"Each level of the mip-map is computed by convolving the environmentmap with a spherical Gaussian that has twice the standard deviation of the previous level"

This is what a confusing line in the paper I'm trying to implement says.

Now for this environment map I have values linked to (theta,phi) pairs. But I'm kinda in the dark as to what they mean with 'spherical gaussian'.

a 2D gaussfunction defined in theta/phi ? If so, how?

As you can see I'm kinda stuck, so any help would be much appreciated