On the OBR, why can the RL value of an event vary when viewed in time domain vs frequency domain?

The RL value for a segment of data in the Time Domain is the average of the RL values in the corresponding Frequency Domain segment.    Since the Time Domain and Frequency Domain data are linked by a Fourier Transform, the conservation of optical power between the two domains is governed by Parseval’s Theorem.  Thus, if you are looking at a device that has a narrow spectral bandwidth, the Time Domain RL value for the device will depend on the wavelength range that you scan over.  

Generally, if you make the scan range much wider than the spectral band limits of the device, you will see a lower RL value in the Time Domian.  You would only see close to the same RL value in the Spectral and Time Domain if you limited the scan range to only cover the flat region between the band limits of the device.  

For example, when measuring a specific wavelength-selective device, Luna expects to get the same RL value  in the Time and Spectral domains if the scan range is limited to the passband.  Widening out the scan range results in finer sample spacing in the Time Domain, but since the device reflects more weakly at wavelengths outside of the passband, the Time Domain RL value is reduced.