HMI Duplicate Region Reduction

Michael Turmon, JPL, October 2011

This page gives some notes on the effect of different time-delays before deciding a region has vanished.

This page is in development (November 11, 2011).

HARP Finalization

A HARP is declared to be final when a certain amount of time has passed without seeing activity in the region where the HARP should be. In our past work, this finalization threshold is 3 hours. We observed some duplicate HARPs in the synoptic maps, and believe raising this threshold would address the duplicate HARPs. But, it could also cause inappropriate joining of AR remnants with newly emerging ARs.

We performed a test, where we raise the threshold from 3 hours to 6, 12, 24, and 48 hours. We used three time periods: ones from late August 2011 and mid-September 2011 that we have examined before, and another one from late April 2011. We used the newer magnetic field threshold to derive masks, from which tracks were found.

The outcome of these tests was that the threshold could be raised a lot before inadvertent linking would happen. So now we are doing experiments with times of 24, 48, and 72 hours.

Also, we maintain a map of locations of past occurrences of a given region; the past in this map decays exponentially with time. We are experimenting with two rates, one implying a half-life of 1/10 day, and another with a half-life of 1/20 day.

Files Here

We have three sets of experiments: April 19/14d, August 28/12d, and September 12/12d.

In each, we used three settings of the region-finalization threshold (24, 48, 72 hours), and two settings of the map decay rate (1/10, "slow", and 1/20, "fast"). We used the new masks.

This shows movies (left link) and synoptic maps (right link) for these experiments. The synoptic maps take about 3-4 seconds to load in my browser (Chrome 15). The synoptic maps might be best loaded into new browser windows/tabs. The maps support zooming. NOAA tracks are in gray, and HARPs are in a variety of colors. Bounding boxes in the synoptic maps encircle only the HARPs, not the NOAA ARs (otherwise it gets too busy). You can use the tooltips as you brush on points to associate the synoptic map points with the movies.

Tag Map Time Delay Map Rate
Apr-24h-fast map 2011.04.19/14d 24h 1/20 (fast)
Apr-48h-fast map 2011.04.19/14d 48h 1/20
Apr-72h-fast map 2011.04.19/14d 72h 1/20
Apr-24h-slow map 2011.04.19/14d 24h 1/10 (slow)
Apr-48h-slow map 2011.04.19/14d 48h 1/10
Apr-72h-slow map 2011.04.19/14d 72h 1/10
Aug-24h-fast map 2011.08.28/12d 24h 1/20
Aug-48h-fast map 2011.08.28/12d 48h 1/20
Aug-72h-fast map 2011.08.28/12d 72h 1/20
Aug-24h-slow map 2011.08.28/12d 24h 1/10
Aug-48h-slow map 2011.08.28/12d 48h 1/10
Aug-72h-slow map 2011.08.28/12d 72h 1/10
Sep-24h-fast map 2011.09.12/12d 24h 1/20
Sep-48h-fast map 2011.09.12/12d 48h 1/20
Sep-72h-fast map 2011.09.12/12d 72h 1/20
Sep-24h-slow map 2011.09.12/12d 24h 1/10
Sep-48h-slow map 2011.09.12/12d 48h 1/10
Sep-72h-slow map 2011.09.12/12d 72h 1/10

If you're looking in a synoptic map directory, you can map the directory name to a delay time and map rate using the following table. The directory name will have the form dupe-MON-NN where MON is a month (apr, aug, or sep), and NN is a number (12 through 18). I changed the movie names to include the hour delay, but the directory names are encoded in this obscure way. This is regretted.

Map Rate Day Delay Directory Name Contains
1/20 72 12
1/20 24 13
1/20 48 14
1/10 48 16
1/10 24 17
1/10 72 18