Guide to Using Cyclone Database Products


Getting Started


1. North Atlantic Cyclone Database products can be found at the 'Products' link above.

2. The menu bar on the products page has two drop-down menus, for selecting 'Run' time (left), covering the last 15 days, and 'Product' type (right). There are also links to the ECMWF home page (icon top left) and to this help page (top right). Two arrow buttons in the centre provide a quick way of changing run time, the left arrow steps down the run time list (ie to earlier times), the right arrow steps up, to later times.

3. To quickly view past evolution use the arrow buttons to step through time. Provided ‘Clickable control T+0 image’ is the selected product, this action will show successive T+0 control forecast analyses.

4. There is also an archive of older North Pacific products, provided for the TPARC experiment, available via this TPARC web page: there is one link for Winter TPARC (2009), and one for Summer TPARC (2008). Some more recently developed products are missing from this archive.



These web-based products aim to represent, objectively, the location and behaviour of near-surface synoptic scale features, in the ensemble and deterministic forecasts, in a variety of ways. The features represented are those typically associated with adverse weather:

-fronts, warm and cold, shown as lines

-cyclonic features, shown as spots, consisting of:

-Barotropic Lows
-Frontal Waves
-Diminutive Waves

A smaller spot (for Frontal waves and Diminutive waves only) denotes when the related front is associated with a less pronounced thermal gradient.

All cyclonic features are kept 300km or more apart, using co-location masking, based on a feature type hierarchy, and a minimum separation threshold.

Mean sea level pressure (mslp), as estimated from 1000mb geopotential height and temperature, is also shown on many plots.

1000-500mb thickness is shown on some plots using conventional colours.

On some plots precipitation totals and dominant precipitation type are also shown (see item 8 below for details).

The product suite relies on each cyclonic feature being represented by a row in a database. Each column then represents an attribute, such as 'latitude of the feature' (intrinsic) or 'mean sea level pressure at feature point' (extracted from model fields). Products rely on displaying these attributes in different ways.

A new tracking algorithm has been used to follow the cyclonic features as they evolve in each ensemble member. This is based on the following:



The time step between frames used in constructing the cyclone database, and on the products themselves, is 12 hours. This was set so as to ensure timely product delivery. Data fields from both determistic and ensemble runs are reprojected onto a 50km grid at the outset, because the focus here is on synoptic scale systems rather than fine mesoscale detail. This is still sufficient however for minor frontal waves to be identified and tracked.


Products and How to Use Them

Sub-section titles below are in red; these replicate items on the 'Product' dropdown menu. Miniature product examples are shown alongside (click to enlarge).

   1. ‘Clickable control T+0 image’
Explanation (follow numbered boxes in order)

   2. ‘Fronts animation’

'Dalmatian charts' (=feature animations)

In constructing each product in section 3 the EPS members are plotted first, followed by the control then the deterministic runs. Highlighters for the control (yellow circles) and deterministic runs (yellow crosses) are added last of all. The aim with each animation is to compress a huge amount of ensemble information into key elements that relate to adverse weather, and that can be visualised on a single plot.
   3.1 ‘Features animation’

   3.2 ‘Feature anim (mslp)’

   3.3 ‘Feature anim (300km 1km wind max)’

   3.4 ‘Feature anim (600km 1km wind max)’

   3.5 ‘Feature anim (thickness)’

  4 ‘Postage stamps Animation’

   5 ‘Postage stamps (detailed)’

  6 ‘Postage stamps (detailed, times)’

'Strike probabilities'

These products are somewhat analagous to the strike probability or 'cone of uncertainty' charts often used to display forecast tracks of tropical cyclones. However there are three main differences. Firstly, the strike probability charts are not cyclone-specific. All cyclonic features, for whom a particular attribute meets a pre-defined threshold, are represented on a single frame, wherever they are in the domain. Secondly, a feature need not be in existence at time zero. Provided the feature exists at the nominal frame time (and meets the threshold criteria) it will be represented. And thirdly, each frame represents only a time window of +/-12h relative to nominal frame time. We represent the full track, within this time window, of features that meet the threshold criteria at least once within that time window. Strike probability map construction is explained at the link below (animated):

It is important to not confuse strike probability with the likely location of wind maxima. What isrepesented is not the tracks of the wind maxima themselves, but the tracks of those features whose wind attributes exceed a certain threshold. In practical terms this means that if we have a localised high strike probability for a windstorm that is moving east, then it is somewhere to the south of that strike probability maximum that will most likely see the strongest surface winds, as the strongest winds commonly occur on the southern flank of such cyclones. Of course the picture becomes distorted if the local topography is complex, or if the cyclonic feature has an atypical structure.

   7.1 ‘Strike probs (1km wind > 60kn)’ and other thresholds

   7.2 ‘Strike prob Postage stamps’

   8 ‘Animation of control’, ‘Animation of member 1’, ..etc..

   9 ‘ T+0 'animation' ’, ' T+12 'animation' ’, ..etc..


Identifying and tracking extratropical cyclones is a complex process, particularly when one is also accouting for less easily discernable features such as frontal waves. The accuracy of the identifying and tracking algorithms used here compares very favourably with other published methods. Nevertheless there will be occasions when the behaviour of a cyclonic feature in one or more members, as represented in particular on feature track and plume diagrams in item 1 above, does not tally with a subjective interpretation of the input data (as represented in item 8). Sometimes there are errors, due to difficulties inherent in tracking. But equally what appears to be an error can sometimes be due to the eye perceiving an evolution that is not happening in reality. For example if the track of a feature is curtailed, it may be because of downstream development, where a low fills in situ, and another one develops downstream, such that for some short intermediate time span there are actually two centres in the model fields.

Erratic track behaviour can also arise in some situations where there is ambiguity regarding which of two subsequent features one initial feature evolves into. In such situations there may appear to be very large (even bimodal) spread in the feature tracking which will not just reflect a large EPS spread, but will rather reflect the tracking ambiguity.

When using the track and plume diagrams the user should generally follow the majority, and reject clear outliers. At the same time however they should be alert to the fact that in dynamically unstable situations what appears to be an extreme outlier, in terms of depth, track etc, may turn out to be correct. And the tracks should, as far as possible, be viewed alongside the synoptic evolution of a representative member (item 8) to ensure the correct interpretation is being made.

Note also that the strike probability maps (item 7) tend to give a visual impression that inherently foccuses on the majority view, so these should also be utilised at short leads, alongside the other products.

Further Reading

Hewson, T.D., 1997: Objective identification of frontal wave cyclones. Meteorol. Appl., 4, 311-315.

Hewson, T.D., 1998: Objective fronts. Meteorol. Appl., 5, 37-65.

Hewson, T.D., 2009: Diminutive frontal waves - A link between fronts and cyclones. J. Atmos. Sci., 66, 116-132.

Hewson, T.D., 2009: Tracking fronts and extra-tropical cyclones. ECMWF Newsletter, No. 121, 9-19.

Hewson, T.D. & H.A. Titley, 2010: Objective identification, typing and tracking of the complete life-cycles of cyclonic features at high spatial resolution. Meteorol. Appl., 17, 355-381.