Return to Home Page
Re: Error reporting > I have a simple (& therefore no doubt dumb) problem about the recommended > way of determining errors, as set out in the last CCD Views #314. It says: > - - - > *Simple* error ranges can be computed with this easy formula: > > > error(mag) = 1.0857 / sqrt(net_counts*gain) > > > ...where net_counts is the count of electrons times the CCD gain. > This provides a first order approximation. This is usually available > in your photometry software. So basically divide 1.0857 by the square > root of the counts of the star as reported by your photometry > software. Tada! This formula should be better understood as the "highest achievable precision" of an ideal instrument, without no sky background, no instrumental noise and all other degrading factors. If this formula gives an error better than 0.01 mag, you had better carefully use it, since it is well-known other external errors can easily introduce a comparable source of errors even in good conditions. If this formula gives several tenth of a magnitude, it would be often a reasonable estimate of your instrument's limitation. However, under such difficult conditions, many commercial photometry packages give incorrect (strongly biased) results. The limit can be only approximated when you use a suitable and reliable photometry package, and adequately use it. In a number of cases, even professional astronomers take these results at their face values, and write them in their papers as the "error estimates". This sometimes resulted in "false positive" detections of variations greater than the inadequately estimated errors. It is often the case that such authors tend to misuse (at least badly use) the photometry package, and don't even question whether they have correctly measured the true variation... X-) The priority thing is to measure correctly, rather than to estimate the (expected) error correctly. The error estimate should naturally come when the correctly measured data sets become available. Regards, Taichi Kato
Return to the Powerful Daisaku Nogami
vsnet-adm@kusastro.kyoto-u.ac.jp
