Flood　is　one　of　the　major　disasters　for　the　mankind.　In　Asia,　riverine　and　flash　flooding　occurs　frequently.　Therefore,　accurate　and　timely　flood　mapping　is　necessary　for　saving　human　lives　and　reducing　economic　costs.　In　this　paper,　the　flood　mapping　for　Pakistan　and　Kashmir　regions　is　done　jointly　using　Multi-Spectral　Water　Index　(MuWI)　and　Machine　Learning　classifiers　that　includes　Support　Vector　Machine　(SVM),　classification　and　Regression　Trees　(CART)　and　Random　Forest　(RF)　with　Sentinel　2　and　Landsat　8　images　dataset　in　2020.　The　overall　average　validation　accuracy　results　before　and　after　flood　with　MuWI　based　SVM,　CART　and　RF　are　evaluated　and　compared　in　Google　Earth　Engine　(GEE).　Flood　difference　maps　were　also　generated.　Results　show　that　the　shadow　areas　are　accurately　identified　with　MuWI.　Jointly　taking　MuWI　and　spectral　bands　obtained　superb　flood　mapping　performance.　SVM　outperforms　the　CART　and　RF　using　Sentinel-2　images.　Using　Landsat-8,　both　SVM　and　Random　Forest　proved　equally　effective　in　identifying　the　flood　pixels　from　non-flood　pixels　as　compared　to　CART.　The　information　derived　with　the　proposed　methods　may　be　important　for　disaster　risk　management　and　urban　planning.　©　2021　IEEE.