Teams / DSLR Documentation and Reduction / Number of stars used in Final Reduction
*Sky & Telescope Illustration by Casey Reed
Privacy Policy | Contact Us | Affiliates | About Us | Participant Sites
International Year of Astronomy | American Association of Variable Star Observers (AAVSO)
Just thought I would post a follow up on this topic. Grigoris sent me a copy of the spreadsheet and it looks like everything is okay. Here is the input data:
And the transformation coefficent output table:
Checking these values against eta Aur, we get a check star magnitude of 3.162 mag (compared to 3.172 optimially) for a difference of 0.01 mag (not too bad). Looking at the TCgraph:
We see that the lam Aur point is fairly far from the line (0.15 in D cat). The great part about calibrating against all of these stars is that one erroneous value doesn't affect the results that much. For instance, if we were to remove lam Aur from the list of calibrators, the resulting check star mag is 3.194 only 0.022 mag higher. I haven't seen any paper on the statistical limits of a DSLRcamera, but I suspect the technical limit in the milli-magnitude level with 10-20 millimag uncertainty being the norm. I don't know how many images were used in determining the instrumental magnitudes, but I suspect with some averaging and a little statistical work could push the uncertainty down a little further.
Hello!! Does this difference in the value of eta aur equals our error in the estimation od eps aur value? What I mean is this, you say that: "Checking these values against eta Aur, we get a check star magnitude of 3.162 mag (compared to 3.172 optimially) for a difference of 0.01 mag (not too bad). " So our estimation on eps aur would be (something) +/- 0.01 ? Grigoris
I'll defer to Richard Berry's advice: the average and standard deviation of several images are best suited to estimate the error in the camera.
Grigoris, The idea of including the additional stars is for outlier rejection. If you are calibrating against 10 stars and you find that one of them is far off of the transformation coefficient curve, you can reasonably asume that some mistake was made (i.e. the wrong star in the FOV). Using multiple calibrators will also help push down the statistical uncertainity in both the TC and zero point offset, therefore making your data less uncertain. Is one star far off of the linear fit in the TC graph? Can you send me the spreadsheet so I can check it out? Brian
Hello!! Yes, i can understand that using more stars the statistics become better. But I was trying to use all the stars so this fails in some cases. From what I understand we can use as many stars as we are able so as to get 3.172 at the output cell for eta aur. Regards, Grigoris ps 1 Brian I send you the spreadsheeto to take a look. ps 2 perhaps i was a little bit hasty on this post...
Grigoris, You do bring up a good point. The tutorial does not discuss outlier rejection, which is a tricky topic especially if you want to do it in a statistically legitimate way. Perhaps I'll make a page discussing what can happen when inputting data... Brian