In geostatistics, we can assess and analyze regionalized variables. Tectonic stresses releases in seismically active regions by occurring earthquakes. We can calculate and estimate the seismic energy released by an earthquake. Now, the question is this: Is it possible to assess 2-D or 3-D spatial variation of the seismic energy released in a region, using geostatistics rules?
An intrinsic assumption in using geostatistics is that one can approximate a distribution by a polynomial surface. In practice, maybe you can do this in a very simple geological situation, such as an isotropic shockwave arising from sedimentary compaction in horizontal sediments. When it comes to major seismic events at tectonic plate edges I think the seismic energy distribution might be too complex. For example, the seismic energy is attenuated at different rates on either side of a subduction zone, and tends to be channeled along the ocean floor according to local bathymetry and structure (especially along mid-ocean ridges). Also, the shock wave distribution will be different according to whether the source is shallow or deep. Trying to account for such features in a geostatistical model will be a nightmare of complexity - good luck!