I am in the process of learning the ins and outs of stellar photometry and plan to contribute measurements to the AAVSO database. I have planned to use the new Baader Bessell UBVRI filters. But I just recently found on the Baader website a document that records spectral transmittance measurements on those filters, and the results make me question my plan. https://www.baader-planetarium.com/en/blog/results-of-the-tests-made-by-galli-gianni-about-the-new-baader-ubvri-bessel-filters/
That document shows that the B filter has a red leak at and beyond 1200nm, the V filter has more transmittace than the Bessell 1990 standard at the Ha line, and the I filter transmittance extends to some 1150nm, far beyond the standard cutoff at about 900nm. The I filter transmittance also shows that it never goes close to zero but climbs again into a red leak at and beyond 1200nm. There are some other much smaller deviations from the standards in the steepness of the band edges and their placement in wavelength. One can question the accuracy of the measurements, but that is largely laid to rest by the fact that Baader has published them.
These results have caused me to reconsider my filter plan, and any comments about the published results and/or answers to the following questions that I have, will be greatly helpful and appreciated. The context for my questions is that I will be employing a QHY268M camera which has the Sony IMX571 CMOS sensor. The only information regarding the sensor's quantuum efficiency that I have found is a graph that extends only to 1000nm where it has fallen to about 4% but is still fairly steeply descending there. Extrapolation by eyeball suggests that its QE will be at most 1% and probably less at 1200nm and beyond.
1. Will the small deviations in band-edge steepness and placement be easy to handle with transformations? just Linear ones?
2. How serious should one expect the V filter's 4% transmittance at Ha to be for objects having strong Ha emission or absorption?
3. Despite the uncertainty regarding the IMX571 quantuum efficiency beond 1000nm, how significant are the B and I red leaks likely to be for late-type stars? Will measurements made with those two filters be of any value at all without transformations? Can measurements with them be adequately handled with appropriate transformations? ... presumably with the addition of quadratic and/or cubic terms.
4. How significant will be the non-zero I filter transmittance between 900 and 1150nm?
5. The AAVSO Extended File Format does not explicity list the Bessell UBVRI filters as options in the DATA/FILTER field. Should Bessell measurements be reported as U,B,V,R,I: Johnson/Cousins filters and then include a note stating use of the Baader-Bessell filters be inserted into the DATA/NOTES field?
6. Does the DATA/NOTES field typically list the transformation equations used?
7. Overall: will these filters likely enable one to produce useful data? Or is detailed testing of them the only way to resolve these questions?
--- Mike
There is an interesting muddle between the filter recipes that Mike Bessell prescribed using colored glass versus what is being marketed, which now are mostly interference filters. One often sees even in the professional literature the implication that Bessell somehow defined a UBVRI photometric standard system, which he did not. He merely gave recipes for colored-glass filters to minimize color terms in transformations to the Johnson-Cousins standards with either photomultiplier tubes or with CCDs (and CCDs as they were prior to 1990, mind you). Thus per item 5 in your note, Bessell-filter photometry should be identical to Johnson-Cousins, and no distinction needs to be made. The source paper is:
https://ui.adsabs.harvard.edu/abs/1990PASP..102.1181B/abstract
See specifically Figure 20 here. Nowadays one sees 'Bessell' filters offered that have rectangular passbands, which as you can see by comparison to the figure are not really anything like what Bessell specified. Calling them 'Bessell filters' seems to be completely bogus to me. See his most recent work on this subject here:
https://ui.adsabs.harvard.edu/abs/2012PASP..124..140B/abstract
...specifically Figure 9, to compare with the Baader or other filters that are available. (It would have been useful if Galli Gianni had plotted the original Bessell passbands over the new Baader passbands in the link you gave.)
What this means from a practical standpoint is that inevitably data from your specific telescope/filters/detector(s) will _always_ need to be transformed to the standard system. Possibly a red leak? Measure it by observing some very red standard stars! Just now (springtime) Wolf 359 is available, which is the very reddest Landolt standard (V-I=4.00), and Arne Henden has taken data that includes the more ordinary stars in that field. In summer/autumn one can observe the Landolt SA 110 field which has several very red (reddened) stars with quite accurate calibration. Don't be surprised if you need quadratic terms in the transformations that include very red stars.
