Does anyone have any resources for automatic parcellation similar to Freesurfer but with smaller subdivisions of the STG? Specifically, I'm looking for separation of anterior and posterior STG.
Are you requiring these parcellations only on the cortical surface, or for 3D fMRI images? Further, do you need these to be functionally/individually defined, or just defined on the standard brain?
In the past we have used a parcellation based on Talairach coordinates, which [if I recall correctly] split STG into more anterior and more posterior regions; it's roughly split into Brodmann areas. Unfortunately it's not exactly automatic. The coordinates come from the Talairach Daemon online, but it took some programming to automatically extract them and map them onto our surface; and the files I have now are specific to the sourcespace I was using (i.e., it's a MATLAB object indicating, for source point n in my surface tesselation, which Brodmann area it sits in) so I don't think they would be directly useful for you. However, if you are more of a coding whiz than me (or have access to someone who is) you might be able to re-create something like that for your own source space.
As a follow up, you can use the Harvard-Oxford atlas (available in FSLView directories) to parcellate STG into anterior and posterior regions based on the standard MNI brain. The H-O atlas (in a nifti image format) uses numeric labels (9= aSTG, 10 = pSTG), if you load this into matlab you can use this as a masking image for only STG by clearing all the other labels (set all voxels not equal to 9 or 10 to 0, which will just leave the STG in the image - you can further separate this by doing it separately for aSTG and pSTG). This will then give you a masking image to use in other analyses with just the aSTG/pSTG defined.
If you need any further help on this, don't hesitate to ask. I could send you some mask files through for STG, aSTG and pSTG (using the steps above) if you need too.