I do not know whether I understand your problem right. I checked the paper of Sandor et al 2017 and I think you can do exactly what they did. You just need to measure just the trace of fundamental frequency (F0) which you call Primary Element.

I am not familiar with SAP2011 or CoolEdit but you can measure F0 in Praat easily:

http://sail.usc.edu/~lgoldste/General_Phonetics/Tone/Measurement/measurement.html

I do not understand what you mean by 'correlation'. Do you mean that PE and SE have very similar traces or something else? If you say: 'frequency and time domain parameters are highly correlating', I imagine this like: short calls have low frequency and long calls have high frequency - or with increasing duration call becomes higher pitched... But this does not prevent you from taking measurments... So I am confused...