Another photometry query (calibration and air mass)

[billyharris72]

Hi all:

Another request for clarification that I'm hoping someone can help me with.

In most of the material I read online about photometry I keep seeing discussion of the need to:

a) Account for extinction by measuring airmass

b) Calibrate the instrument on a standard star field to correct for differences in instrument response.

In both cases I am struggling to see the need in the case of differential photometry and am hoping someone can set me straight. My reasoning is as follows.

As regards extinction

a) If the FOV is relatively narrow the we can ignore extinction within an image (depending on how accurate we need to be, obviously).

b) If we are taking multiple exposures that between them cover a wider area of sky then we would expect the ADU registered to vary between shots. We would also expect the colour (B-V) response in the instrument to change. But, given that we are always taking our comparison stars from within the same image (i.e. we calculate the relative brightness of the target and comparison stars for every individual exposure) I don't see how this introduces any error. If star A is twice as bright at a given wavelength as star B, that is not going to change with atmospheric extinction. If, between exposure 1 and exposure 2, the ADU drops by 10%, the calculated magnitude for the target should not change.

As regards colour response:

a) Instruments may well respond differently to different wavelengths of light. If a sensor operating without a filter is more sensitive at the blue end it will tend to see "equally bright" objects as brighter if they are bluer. This can't be calibrated out (as we don't know the colour of the various photons), which is why we use filters.

b) Within a given filter's spectral response, an instrument may be more sensitive at one end. With images from a single filter this also can't be calibrated out, as we don't have information about the distribution of photons within the transmission range of the filter. (I assume we could carry out some kind of correction if we were doing BVR photometry, as the relative brightness in each band would give us more information that would let us correct the red and blue ends of the V band response).

Can someone tell me what I'm missing? My own approach when conducing photometry (which may or may not be relevant) is:

1. select a target star, (sometimes) a check star and a group of comparison stars.

2. select an appropriate annulus and grab the ADUs.

3. Log transform the ADU values.

4. Fit a regression model using the magnitude as the dependent variable and log-ADUs as the independent.

5. Check the residuals (for wide field shots with a DSLR I'll tolerate about 0.05 mag provided the error does not appear systematic) and make sure the model is generally behaving itself.

6. Calculate the target value from its log-ADU.

7. Repeat for every exposure or batch of exposures (i.e. don't re-use the regression model).

So, within the limits of this approach (and assuming I'm using a V filter only, or DSLR green filter) should I be correcting for either of the above and if so how would I go about it?

Thanks,

Billy.

 

[vlaiv]

Both mentioned things should be done for absolute photometry.

For relative photometry, if you take such exposure to have your reference star and measured object within same FOV, and your FOV is relatively small, you don't need to do these adjustments.

However, this is often not the case (well it used not to be the case) - very few stars were used as a reference and one needed to move the telescope to different position in order to do reference exposure. If reference exposure is at different ALT than the target, and substantially different part of the sky (meaning possibility of different transparency due to different direction) you need to account for difference in air mass. Nowadays there is enough information in catalogs that most of fields will contain star with a known / measured magnitude (in different bands) so you can just use any available reference star in your FOV.

As for standard filter responses, if you are doing relative photometry and depending on required accuracy (can you even achieve needed SNR in the first place to go that accurate?) - you don't have to worry too much about different responses - both that of a filter and that of a sensor. This is due to fact that most stellar objects are pretty close to black body radiation curve - and that one does not change significantly over the observed wavelength ranges. There are absorption and emission lines, but those tend to be very narrow and account for very small percentage of difference to black body. All of this is of course related to relative photometry, where you use same filter / sensor response curve for both reference and target objects.

If you really want to get close to true photometric standard filters with non standard filters and arbitrary sensor response curve, there is a "simple" way to do it. It would involve simple objective grating (or in converging beam like star analyzer or DIY printed grating) - doing spectrum of both reference and target objects, and then calibrating that spectrum with photometric data that you obtained to get proper/calibrated photometric flux spread over spectrum. Then it would just be the case of integrating spectrum multiplied respective filter response curve.

[billyharris72]

Thanks Vlaiv, that makes sense. I'd wondered about that idea with a Star Analyser or similar before - might be worth a try. Billy. 

[robin_astro]

On 12/08/2018 at 16:48, billyharris72 said:

 I'd wondered about that idea with a Star Analyser or similar before - might be worth a try. Billy. 

Yes you can do photometry with a Star Analyser and it is fun to try it but in practise it is more involved  than it appears and not really worth it except perhaps in specific circumstances eg where you need to measure high speed events in several wavelenths simultaneously.  If you are interested and you are a BAA member you can listen to the talk I gave on this at the the joint BAA/AAVSO meeting last month entitled "pushing the limits using commercial spectrographs"

https://britastro.org/video/13862

In spectroscopy it is more commonly found  the other way round where amateurs convert their relative flux calibrated spectra to absolute flux using photometric V measurements

Robin

 

[robin_astro]

I am not a photometry specialist, hopefully one will come along, but regarding correcting for extinction and transforming data to a standard photometric system, the AAVSO manuals have some good information

https://www.aavso.org/dslr-observing-manual

https://www.aavso.org/ccd-photometry-gude

particularly the chapters on transforming the data

Again there is a video of a talk at the BAA/AAVSO meeting by Gordon Myers entitled "Applying transformation and extinction to magnitudes estimates"

Robin

[Whirlwind]

Hi 

It depends on what you are trying to achieve.  If you want to upload data to AAVSO etc then you do need to keep to their general methods.  That is because they want data to be relatively comparable by method.  However for you own purposes, out of interest, then there are some things you can 'drop' because they are only undertaken to standardise data reports (otherwise the data would be all over the place when you compare to other people's data)

If you are doing it out of your own interest then dropping conversions to magnitudes is best and just record data by flux.  This is how most professional observatories undertake photometry.  Except for absolute photometry (e.g. SDSS type surveys) then time photometry surveys do not need to be converted to magnitudes as it is an arbitrary scale.  If you want to know the depth of the transit then it is easier to know this from normalised flux photometry rather than via magnitudes.  In effect you count the total flux in an aperture of the target and divide this through by the flux of a brighter comparison star or stars (in case one of the comparisons is variable).  This gives you a relative flux and generally you then normalise this by dividing through by the median value of the relative flux.  This provides a normalised flux light curve.  

In terms of airmass there is usually a residual effect but is limited to basic second order curve.  This arises because the comparison star is likely a different temperature and the atmosphere has greater/lesser effects dependent on the colour of the stars.  It is most pronounced when the target star is hot and blue and a comparison star is cool and red (or vice versa).

When undertaking photometry in this way you don't have to worry about instrumental response (especially at the levels of accuracy we are talking about here).