Return to Home Page
I take it that Gary refers to CCD observations. This then applies to what I do with timeseries photometry and especially in monitoring faint CVs (15 to 19th magnitude). Although I measure unfiltered, the following would also apply to measurements with a V filtered CCD camera. In the latter case, the accuracy would be better, but the precision would be worse (on the same faint variables). I reply in between the questions /// I guess I would like to hear what steps other folks are taking to make sure that their observations are as correct as possible. 1. What checks are being made? The proper identification of the variable is very important. I use one comparison star (C) and one or more check stars (K). All of them have R magnitudes traceable to a calibrated sequence from published data (Henden, Loneos, GSPC.....) or from data bases with reasonably accurate values (+ or- 0.2 magn); Tycho2, UCAC1 I try to achieve the following: C and K stars have non red colours (B-V<0.6) since most CVs are very blue (B=V=R approx). One K star is located close to the variable under test (V) and has about the same R brightness. This is especially important for timeseries photometry in order to probe the signal to noise ratio of the variable itself (in addition to prove that C is of constant brightness). C is at least 2 magnitudes brighter than V in order to reduce the noise level introduced by the C star in the C-V differential. The differential can be larger as long no saturation (of C) occurs. I do make at least two images of the field. For timeseries (hundreds of images) the integrated value of C must be constant and only changing slowly in time (due to sky extinction). Intermittent clouds and badly tracked images will affect the value of C. The value C-K ought to be constant and equal to the difference between those stars' R magnitudes. If not, this is used to discard images in addition to those with a smaller C value (caused by a bright sky background pe due to clouds..). See 3. 2. How many comp stars do you use? One (C), but indirectly 2 or 3 from K-C differentials 3. How large a spread of your estimates between comps stars do you allow before you reject an observation and not submit it? With CCD images, differentials are measured between total nett pixel readings between the relevant stars (measured over 4 to 8 pixel radii /using software mostly), no visual comparison estimates ought to be done by the operator (except for quick verification). The value(s) C-K must correspond to the expected R differential(s). If not, other 'K' stars in the field need to be checked. The mean deviation of C-K values must be smaller than 0.15. Otherwise it's reported as 175:CR. Usually this testing will take place on the first such measurement of the variable. This will establish a proper set of constant C and K stars for future use. 4. Do you look at your previous observations of the same star to make sure that the current estimate is not too far off? How far off is too far? No! Every measurement has to fulfill the requirements above. They will give confidence in the measurements. 5. Do you look at the quick look file or the light curve generator and see what a star is doing to make sure that your observation is not too far off? Not applicable here. It generally is a bad approach. 6. What do you do with a CV that is in outburst and you are the only observation in outburst and your estimate is 1-2 magnitudes brighter than typical that previous observations in the light curve generator? Report it in outburst! 7. Suppose a star is an LPV. Your observation differs from the last one, 10 days ago in the quick look file. How large a deviation would you accept before discarding it. N/A 8. How much apriori information do you collect to make sure that every observation is within limits. How big are your limits? What type of star do you impose this process on? Requirements under 1. specifically the proper identification of the variable is important. 9. What other suggestions do you have for screening observations? Specific things you do. I am sure that everyone want to turn in accurate data. What is needed is a set of guidlines. I would like to hear just what others are doing. None! It is important to know what you are doing and to be confident by doing it properly. I know that experts like A Henden give courses on this. If you feel unsure, just attend one of those. BTW I measure unfiltered and although unstated in the reports I dare to put claim to accuracies well within 0.3 magnitude CR (which approximates R, V and B for CVs) up to magnitude 17, and 0.5CR up to 18.5. I hope that simultaneous observations with larger telescopes and with filtered CCDs would periodically scrutinize some of those measurements, just to validate that claim. Regards, Berto Monard / MLF Bronberg Observatory / CBA Pretoria >>> <BailyHill@aol.com> 02/12/03 03:39AM >>> In a message dated 2/11/03 12:14:24 PM, lolife@bitstream.net writes: << For non-survey type CCD observations it just baffles me why anyone would submit any observations that were not as careful and as correct as possible. Observations are worse than useless if you don't take care in the reduction and reporting. >> I guess I would like to hear what steps other folks are taking to make sure that their observations are as correct as possible. What checks are being made? How many comp stars do you use? How large a spread of your estimates between comps stars do you allow before you reject an observation and not submit it? Do you look at your previous observations of the same star to make sure that the current estimate is not too far off? How far off is too far? Do you look at the quick look file or the light curve generator and see what a star is doing to make sure that your observation is not too far off? What do you do with a CV that is in outburst and you are the only observation in outburst and your estimate is 1-2 magnitudes brighter than typical that previous observations in the light curve generator? Suppose a star is an LPV. Your observation differs from the last one, 10 days ago in the quick look file. How large a deviation would you accept before discarding it. How much apriori information do you collect to make sure that every observation is within limits. How big are your limits? What type of star do you impose this process on? What other suggestions do you have for screening observations? Specific things you do. I am sure that everyone want to turn in accurate data. What is needed is a set of guidlines. I would like to hear just what others are doing. Clear Skies Gary -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. "The CSIR exercises no editorial control over E-mail messages and/or attachments thereto/links referred to therein originating in the organisation and the views in this message/attachments thereto are therefore not necessarily those of the CSIR and/or its employees. The sender of this e-mail is, moreover, in terms of the CSIR's Conditions of Service, subject to compliance with the CSIR's internal E-mail and Internet Policy."
Return to the Powerful Daisaku
vsnet-adm@kusastro.kyoto-u.ac.jp
