[Date Prev][Date Next][Thread Prev][Thread Next][Search]
Overscreening of GIS data caused by mkf bug
Dear ASCA PIs and Archival Researchers:
[The WWW version of this bulletin can be found at
When mkfilter, the program which creates mkf files, was upgraded to
mkfilter2, a bug was introduced into the mkf file. The bug arose because
mkfilter2, unlike its predecessor, uses housekeeping files as input.
As a result of the bug, the G2_L1 and G3_L1 columns in the mkf file
remain defined for most of the file. Since these parameters are used in
the routine cleaning for the GIS instrument, this bug will result in the
rejection of a substantial fraction of your GIS data.
To see if this is the case for your data, start up xselect, set the
datadirectory to wherever your data resides, and issue the commands:
xsel:ASCA > mkfbin "G2_L1 G3_L1"
> Give print-out time interval >
xsel:ASCA > plot mkf
The available parameters are:
0 - TIME
1 - G2_L1
2 - G3_L1
> Enter independent variable ( 0 for TIME ) > 0
> Enter dependent variables (e.g. 1-6) > 1-2
Look at the plot, and if the values are -99.9 for most of the plot, your
mkf file has been corrupted by this bug.
The bug was fixed in August 1995. For affected data processed before
this date, the solution is to remove the G2_L1 or G3_L1 criterion from
the mkf selection. To do this, run ascascreen with:
Ascascreen will not start up xselect, but just write the script files
for the run, and quit. Now the mkf criteria are stored in the file
prefix_mkf.sel, where prefix is the session prefix you chose. Edit this
file, and remove the sub-expression that refers to G2_L1 or G3_L1. This
is currently the last line in the prefix_mkf.sel file.
Next you should rerun xselect, using the modified criteria, by saying
Finally you will have to clean up by hand the data which should have
been cleaned by the G2_L1, G3_L1 criterion. Fortunately that is fairly
easy. This criterion eliminates times during which the instrument was
not taking counts, though it was nominally on. These intervals will show
up on the light curve plot as regions when there is supposedly data
(i.e., there is a red line on the plot), but the counts are consistently
zero. You can remove these regions by doing
> extract curve
> filter time cursor
> extract event
Then you should be ready to continue the analysis.
Keith Arnaud & Charles Day, ASCA GOF
Please send comments and questions to firstname.lastname@example.org