The figure looks like the gain should depend only on a single angle, namely the inclination between the direction in question and the axis of the diagram. It appears that this is usually notated $\theta$.
Something like $s=|\cos\theta|$ would fit the figure fairly nicely _and_ be plausible as a physical idealization -- it would mean that the microphone picks up the orthogonal component of the sound wave that goes along the axis.
$s=\cos^2\theta$ also looks like your diagram, and might be more appropriate if you're measuring sound effect rather than amplitude.