Notes on processing drifter tracks for the BWZ project


Mati Kahru,


·        Drifter tracks are received as emails every week from MAYRA C. PAZOS []. The emails are saved from MS Outlook with “Save as”. This produces a text file with the email header. The email header does not interfere with the processing but is not necessary.

·        A utility DrifterRead was created in C# and reads drifter track files in various formats (text) and saves processed files in text files.

·        The two main tasks of DrifterRead are

o       Interpolation and resampling of the variable interval data to equal interval time series

o       Splicing tracks from multiple files with partially overlapping tracks

o       Before doing any interpolation DrifterRead re-arranges the text file into correct lines as in the past many files were corrupted.

·        Resampling is currently done at exactly 1.0 days (that could be changed to any interval), i.e. to each day at 0:00 GMT. A cubic spine is fitted separately through the latitude and longitude time series and resampled at 0:00 GMT each day. Currently no quality checking is done, i.e. the apparent distance and/or the apparent trajectory are not checked. This could be added in the future.

·        Splicing of partially overlapping tracks from multiple is done by appending any non-overlapping data from a new file to the existing accumulated track.

·        The output format is currently “Longitude, Latitude, Time, zDay” but is easy to change. The “Time” is in decimal years (double precision), e.g. 2000.0 is January 1, 2000 at 0:00 GMT. The “zDay” column is actually redundant and is provided just for convenience. It shows the current decimal day of the year. Please note that January 1 is zDay 0 and not 1. This is different from commonly used Julian day or year day. For example, January 1 at 12:00 GMT is zDay 0.5 and not 1.5.

·        The following figures show examples of resampled Longitude and Latitude (drifter 43569 “Miguelito”, data of May 17, 2005).



·        The following figure shows an example of drifter tracks on top of a composite January-February 2004 chlorophyll image



·        Currently the following potential problems may occur:

o       Apparent straight lines – may be caused by missing or missed (!) data

o       Apparent tracks “through” islands or other dry land – may be caused by the spline interpolation whereas the drifter actually traveled around the island or cape. Another option is just errors in the latitude and/or longitude.