Downsampling source localised data along spatial dimensions

Dear Francois,

I have the following data structure -

40 participants, 10 x 60sec epochs, MEG data Fs 250Hz
The data has been source localised (LCMV) using the 15000V cortical surface (generated using FreeSurfer). As you can imagine, the output has become huge! Furthermore, I also want to estimate source level coherence in 4 frequency bands. This will replicate these files 4 times (400x4), this isn't the most efficient use of the storage space.

So I was wondering if it is possible to downsample the source localised data in spatial dimension?
I know how to downsample data along the time dimension. Ideally, I would like the final size to be 8k vertices x 6000 samples in the time domain (Fs: 100Hz).

Please let me know.


Two approaches: