When do we get to the hard questions?

This is not completely "reciprocity failure" - a phenomemon directly linked to time. As we all know, "reciprocity" is the idea that an exposure of 1 second at f/16 is equal to an exposure of 1/2 second at f/11; and equal to 1/4 second at f/8; and 1/8 second at f/5.6. However, that "law" fails at *very* long times ... an indicated exposure of 10 seconds may well require 20; one minute exposure may require two minutes, etc. - depending on the characteristics of the film. The data sheet for the film usually specifies *rough* compensating factors for long exposures.

What you describe is something more... and could get theorectically *complicated*.
There is no assurance that the increase in light at "dawn" would be linear; if so, one might "strike an average" - but even so, reciprocity compensation is definitely NOT linear.

So - it is a problem. I was going to use the catch-all for exposure problems - "bracket" ... but that wouldn't work either.

Uh ... try it and build up experience? Take notes and ... hope for the best.