This page's content is no longer actively maintained, but the material has been kept on-line for historical purposes.
The page may contain broken links or outdated information, and parts may not function in current web browsers.

Atlas of Extratropical Storm Tracks (1961-1998)

Extratropical storms directly affect the environment and economy throughout the mid-latitudes, greatly impacting the lives of billions of people on a daily basis. This on-line atlas provides fundamental information about extratropical storm systems, including maps of storm frequency and intensity as well as plots of individual storms paths. Additional plots of the "most severe" storms are also available and are defined as the storms of maximum intensity within the selected time span. Maps are available as monthly and seasonal means for the years 1961 through 1998 in either global or hemispheric projections.

The entire data set is available for download as well, along with a FORTRAN program for extracting time-specific and geographic-specific information. Together, the images and data presented here describe the state of the mid-latitude storm tracks during the last half of the 20th century.

Note: If you select a Winter (DJF) graphic, the season overlaps the boundary of two years. The appropriate image can be obtained by using the year in which the season ends; e.g., the images for Winter (DJF) of 1970-71 may be obtained by requesting Winter 1971.

There are many methods used to define the position of the extratropical storm tracks, probably the most common being to plot the variance of the 500-mb geopotential height field. That technique, however, does not yield information on the transient position of individual storms. Here we have instead track the position of each low pressure center within a series of gridded sea level pressure (SLP) fields. The technique was originally developed to examine the storm tracks produced by atmospheric general circulation models (GCMs), but it is directly applicable to other gridded SLP datasets, such as those derived in weather forecasts or reanalysis projects. The SLP fields used in this atlas are derived from the 12-hourly (0Z and 12Z) 500 HPa and 1000 HPa geopotential heights generated by the NCEP Reanalysis project.

It is important to note that neither the SLP data nor our storm products represent raw meteorological observations. They are a data assimilation product, based on both observations and computer modeling techniques. The advantage is that we are able to produce storm data sets that are gridded and global in extent. However, users should be aware that the input data, and thus the results, are not of equal quality at all locations (e.g., over remote ocean regions) due to the nonuniform distribution of observations used in the assimilation process.

Users of this atlas should remember that this data product is based on the results of a data assimilation model forecast. It is, therefore, subject to all the errors that exist in the input data. The NCEP/NCAR Reanalysis Project provides regular updates regarding errors and problems in the data products and model. These updates can be found at the NCEP/NCAR Reanalysis Problems List.


Storm track data from the 1961-1998 Atlas of Extratropical Storm Tracks is available for download along with an associated FORTRAN program that can be used to extract subsets of the database.

  • stormtracks.bin -- a 4.8 MB binary data file containing the database
  • getstms.f -- a FORTRAN program which retrieves storms from the data based on a user-defined date range.
  • st_report.txt -- a sample output text file which includes a table of information about the location and central pressure of all the cyclones in the atlas during the specified date range.
  • README.txt -- further information about the output report's contents and usage.

Computation and Plotting of Storm Tracks

In order to create the "storm" paths and statistics presented in this database we use an algorithm that identifies and tracks sea level pressure minima. The raw data have been taken from the NCEP Reanalysis Annual CD-ROMs for the years 1961-1999. Twice-daily sea level pressure fields, gridded to 2°×2.5° (lat×lon), are calculated using the method recommended by the NCEP Reanalysis project, which employs the 1000 HPa (z1000) and 500 HPa (z500) geopotential height fields in the following formula:

SLP = 1000 × e(z1000 / (1.5422885 ( z500 - z1000)))

If your browser supports MathML, this may be written as:

SLP = 1000 e z1000 1.5422885 (z500 - z1000)

As there are numerous techniques for determining sea level pressures from atmospheric observations, all having limitations, we also compared the SLP fields generated in the above way for general consistency with those generated using an independent method. That method, which we use in the NASA/GISS GCM, extrapolates the surface pressures to sea level based on elevation and the atmospheric lapse rates calculated from surface air temperatures. This second technique is preferable for use with fields produced by grid point models, like the GISS GCM, but is not wholly suitable for use with the spectrally calculated fields of the NCEP reanalysis, in which negative elevations exist over many regions. However, similarities between the two SLP datasets gives us confidence that we have obtained the best-possible set of storm statistics given the variables available from the NCEP Reanalysis fields.

With the 12-hourly SLP fields in place, our algorithm searches for and identifies all local minima from one gridded field on any user-designated date. With the latitude and longitude positions of those minima stored, the minima for the ensuing 12-hour SLP grid is searched, and the positions of minima are again located. Next, the results of the two searches are compared and any SLP minima within a critical radius of 1440 km are joined by a segment, representing the path of the low pressure center during that 12-hour period. The 1440 km critical radius (the maximum we found to be reliable) represents the maximum distance a low pressure center may be displaced during any 12-hour period and still be considered the same genetic system. Put another way, it implies that the cyclone's center can travel at a mean speed of no more than 120 km/hr (1440 km / 12 hr).

As the SLP minima from subsequent 12-hourly grids are analyzed, other criteria are also invoked: Any two associated minima identify one storm track segment, but we have found it reasonable to limit plotted "storms" to those which last at least 36 hours. That is, all storm tracks compiled in this database consist of at least three 12-hour segments, which are defined by a SLP minimum appearing in four contiguous 12-hour grids. In addition, if at any time two segments on the same track were found to define an acute angle of less than 85° the low pressure centers are considered to represent separate storms. This criterion was invoked because, in general, extratropical cyclones are not found to "double back" on themselves over the course of only one 12 hour period. By setting this limit we avoid the possibility that the algorithm will misidentify secondary lows (which can form in the wake of extratropical cyclones) as a reversal of the primary low pressure center.


Please address any questions about this research to Dr. Mark Chandler.

Also participating in this research was Jeff Jonas.