Tuesday, April 4, 2017

A conclusive non-conclusion about dimming in the AAVSO data

I'm spending too much time on this, so will have to bring it to a close until the summer's observing is done.

I took one more look at the AAVSO data, this time doing something called binning, similar to what Brad Schaefer did with the DASCH data in his paper on dimming in the historic photographic plates. Binning takes several observations within a defined time period and averages them before attempting to fit a model to them. In this case, the model is a simple straight line. This has the effect of giving each time period an equal "vote" in the best fit to the model, even if there is much less data in one time period than another. In the case of the AAVSO data, some observers would report many observations over a short period of time, which tended to overweight their observations. Binning mitigates that.

Of course, you have to decide what period of time you will use for binning the roughly 500 day span we have so far. I arbitrarily picked 10 days, and averaged the observations for each observer over that time period. There were 47 AAVSO observers in all whose measurements survived the filtering process in the "V" passband.  There were 48 observers, but I identified one who temporarily had apparent problems with respect to the others, so was filtered out to make it simple.
The V Band Fit with 10 day binning