|
Linuxpops
3.01.04
Spectral Anaysis Package for Single Dish Data
|
|
Linuxpops
3.01.04
Spectral Anaysis Package for Single Dish Data
|
FB-Offsets lets the user correct anomalies in the band-pass caused by DC offsets in the baseline due to hardware issues. These issues are most common in analog filter-banks. Sometimes referred to as 'platforming'. These procedures should be used judiciously.
First lets look at some scans that exhibit this problem.
The resulting plot looks like this:
As can be seen, there is a slight offset in the first half of the plot. Let's fix that.
First set the environment variable, 'markers' to a value that represents the dividing line between the two halves. In this case, there are 1024 channels total, so, set markers to 512.
It's also necessary to have baselines defined, as can be seen in the example above. It's best not to include the faulty sections in the baseline regions if you plan to use the automatic mode.
Now, there are two ways to execute this procedure, manually or automatically.
NOTE: If there is a strong line in any of the segments requiring offsetting, then the only proper way to perform these functions is manually as the automatic process will end up averaging in the strong line and that could effect the results.
To perform the fix automatically, type:
The results are show here:
Alternately, you can try manual. This requires you to click the mouse on the bad sections at approximately the average level of that section. Then click the mouse a second time on the average level of a good section. The program then offset the bad section to bring it into alignment.
Start be entering the segment that's needs adjusting, counting left to right starting at 0. Markers should appear to allow you to identify the boundaries. When clicking the plot, only the Y-Axis value is important, so you can move freely, left or right, to aid in you selection. i.e., you don't have to click in the actual area of each section, only the vertical axis matters.
Just use your best guess as to what the 'average' value for each section is.
Again the results are fairly the same, although the automatic method can be more reliable.
Alternatively, you can have Linuxpops adjust the offset using a slightly more accurate method. First you must define at least two baseline regions, one encompassing the bad region, but not including and line features, and a second region containing the 'good' sections you wish to match.
Once again, accumulate the scans as above and define two baseline regions outside the central emission line.
NOTE: Only define one baseline region per markers segment.
Now, perform an average on each baseline segment:
Now tell Linuxpops to adjust bandpass segment 0 to match segment 1 using the baseline averages.
We now recommend removing a baseline to eliminate the remaining dc-offset in the data.
1.8.5 
