Long-term post-seismic gravity change (PGC) of the 2004 Sumatra-Andaman earthquake (Mw9.1-9.3) detected by GRACE + GRACE-FO is modeled sufficiently by an exponential function until the 2012 Indian-Ocean earthquake (Mw8.8) but gravity increasing became constant after that. An interpretation is PGC of the 2004 earthquake was almost completed in 2012 and that of the 2012 earthquake proceeded constantly. Assuming this, I calculate their spatial distributions. PGC of the 2012 earthquake was expanded along the faults of the 2004 earthquake and the peaks of PCGs of the two earthquakes appeared at a same location. This suggests another interpretation, namely, PGC of the 2004 earthquake has another long-term component expected by Burgers rheology and PGC of the 2012 earthquake is slight. Spatial distributions of PGC on this assumption support this interpretation. This is the first result showing Burgers rheology is appropriate to consider the viscosity of upper mantle based only on satellite gravimetry.