12Z, for example, shows only midday values, making day-to-day changes easier to compare.
Whisker Mode switches the outer range between the 10th–90th percentiles and the full minimum–maximum.
Quick Select lets you choose a city by region, while the latitude and longitude dropdowns allow any grid point to be selected. Cities outside the selected model's coverage are greyed out.
Charts can be downloaded as images from the chart menu, with the model, run, location and variable included, which makes them suitable for presentations or sharing. On phones and tablets, zooming is disabled so that scrolling past a chart doesn't accidentally zoom it.
The Ensemble Dashboard is available with free access, but much of the data and functionality requires Premium access. The aim is to make sophisticated ensemble analysis accessible without specialist software or expensive data platforms.
For forecasters, enthusiasts and professionals who need to understand uncertainty rather than follow a single forecast line, the combination of multiple models, multiple chart types and the new Cluster and Scenario tools is a powerful way to interrogate the data. If you would like to discuss commercial access, please contact us directly.
Comments are available to logged-in members.
No comments yet. Be the first to post.