Just finished taking a look at the .mat file provided. It looks like there are more TF rows (15002) than there are rows in Options.FOOOF.data (8002). There may also be an issue with using doubles to refer to RowNames instead of a cell array of strings.
If I am understanding this correctly, the aim is to plot FOOOF models of sources on the cortex, yes? If so, I have not yet looked into this (although I could try to add it by March) and so for the time being, computing FOOOF models on sources is not yet a 'vetted' process. Essentially, while no errors might be thrown for computing FOOOF models at the source level, the process interprets every source as a channel, and requires there to be an equal number of TF rows to the number of FOOOF models (visualizing the data requires 1 - 1 replacement of rows).
For a quick fix, you can visualize the FOOOF models by trimming the TF matrix to 8002 rows and re-importing the data, as well as converting the RowNames to strings in a cell array:
a.RowNames = cell(1,8002);
for k = 1:8002
a.RowNames{k} = num2str(k);
end
However, you may notice that some of the FOOOF models are not great fits (although I cannot explain why they are this bad).