The UKFSST package

UKFSST is an add-on package for the statistical environment R to efficiently estimate movement parameters and predict the most probable track from raw light-based geolocations and sea surface temperatures (SST). This package offers an approach to use both types of information in one coherent state-space model and carries out the estimation via the unscented Kalman filter (UKF). Read all about the model here (html / pdf).

How to try the package

The UKFSST package depends on two other packages (date, and kftrack).

To make installation a bit easier, a script has been prepared to set up all of these five packages.

To run it fire up R and type (or paste) the following line:

  source('http://www.nielsensweb.org/ukfsst/install')

This should be done in root/administrator-mode on Linux/Windows to allow each package to add to the R-documentation and library tree.

How to use

UKFSST comes with a function explaining how to fit a track. It covers all steps from how to read in the raw track, automatically obtain SST-field, fit the track, and plot and investigate the final track. This help is displayed by typing:

  road.map()

Also the package comes with a complete example. To see it type:

  example(blue.shark)

Thank you for your efforts in downloading and trying this package. If you encounter any bugs or have any questions please don't hesitate to contact me.

Latest update: 14 Sep 2009 (email: anders 'dot' nielsen 'at' hawaii 'dot' edu)