Case Study: Ulysses

Dust Experiment

This data set consists of 968 data points acquired over 2 years from the Ulysses mission. 33 parameters are included: Raw data: No., date, impact time, CLN, AR, SEC, IA, EA, CA, IT, ET, EIT, EIC, ICC, PA, PET, EVD, ICP, ECP, CCP, PCP, HV and evaluated data: R, LON, LAT, D_Jup, rotation angle (ROT), instrument pointing (S_LON, S_LAT), speed (V), speed error factor (VEF), mass (V) and mass error factor (MEF). It was obtained in ASCII format via the Ulysses/Galileo Dust Experiment Home Page ( and converted to XmdvTool format. Details on the data sets are available in a series of articles by Grun et. al. in Planetary Space Science, Vol. 43, No. 8 (1995).

Overview Images

We've eliminated the raw data fields except for the impact time, and then displayed the data both as a scatterplot matrix and using parallel coordinates. The Longitude/Latitude plot does not have a smooth flight path. This turns out to be caused by the flight plan, which diverted Ulysses to Jupiter during the time of this data gathering. The missing values (which are assigned high values for completeness) tend to through off the distribution of the plots. A better method for dealing with these missing values is needed. This data also highlights a limitation in XmdvTool, namely that data are imported as single-precision rather than double-precision floats. The masses of all dust particles mapped to the same value (0) because of this.

Another problem with plotting this data is when using the Date field, which is specified as the year (90, 91, 92) followed by the date count from the start of the year (1 - 365). We concatenated these 2 numbers, which left gaps between the end of one year (90365) and the start of the next (91001). An interesting problem would be to eliminate these gaps to avoid distortions in the plots. This also occurs with mm-dd-yy date representations. The scatterplots below show an example of these gaps.

Matthew O. Ward (