Guilin Sun
Ansys Employee
If so, I have not seen any theory that guarantees the sum of overlaps to be unity, like transmission and reflection. "overlap" is to compute the similarity of the modes. The overlap assume no reflections, but in physics there is some reflection. Some details can be found here:
https://support.lumerical.com/hc/en-us/articles/360034396834-Understanding-the-mode-overlap-calculation
ideally, we would know the reflection, which is not necessarily part of the incident fields.
So, I would suggest that you can simply ignore the difference to unity.