Tue, 02/13/2024 - 01:06
Hi,
Instead of transforming tricolor magnitudes to Johnson magnitudes, what variables would comparing BVR and tricolor be useful? For instance, there may be novae/supernovae where TB may be more sensitive to Hb compared to B, and R is more sensitive to Ha compared to TR if tricolor OSC has IR filter (i.e. unmodified DSLR / OSC CMOS with UV/IR cut).
I am exploring ways on how tricolor data from smart scopes would be useful without transformation.
Thanks,
Raymund
I’ve been doing YSOs with DSLR equipment for several years now. Most are neutral in color, spectral classes from O and A to K, with good results as TG. I had much more trouble getting good results on semi-regular variables or other red stars. I found that I had a lot of trouble getting the transformation spreadsheet to work as well. Tricolor values are different from the comparison values provided in the photometry tables, so there is a genuine need to work these problems out. One-shot color imaging is here to stay- I’ll be watching this thread!
Thank you for mentioning YSOs, John. I will keep that in mind - in addition to novae.
The spectral profile of OSC colors have more compressed peaks than BVR - which is farther apart, particularly B and V. I actually tried combining all 3 channels into a single luminance with IR cut (since the 3 channels have peaks that are compressed), and did photometry of an RR Lyrae (so neither too blue or red) star as CV filter, and had good results.
Instead of transforming, we can have the TB peak cover the range in between B and V; and using an IR cut (i.e. 400 - 700 nm) on the OSC, comparing R and TR could yield information with emissions present past the 700nm mark that R can capture, but not TR.
On the other hand, with modified DSLRs or OSC without IR cut, if all tricolors are significantly brighter - particularly TG - this could yield information on the IR emission of the variable (due to IR leak on all 3 channels). I think there is value in preserving the TB, TG and TR magnitudes and not necessarily transform them.
And you are right, OSC is here to stay. With smart telescopes already on the rise, this coverage between B and V (covered by TB), and emissions in R that TR could not detect due to IR cut could also be valuable; as well as IR emissions from unfiltered OSC / modified DSLR.
I think there is also a need to make the "Notes" column of the AAVSO extended format as a required field when submitting TB, TG or TR (or even CV or CR) since it is important to know whether the observation made use of an IR cut filter, and the type of sensor (since different colored sensors can have varying profiles at the IR leak region).
Raymund
Raymund wrote: I think there is value in preserving the TB, TG and TR magnitudes and not necessarily transform them.
I agree with this comment, but it is my opinion that anyone calculating non-transformed magnitudes (whether with RGB OSC cameras, or with monochrome cameras and photometric filters) should know the transformation coefficients of their equipment.
The reason for this opinion is that TCs will show you the component of the error in your measured non-transformed magnitudes due entirely to the colour index difference between the target and comp stars. In the case of DSLR cameras (the only type of OSC camera I have experience with) such errors can be substantial.
For example, the TCs for my Canon EOS DSLR from a few years ago were: Tbv 0.409, Tvr 0.711, Tv_bv -0.124, Tv_vr -0.241, Tr_vr -0.531.
The colour transforms (e.g., Tbv) for monochrome cameras with photometric filters may approximate unity. The transforms above are far from unity.
The filter transforms (e.g., Tv_bv) for monochrome cameras with a photometric filter may approximate zero. The DSLR transforms above are far from zero.
For the filter transform Tr_vr, a value of -0.531 means that, if the target and comp star V-R difference is 1.0, the error in TR will be 0.5. For Tv_bv, the error in TG for a target-comp B-V difference of 1.0 will be 0.124. These relationships are linear, so in the case of the TG value if the target-comp B-V difference is only 0.1, the TG error will be only 0.012.
Concerning smart telescopes, the Seestar50 was found by Andrew Pearce to have a Tv_bv of only -0.02 (I hope I quote him correctly from this Forum), a usefully low value for determining TG and comparing the result with V.
Roy
While I agree that these devices are "here to stay", unless you have decades worth of consistent data on the same instrumental system, then your results for the variables stars will be difficult to link with other stuff past/future. And in 10 years you are likely to be using some other hardware, introducing yet more inconsistencies into the literature. From previous discussion threads (large catalogues of tri-color photometry now available), it seems like the generic TG might transform with modest color terms to either Sloan g or to the GAIA BP zero-point. The TR might go to either Cousins R or Sloan r. In any case you will need to do the transformation using standard stars for every camera + lens or telescope set-up involved, and to see which standard system is most reasonable.
