Opened 16 years ago
Closed 15 years ago
#44 closed task (worksforme)
Invert - fix theoretical noise calculation for CABB data
| Reported by: | MarkWieringa | Owned by: | MarkWieringa |
|---|---|---|---|
| Priority: | major | Milestone: | 7. Stage 1 - available for testing |
| Component: | MIRIAD - CABB branch | Version: | |
| Keywords: | Cc: | ||
| Estimated Number of Hours: | 5 | Add Hours to Ticket: | 0 |
| Billable?: | yes | Total Hours: | 3 |
Description
Invert calculates the theoretical noise based on the integration time, Tsys and uv weighting scheme. It knows about frequency overlaps between channels for the AT correlator.
With CABB, we need to update the algorithm to take into account the way Tsys is calculated (spectrum of values) and the channel overlaps in the 2GHz band.
Change History (2)
comment:1 Changed 15 years ago by
comment:2 Changed 15 years ago by
| Resolution: | → worksforme |
|---|---|
| Status: | new → closed |
| Total Hours: | 0 → 3 |
2009-09-03
Looking at invert more closely, it does not take overlap into account. Single visibility: sigma=JyperK * sqrt(Tsys1*Tsys2/(2*BW*inttime)) The theoretical noise is sqrt(sum(wt*wt*sigma*sigma)/sum(wt*wt)), where wt is the visibility weight including tapering etc.
Using the band average Tsys is probably a reasonable approximation for mfs images. Channels at top and bottom of band may have significantly higher Tsys though.
At present we are still using an average Tsys across the band and it seems to be working reasonably well.
The calculation code within INVERT is still not completely understood - doesn't have any terms allowing for channels that overlap in frequency?