Check to make sure original fits were/were not forced through 0,0
It is somewhat clear that sampling isn't causi...
It is somewhat clear that sampling isn't causing huge issues in the sense I was imagining... the highest pixel ranges from 0.98 to 1.04 depending on how the PSF is made anyway, so 5% errors are totally unavoidable.
It looks like, for AFGL 4029 combined images, ds1 ...
It looks like, for AFGL 4029 combined images, ds1 and ds5 agree to within 5% in the Dec2010 data. This implies that the offset is a result of the individual scans instead of coadds, though for the individual observations ds1>ds5 pretty uniformly.
From a comparison of 3 different epochs using abou...
From a comparison of 3 different epochs using about a dozen different reduction techniques, but all with pairs of cross-scans, ds1 is uniformly higher than ds5. The only thing left to do is compare the individual scans, I suppose... if the cross-scans behave differently, that explains the difference between 2010 and other epochs
Next step: prove that this works in maps...
Next step: prove that this works in maps...
Bleh, a whole day's work dedicate to new delin...
Bleh, a whole day's work dedicate to new delining methods and testing. Very unrewarding. However, there are results that I need to FINALLY prove and show tomorrow: 1. amplitude suppression + phase preservation is a good alternative to zeroing/scrambling or replacing with noise 2. line fitting is effective but slow 3. Assuming a line profile and suppressing is nearly as effective as line fitting 4. scan-by-scan is less effective than observation-by-observation for fitting (but it's not clear if that holds for other methods)
What about the effects on the map? Unfortunately,...
What about the effects on the map? Unfortunately, in the tests I've run, delining actually INCREASES the noise in the map! How is this possible? My only hypothesis is that PCA cleaning does an excellent job of removing the line noise, but when it is replaced with truly uncorrelated gaussian noise, the PCA cleaning no longer removes that component. But we don't want PCA cleaning to grab instrumental correlated noise; we want it to be limited to atmospheric noise. So ideally we want to completely remove the line noise without adding noise to the image... maybe the line regions should just be set to zero after all....
Downsampling by a factor of 3 appears not to be an...
Downsampling by a factor of 3 appears not to be an option because this error: % NCDF_VARGET: Requested read is larger than data in dimension 0. Reducing COUNT to 615. % REBIN: Result dimensions must be integer factor of original dimensions % Execution halted at: NCDF_VARGET_SCALE 133 /Users/adam/work/bgps_pipeline/support/ncdf_varget_scale.pro occurs.
l020 too
l020 too
redo targets:<br>l068<br>l086<br>l111<br>l027
redo targets: l068 l086 l111 l027
Page 1 / 2 »