You wrote: "...I had a lot of trouble getting the transformation spreadsheet to work as well..." Could this be because the transformations are not linear, possibly bi-linear or needing a quadratic?
\Brian
I’m all for reporting TG as TG. One-shot color imagers are specifically designed and engineered to REPLICATE HUMAN VISION.
They obviously do this extremely well, as photometric systems have a hard time approximating something as messy as human perception. It’s no surprise that transformations are difficult.
The value in TG observations is their connection to actual visual observations. It’s truly a way to quantify a visual observation accurately. I don’t think this has been appreciated by the community in general.
Photometry has been a very challenging subject for me personally. I find the math hard to understand with a high school education- I have no working experience with Excel, and the terms used don’t help. I’m sorry I can’t answer your question as put. I suggest that if transformation is so important, you should all apply yourselves to developing methods, techniques and language that makes the subject easier for laypersons.
I’m grateful that the AAVSO has a very inclusive policy on observations. I certainly hope that continues.
Raymund (correction, John) wrote: "The value in TG observations is their connection to actual visual observations. It’s truly a way to quantify a visual observation accurately."
I presume you perform differential aperture photometry, and that you use AAVSO charts and comparison stars for calculating your TG values, or your own charts with comparison stars chosen by you. The magnitudes of the comparison stars you use would, I presume, be Johnson V magnitudes.
The basis for your TG values is therefore a standard photometric system, not the response of the human eye.
Your also wrote: "photometric systems have a hard time approximating something as messy as human perception."
The purpose of photometric systems is not to approximate the human visual response, but to standardize measurements of magnitudes within defined passbands, and those passbands extend well beyond the capability of the human eye, toward both the blue and red ends of the electromagnetic spectrum.
If you are unable to determine transformation coefficients and use a DSLR camera to measure TG, TB and TR magnitudes, it is important that you select comparison stars with colour indices as close as possible to those of the target. The scientific reasons for this are in my post above.
Roy
TR, TG and TB comprise a separate filter set (just like Sloan - although have more degree of variation related to spectral efficiencies of colored sensors and use of IR cut), and TG is close to V mag that transforming with appropriate stars would yield good V mag values.
Transformation can make magnitude estimates closer to V mag. However, it could also be possible that it may be off due to observer (not adept in transformation) or instrument (filter wheel issues - in case of V filter transformed to V to eliminate possible variations in manufacturing).
With more data points in tricolor, we could appreciate more the mag difference with BVR - which could provide importance in terms of capturing spectral lines not strongly detected by another.
Raymund
I have been photometrically reducing the Seestar observations of observer EJOC for nova V1723 Sco using Tycho Tracker. The TR magnitude resembles closely the R magnitude of observer HMB. The TB magnitude is much brighter than B - probably due to difference in peak sensitivity / QE. TG is brighter than the reported V magnitudes. It is interesting to note though that the V mags themselves are scattered (observers HMB vs LBEC).
However, the V and B mags of observer HMB are much closer compared to R mag - which also resembles EJOC's data (i.e. TG and TB are much closer compared to TR).
I think this is where appreciation of BVR and TR/TG/TB comes in as separate filter sets - not necessarily needing to transform.
Raymund
Hi
There appears to be a systematic offset between HMB and LBEC observations, presumably due to different comparison stars used. That is easy to account for as the observation datasets themselves appear consistent. My own more limited BVIR photometry in the last few weeks more lines up with LBEC observations.
Regards
Andrew
I believe the Seestar has the IMX462 sensor. If the QE curve I have seen is correct, the green and blue filters have very large red leaks, obvious by 700nm and peaking at 800-850nm. The QE peaks at 800-850 are higher than the main green and blue peaks.
Roy
Hi Roy,
The images had IR cut on, but it may be possible there was still IR leaks brightening TG and TB mags.
Raymund
Hi Raymund,
Yes, I think you may be correct. The transmission plot from ZWO for the IR cut filter shows that more than 90% of light at about 680nm passes the filter. If the QE plot I have seen for the IMX462 sensor is correct, then there may still be some red leak from both the green and blue filters at about 680nm.
The ZWO IR cut filter has a nearly rectangular profile from 400nm to about 700nm, but the trajectory of the transmission plot starts to fall steeply from about 680nm.
Roy