Return to Home Page
I agree with most of what Brian says in his excellent, lengthy, but reasoned email. Some comments follow. Arne Brian Skiff wrote: > The silly idea of the AAVSO separating CCD V-band observations from > single-channel photoelectric V (and coding them as though they are different) > has led to the widespread notion that somehow CCD is "better" (or the reverse!) > and that they need to be kept separate (and the old-time photoelectric guys > won't talk to the new-fangled CCDers). The photometric scales are defined > independent of the detectors, so if data from two devices don't match, > then at least one of them is wrong. This is completely correct: "V" is "V" and doesn't depend on what detector you are using. In defense of the AAVSO, I think the main reason for separating CCDV from PEPV is for observer awards and has nothing to do with the science. Perhaps there is a better way to do this. > A comparable charge can be lodged against the lack of vetting in > the vsnet data. It ultimately means that if you want to use datasets, > you have to do the culling of bogus data yourself. I can imagine there > are cases where you would end up with _zero_ useable data because _all_ > of it has some systematic problems or its provenance is unrecoverable. > Finally you either take new data yourself, or identify a couple of > first-rate _visual_ observers you can rely on, and exclude everyone else > (that 80 percent mentioned above). We often do just that in the campaigns that I have been involved with (especially GRBs). If there is standard photometry available that covers the same timeframe as an unfiltered amateur observation, that unfiltered observation is usually dropped since we do not know how the observation was made nor can we transform it without further information, usually not worth the trouble. If we know the amateur in question and have confidence in his/her ability, we will sometimes make contact and figure it out. However, we usually (and probably unjustifiable) assume someone who is using standard filters knows a bit more about photometry and use their data first. > I'm in the camp of wanting folks to use standard filter passbands, and > encourage amateur variable-star observers to do so as well. Yes, there > are some applications where sheer signal-to-noise/time-resolution matters, and > an unfiltered system will work. But ultimately you need to place your data > on a standard system, and despite the loss of limiting magnitude, you must use > filters (and the correct ones, as Ulisse Munari pointed out!). The old > American phrase is that "there's no free lunch", and it applies everywhere > in photon-starved astronomy. You match the capabilities of your > telescope/detector to the stars that are available and the time you have. > Don't bother taking crappy data just because it's some amazing faint object--- > CCDs are not magical, and the GIGO law still applies. I am not against unfiltered photometry (and have a hundred-plus nights of CV photometry of my own to prove it), but for scientific purposes, you have to be careful with how you use unfiltered magnitudes and understand their limitations. It is a great way to start out in photometry since you can work faint with little expense, but I am definitely in the "filtered camp" along with Brian for most projects. > If you are using the magnitudes from an old chart (typical older AAVSO chart, > for example), this is also not V, because the chart magnitudes are given to > only 0.1-mag precision and most of the time there are systematic errors in > zero-point, in scale, or comp-star color---or all three(!). It is possible to > do strictly differential photometry with a V filter using a single star from > one of those charts, which you hope will have standard V somewhere downstream, > but these data need to be reported only as magnitude-differences, not on an > absolute scale, since it will inevitably be wrong. This problem is slowly going away; there are good chart initiatives in progress. It takes time to revise a few thousand charts, so bear with the AAVSO and other collaborating organizations. All new charts have been produced correctly, with good supporting photometry and a database of magnitudes and errors. The others will come in due course. > ...All the > 'unfiltered' stuff would go whether transformed or not, all the "orange" stuff > would go, all the etc. In any publishable analysis it would all be omitted > anyway, so don't keep it, and certainly don't encourage people to waste time > taking/reducing such data. There are enough problems with data nominally taken > on standard systems that dealing with non-standard results is a waste of time. Here I disagree. There are uses for unfiltered or even non-standard filtered measures, such as time-of-minimum for eclipsing binaries, superhump observations of CVs, etc. What is needed is a clear set of instructions as to how to report those observations. You don't need to purge them from the database. Reporting errors occur for any organization. I am sure there are just as many incorrect observations reported to vsnet as to the aavso; highlighting one particular example is counter-productive. > Another source of problems I've noticed is that folks are believing > without question whatever numbers come out from a piece of commercial > software. (It's the "it comes from a computer, it must be right" syndrome.) > Uncertainties seem to be underestimated by factors of several as well. You can > test the software of course by measuring known constant standard stars and > seeing how close you get to "truth", and work on figuring out why errors > occur---is it the software or your observing? By doing such tests on many > nights (every night you observe), you'll get a feeling for what your errors > really are, and learn a lot in the process. Yup. The one that really bugs me as a scientist is reporting magnitudes and times to vast numbers of "significant" digits. That is what comes out of the software, so it must be correct! Arne
Return to the Powerful Daisaku Nogami
vsnet-adm@kusastro.kyoto-u.ac.jp
