Hello,
I am currently working on OPM forward modeling beyond a single point, for a specific prototype system. This is complicated to discuss generally however because many different flavors of OPM exist and it's not as simple as integrating the field over the sensing element (gas cell) as it is for SQUIDs (coil). The technology is based on magnetic resonance. The sensor I'm working on measures the atom polarisation precession frequency. Others measure light absorption/transmission. Both use one or more lasers. So we need to take into account many aspects: shape and size of the cell, directions of the lasers, intensity profile and diameter of the laser beams, and in my case integrating the signals with different phases and frequencies due to field inhomogeneities, and then taking the max spectral peak as the value extracted by the lock-in amplifier.
There are yet other considerations even with a single point approximation: the sensitivity of the design (1-d, 3-d, scalar), if it operates in earth's field, the effects of that field.
In my case, for simulations it's also non-trivial to model the noise. I have to look further into that.
So in conclusion, we need to know many details of a specific OPM design to properly model it, especially if we want to go beyond the single point approximation. So this might have to be done on a case by case basis.