Michael Turmon, JPL, February 2012

The results here are two movies of NRT HARPs from 2011 October and November.


I ran a few days of images in this mode (2011.10.01/15d and 2011.11.01/10d). Each frame, out of 120 per day, required a run of Matlab and the HARP-ingestor. Processing (not including mask-making) took about 90 sec per frame.

These runs fix a problem with earlier code, in which HARP numbers could change in an undesirable way. Label changes as merges occur are inevitable, but now the label of the merged HARP is taken to be that of the longest-lived of the HARPs that were merged. This gives more predictable results.

I also noticed an issue related to cosmic rays. For definitive runs, I introduced a threshold that we have to see 3 appearances of a HARP for it to be ingested. Short tracks of 1 or 2 HARPs are produced by the tracker, but these HARPs are not ingested unless they are three long, so they will not appear in JSOC. In NRT mode, I am not doing this filtering (easy to change if desired) because it would delay seeing a new HARP for two frames.

I introduced the threshold because CRs were causing occasional length-1 HARPs at the limb. The CR screening is not done (at all/as well) about 20 or 50 pixels short of the limb; we were detecting CR-induced HARPs in this little ring. About 5% of HARPs are screened in this way.


The two resulting movies are: NRT HARPs for October 2011 and NRT HARPs for November 2011.

This movie can be compared to the definitive movies: definitive-HARP October movie and definitive-HARP November movie from the large set of draft HARP runs.

One difference you might notice, that is not unexpected, is that the definitive-HARP movie gets a "running start" from late September -- that is, it had some history of prior HARPs that makes the gray outlines around HARPs, and some HARP groupings, especially in the first couple of hours, look different. In production, they would both have the same history, so this would not happen.

Near the end (2011.10.10_23:00) you can see a 3 or 4 way merge that is several different HARPs in the NRT series. The NRT and definitive diverge significantly there.