search

GBTAmmoniaSurvey / keystone

KEYSTONE codebase
GNU General Public License v3.0
0 stars 3 forks source link

Negative baselines #17

Open low-sky opened 6 years ago

low-sky commented 6 years ago

When averaging over large areas, negative baselines show up.

low-sky commented 6 years ago

See here

keflavich commented 6 years ago

FWIW, this is a pretty systematic problem across all mapping surveys; it's not unique to GBT GAS. Without looking closely at the issue again, this appears to be a general problem when the emission fills a substantial fraction of the bandpass, especially near a point where baseline curvature is a problem. It's likely a mix of inadequate masking (since you can only mask down to a few sigma, you're including some of the line wings in the fit) and the resulting inadequate fitting area.

This seems like a fun problem to tackle in the general case in the limit of infinite time.

low-sky commented 6 years ago

After screwing around a bit with all our strategies (and getting a multiprocesing framework working), I think that our best approach may be to fit a simple (thin) ammonia model + baseline to the individual cubes. This would take care of the issues of baseline window selection especially when there's no space between the hyperfines. It would not deal with multi-components perfectly but it seems like a good "first pass" attempt.

low-sky commented 6 years ago

Proof of concept code here More efficiency needed, especially with this.