gbtgridder for non-GBT data

[teuben]

During the single dish school I patched the code to work with 20m data, and added a --diam parameter to play with the convolution independently of the presumed setting.
However, we did run into some issues which may require more testing. From the presentation slides of group2 I'll just mention them:

1. Line cube still has residual baseline (unavoidable?)
2. Spatial edge effects from gridder (unavoidable, or set a better mask?)
3. Small scale structure < beam ? (maybe need a test on a point source?)
4. First channel wrong? (this was kind of obvious when we used the --channel flat)

[astrofle]

Thanks for sharing that Peter. The latest version of the gbtgridder 2.0, also has a --diameter flag that defined the telescope diameter. About the points you shared

1. The gbtgridder used a very crude baseline subtraction method. It subtracted the mean along the spectral axis. In the latest version we have opted for not producing a "line" cube.
2. In the version 2.0 of the gbtgridder the padding around the mapped area has been reduced. This results in "less" edge effects in the cubes.
3. I had observed something similar, but was attributing it to using the wrong beam size. For 2.0 we did try to test on a point source, and were able to recover the expected source size in the gridded data.
4. Could you provide more details about this? What do you mean by wrong (for example is it the frequency, or the data, or something else)?