To determine the annealing parameters for the disambiguation module, we need to characterize the sizes of the HARPs. In what follows, the area is the area in pixels of the rectangle enclosing the active pixels, since this is the relevant quantity for the speed of the annealing, determined from the product of the keywords crsize1, crsize2.

Start with the definitive HARPs, from 2010.05.01_00:00:00_TAI to 2012.08.19_00:00:00_TAI (noting that the last few weeks may be incomplete because of the use of the full disk crossing for determining a HARP size) in the series hmi.Mharp_720s. The dashed horizontal line marks the median value; the dotted horizontal line marks the 95th percentile. Time is in days since 2010.01.01_00:00:00_TAI, and only the HARPs present at 00:00:00_TAI are included. (I did not properly account for leap year, so days since 2012.02.29 are offset by one, and the number/area on 2012.03.01 are incorrect.)

If we want to be able to keep up approximately half the time, then we need to be able to disambiguate approximately 10⁶ pixels at each time; if we want to keep up approximately 95% of the time, we need to be able to disambiguate more than 3x10⁶ pixels at each time.

This needs to be revisited for NRT HARPs, since the areas will presumably be smaller than for definitive HARPs.