Spatial Autocorrelation for Surface Data


Is there a way to account for the spatial autocorrelation issue for surface data? I am aware that the autocorrelation function (acf) implemented for volume data would need to be adapted to work with surface data given that surface data are not in a grid (see here). But I was wondering whether there was a solution or a recommended approach.
Would using 3dttest++ permutation method be appropriate to determine cluster thresholds if using surface data?

Thank you very much!