Is it possible to do photometry and reliably (whithin reasonable errors) calculate magnitudes on standard passbands from unfiltered observations/measurements? Does the answer to this first question vary if a Luminance of Clear filter is used instead?
If taking measurements without photometric filters, how large can the passband magnitude errors be expected, as you have no colour information at all and only brightness information - would an error bar of about 0.3 mag be a reasonable colour error estimate?
Thanks and CS,
I regularly use a Clear filter (which is compatible to no filter) for observations of Cataclysmic variables. If the B-V color is close to 0 then there is no issue with using a clear or no filter.
For magnitude estimations I use the V references of the AAVSO seuqences or UCAC4 magnitudes.
All this is done using LesvePhotometry as analysis program.
I believe AAVSO recommends a standard filter set for all photometry. No filters means you are just relying on the detectors response curve, the stars colors, unusual emission or absorptions, etc. all of which can vary enormously. Such results are frequently inaccurate, can be misleading, of much less scientific value than properly filtered ones!
Although there are some types of observing in which unfiltered photometry is undesirable, it is an excellent choice in others. It really depends on what your scientific objectives are. It is most useful (1) if you want to study an object that varies extremely rapidly or (2) if small zeropoint offsets are acceptable. Otherwise, a filter would be better -- especially if the target is very red.
Cataclysmic variables are very well suited to unfiltered photometry. Some show variability on timescales of seconds, making it important to observe at a fast cadence. Filters require longer exposures, and longer exposures diminish your ability to detect rapid variations. As an example, let's say that you're observing the famous intermediate polar DQ Herculis in order to study (1) the 71-second spin period of the white dwarf and (2) structure in the eclipse light curve. Both objectives demand a fast cadence (less than 30 sec per image), but they don't require an accurate zeropoint. I think that many amateurs would struggle to get an acceptable signal-to-noise ratio with a filter on DQ Her in such a short exposure time, but it would be easy to do so without a filter. In this scenario, the improvement in time resolution by going unfiltered greatly enhances the scientific value of the data, outweighing the comparatively minor drawbacks.
If you're careful to choose relatively blue comparison stars (B-V < ~0.6) for a CV, my experience is that the resulting zeropoint errors are only a couple of tenths of a magnitude at worst, which is typically quite acceptable for CVs. CVs tend to be rather blue, so bluer reference stars should reduce the zeropoint error.
However, I wouldn't recommend unfiltered photometry of very red stars, as the resulting zeropoint error can be massive (potentially several magnitudes in extreme cases). Likewise, there are some projects in which color information is quite important, such as observing pulsations of RR Lyrae stars. In short, unfiltered photometry is fine in some circumstances, but having filters does increase your observing options.
Thank you all for your answers. I once had a small school project and used a "non-filtered magnitudes to standard magnitudes" method by Brian Warner. I asked for methods in general in the first post in the hope that there would be other methods suggested besides this one. But if anyone wants to look at it, it's in the great photometry book by Brian D. Warner.
Ok. I once had access to several hours of unfiltered image time of the same patch of sky from a meter telescope and wanted to determine the limiting magnitude. I only could acquire a few filtered images from another part in the sky and applied Warner's method, calculated the conversion terms and applied them to the instrumental magnitudes of the very faint point sources from the stacked unfiltered images. Well knowing that I had no true colour information for this part of the sky, I arbitrarily included several tenths of magnitude in the final standard magnitude error bar calculation. Would there have been any possibility of doing better in this situation without having filtered images from the deep observations sky region?
With unfiltered photometry, I'd recommend against arbitrarily increasing the uncertainty to reflect the estimated zeropoint error. Unlike filtered, transformed photometry, it is assumed that there is a zeropoint error for unfiltered photometry, as the CV and CR bandpasses are by definition approximations of the V and R bands, respectively. The reported uncertainty should reflect the signal-to-noise ratio of the data, and if I downloaded unfiltered photometry with typical uncertainties of several tenths of a magnitude, I would interpret the data as having a very low signal-to-noise ratio. Let me give you an example. My colleagues and I recently requested an AAVSO observing campaign on FO Aquarii, specifically asking for unfiltered, fast-cadence photometry. We knew that there would be small zeropoint offsets between different observers, but we had a plan to deal with this. When analyzing the periodicities in the data, our code weighted each photometric observation based on its reported uncertainty. Had observers attempted to incorporate an estimate of the zeropoint error into their uncertainties, it would have caused even high-quality data to become heavily de-weighted. (As an aside, the AAVSO really came through for us; the response to our request was fantastic!)
I haven't read Warner's method, so unfortunately, I can't comment specifically on that.
This comes about a year late, but if this is still an open question for you, I would point you to the following:
This is a follow-on to unfiltered photometry experiment by Arne Henden years past. And similar to Brian D. Warner's work.
The comp star data is taken from APASS data (or other) real-time and applied image by image.
The spreadsheet is straight forward and directly calculates transforms and standard magnitudes.
Joe Garlitz, Elgin, OR