During　heat　treatment　or　mechanical　processing,　most　polycrystalline　materials　experience　grain　growth,　which　significantly　affects　their　mechanical　properties.　Microstructure　simulation　on　a　mesoscopic　scale　is　an　important　way　of　studying　grain　growth.　A　key　research　focus　of　this　type　of　method　has　long　been　how　to　efficiently　and　accurately　simulate　the　grain　growth　caused　by　a　non-uniform　temperature　field　with　temperature　gradients.　In　this　work,　we　propose　an　improved　3D　Monte　Carlo　Potts　(MCP)　method　to　quantitatively　study　the　relationship　between　non-uniform　temperature　fields　and　final　grain　morphologies.　Properties　of　the　aluminum　alloy　AA6061-T6　are　used　to　establish　a　trial　calculation　model　and　to　verify　the　algorithms　with　existing　experimental　results　in　literature.　The　detailed　grain　growth　process　of　the　6061-T6　aluminum　alloy　under　different　temperature　fields　is　then　obtained,　and　grain　morphologies　at　various　positions　are　analyzed.　Results　indicate　that　while　absolute　temperature　and　duration　time　are　the　primary　factors　determining　the　final　grain　size,　the　temperature　gradient　also　has　strong　influence　on　the　grain　morphologies.　The　relationships　between　temperatures,　temperature　gradients　and　grain　growth　process　have　been　established.　The　proposed　MCP　algorithm　can　be　applied　to　different　types　of　materials　when　the　proper　parameters　are　used.