[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: another check on quality of Mark IV ensemble photometry
Hi Michael,
I stared at this for a while and decided that I did not know the
definition of "scatter"
# min max N unclipped clipped median
interquartile
# mean stdev mean stdev mean
stdev
#----------------------------------------------------------------------------
V-band
7.0 8.0 1398 0.032 0.016 0.030 0.012 0.029 0.029
0.005
8.0 9.0 4098 0.038 0.025 0.037 0.015 0.035 0.035
0.006
9.0 10.0 10887 0.043 0.026 0.041 0.016 0.038 0.039
0.006
Under unclipped, what is the mean of? The first entry shows 1390 stars.
Each of these has n measurements. How does one compute the "unclipped
mean" scatter? Is the scatter the sigma of the n measurements of each
star, and then the mean the mean for all the stars and the stdev the
sigma of the sigmas?
OK I will stop here as an answer will change all my other questions.
One more thing. Is the unclipped value before the application of the
ensemble photometry and the clipped after? There seems very little
difference.
Tom
On Tue, 22 Nov 2005 15:33:33 -0500, "Michael Richmond"
<richmond@stupendous.cis.rit.edu> said:
>
>
> (apologies if you have seen this before -- I had E-mail issues
> yesterday)
>
>
> I've mentioned before my plan to reduce the Mark IV photometry
> by breaking it up into "patches" of roughly 1 degree on a side,
> and then considering the data within each patch as an ensemble.
> Tech Note 101 describes the idea and gives a few initial findings.
>
> http://stupendous.rit.edu/tass/technotes/tn0101.html
>
> I have finished running a first pass of the ensemble photometry
> code on most of the Mark IV data. Below is a table which summarizes
> the properties of the resulting ensembles. The quantity of interest
> is the "scatter around mean ensemble magnitude"; that tells us how
> _precise_ the Mark IV photometry of a single star is from one night
> to the next to the next.
>
> Caveat: to appear in this analysis, I required that a single star
> be measured at least 5 times in both V and I.
>
> In the table below, I break all the data into bins by the magnitude
> of stars. Within each bin, I list
>
> N number of stars in this bin
>
> unclipped mean and stdev of the scatter within each
> bin
>
> clipped ditto, after 1 round of 3-sigma clipping
>
> median median of scatter within each bin
>
> iqm mean and stdev of all scatter values
> between the 25'th and 75'th quartile
>
>
>
>
> # min max N unclipped clipped median
> interquartile
> # mean stdev mean stdev mean
> stdev
> #----------------------------------------------------------------------------
> V-band
>
> 7.0 8.0 1398 0.032 0.016 0.030 0.012 0.029 0.029
> 0.005
> 8.0 9.0 4098 0.038 0.025 0.037 0.015 0.035 0.035
> 0.006
> 9.0 10.0 10887 0.043 0.026 0.041 0.016 0.038 0.039
> 0.006
> 10.0 11.0 26699 0.049 0.031 0.046 0.017 0.043 0.044
> 0.006
> 11.0 12.0 63549 0.067 0.035 0.063 0.019 0.061 0.061
> 0.007
> 12.0 13.0 72528 0.110 0.042 0.105 0.030 0.103 0.104
> 0.012
> 13.0 14.0 80510 0.161 0.056 0.155 0.045 0.153 0.154
> 0.018
> 14.0 15.0 2290 0.231 0.079 0.225 0.069 0.223 0.225
> 0.029
>
>
> # min max N unclipped clipped median
> interquartile
> # mean stdev mean stdev mean
> stdev
> #---------------------------------------------------------------------------
> I-band
> 7.0 8.0 19210 0.045 0.039 0.039 0.023 0.033 0.035
> 0.008
> 8.0 9.0 46670 0.035 0.030 0.030 0.017 0.026 0.027
> 0.006
> 9.0 10.0 115141 0.034 0.028 0.031 0.016 0.028 0.028
> 0.006
> 10.0 11.0 276427 0.037 0.026 0.035 0.014 0.032 0.033
> 0.005
> 11.0 12.0 628596 0.049 0.023 0.047 0.015 0.046 0.046
> 0.006
> 12.0 13.0 789607 0.077 0.031 0.073 0.024 0.072 0.072
> 0.009
> 13.0 14.0 111809 0.118 0.050 0.112 0.039 0.111 0.112
> 0.015
> 14.0 15.0 371 0.275 0.096 0.273 0.085 0.275 0.274
> 0.034
>
>
> You can find a figure which illustrates some of these columns
> at the URL
>
> http://spiff.rit.edu/richmond/temp/scatter.gif (GIF)
> http://spiff.rit.edu/richmond/temp/scatter.ps (postscript)
>
>
> The bottom line is that the typical scatter left over after
> doing the ensemble photometry is about 3 percent at best, and increases
> to around 12 percent at the faint end.
>
> This is right along the lines of earlier work I and others have
> done, so no big surprise. This just confirms that the results
> remain when you consider all the data.
>
> Very slowly, I am making my way through this sort of analysis.
>
>
> Michael Richmond
>
--
Thomas F. Droege
droege@fastmail.fm