\Brian
We've been wrestling with this problem for at least the last three years, and more so as the supply of the dyed glasses that Bessel specified have dried up. To a filter manufacturer interesting in making filters for amateur astronomical photometry, the "sloppy" bandpasses of the J/C system must have seemed silly when it is so easy to design interference filters with clean, sharp well defined passbands*. Thus we have Bessel filters and "Bessel" filters. My understanding is that the situation is in process of being rectified with interference filters having passbands that match dyed-glass UBVRI filters and have deep blocking in the near IR.
As I recall, Arne Henden was working on a synthetic photometry program to explore the "solution space" of filter passband shapes about a year ago. While he did not describe it in detail, I asked and gathered the idea was to perform synthetic photometry on a catalog of spectra, then compute the transform coefficients for various filter inputs to the standard filters. The model should reproduce known transforms. Arne might care to update us on this initiative.
--Richard
* i.e., what the much larger market for RGB and narrowband filters wants.
Yes, I understand the practicalities of the interference filters versus colored glass, but just don't call the former 'Bessell filters'. (And it is two l's --- Bessel with one l is the 19th Century mathematician). Mike Bessell, and Andy Young before him, have published a long series of papers decrying the untransformability of the rectangular passbands. But that's what we have, so we have to deal with them.
In re synthetic photometry on spectra: alas, the systems are defined by color-indices of specific real stars on the sky, and in most cases there is no calibrated spectrophotometry for those specific stars. So (my opinion) defining passbands with model spectra is not really going to help somebody with an off-the-shelf system, and each set-up will have to be transformed using lots of observations of lots of standard stars on many nights. Again, a large literature on this from folks like Vytautas Straižys and Ulisse Munari. There's also the trenchant comment on this in the Allan Sandage paper about the Mount Wilson photometry:
https://ui.adsabs.harvard.edu/abs/1997PASP..109.1193S/abstract
...see the long footnote on page 1196.
\Brian
Thank you very much Richard and Brian for responding to my posting. The points that you have made have caused me to read and think a good deal more deeply about photometry and filters. I think that I have now come to a somewhat different but more accurate understanding of the situation than I previously had. Some observations and puzzles remain, however.
I spent a goodly amount of time more carefully reading Bessell's and some other papers. This made me wonder why the Baader-Bessell filter transmittances at Baader's website look nearly identical to Bessell's system passbands which also include the detector quantum efficiency. Putting all of these into graphs for comparison and then adding a typical CCD or CMOS detector let me see that the detector response changes rather slowly across the narrow filter passbands, leaving the resulting passbands looking about the same as the transmittances. So that question seems to be resolved.
I also found that, contrary to what I have heard, the glass filters that Bessell specifies in his publications are still available to buy. The sole exceptions are the GG385 and BG12 for the B passband whose production by Schott was discontinued for reasons that I have not found. Looking at Hoya colored glass filters, I suspect that their L38 and B390 glass filters might yield a useful substitute, although the edges would have different slopes, and it would be some 5nm bluer than the Schott. All these glass filters are available in polished, single, 50mm square pieces from Edmund Optics, and I'm toying with the idea to put together a V filter for myself to see how that works out. One caveat is that the surfaces of these are all 60-40 rather than 80-20 which I think would be better.
Before launching into cutting and gluing glass filters, I would model the anticipated performance by synthetic photometry. I would plan to use the Pickles spectra in that job which, despite their age and pedigree, would be adequate for my purposes. That should rather quickly show whether the idea has a fatal flaw. For this and similar reasons, I think that synthetic photometry can be a very useful tool so long as one recognizes its limitations.
A final thought that has me wondering ... It is clear that, besides historical precedent and practice, a principle advantage of the UBVRI broad passband photometry is that amateurs can use it to reach dim stars. For brighter stars it would seem that low resolution spectroscopy, which contains a lot more physics than broadband photometry even for slitless spectroscopy, might be a profitable general undertaking. Properly taken, calibrated and recorded, such spectra would be pretty easy to reduce to UBVRI passbands, either as a standard practice or on an as-needed basis. Is there a general consensus of thinking along these lines?
Thanks again.
--- Mike