This forum seems to be closest to this subject; I am hoping to find someone with uncertainty/calibration experience here!

I am calibrating a reflectometer which measures 0/45 reflectance from 100.00% down to 0.00%.

The standard I have is a white plaque (traceable to NRC and NIST by 10nm wavelengths) with about 80% reflectance.

The stated std uncertainty of the original NRC plaque is 0.25%, then my expanded uncertainty ends up being 0.56%.

Now the question: can I scale this uncertainty down so that at 40% reflectance it is half, then 20% reflectance half again?

Otherwise for a 1 or 2% measurement I end up having to state the uncertainty as 0.56% which is to big a part of the chunk!

I am calibrating a reflectometer which measures 0/45 reflectance from 100.00% down to 0.00%.

The standard I have is a white plaque (traceable to NRC and NIST by 10nm wavelengths) with about 80% reflectance.

The stated std uncertainty of the original NRC plaque is 0.25%, then my expanded uncertainty ends up being 0.56%.

Now the question: can I scale this uncertainty down so that at 40% reflectance it is half, then 20% reflectance half again?

Otherwise for a 1 or 2% measurement I end up having to state the uncertainty as 0.56% which is to big a part of the chunk!

I'm not familiar with your instrumentation, but apparently the calibration procedure involves measuring a reflected power level and calibrating this at 80% +/-0.25%. This provides a calibration factor for your power measurement so I would also say that the uncertainty is a % of the reflectivity results and not the reflectivity error itself. So it scales, as you say.

For measuring lower reflectivities, the uncertainty will have contributions from this calibration factor uncertainty and instrument uncertainties that you are apparently already including in the expansion, as well as from nonlinearity between the calibrated level and the measured level, and from noise, background signal and drift effects that may be more significant at low signal levels. So not all of the contributions to uncertainty will scale with reflectivity.

-Mike