In　an　air-space-ground　integrated　network　for　emergency　scenario,　assisted　by　the　satellites,　the　UAVs　(Unmanned　Aerial　Vehicle)　are　used　as　base　stations　that　serve　the　ground　users　to　recover　the　communications　of　the　disaster　area.　The　deployment　delay　of　the　UAVs　greatly　affects　the　rescue　efficiency.　Meanwhile,　due　to　limited　number　of　the　UAVs,　each　UAV　should　spare　it　effort　to　serve　the　ground　users.　In　the　aim　of　solving　above　problems,　we　have　designed　a　method　for　deploying　UAVs.　In　this　paper,　we　study　how　to　minimize　the　number　and　deployment　delay　of　UAVs,　which　is　divided　into　three　steps　to　solve　the　problem　of　UAVs＇　deployment.　First,　a　K-Means　algorithm　(UB-K-Means)　based　on　users＇　bandwidth　requirements　is　proposed　to　cluster　the　ground　users　and　determine　the　two-dimensional　coordinates　of　UAVs.　With　such　cluster　algorithm,　the　UAVs＇　number　can　be　decreased.　In　addition,　we　considered　the　bandwidth　carried　by　the　UAV.　Then,　the　UAV＇s　transmit　power,　the　channel　gain　of　per　unit　distance,　the　threshold　of　SNR,　the　ground　user＇s　noise　power,　and　the　UAV＇s　coverage　radius　is　used　to　determine　the　UAV＇s　altitude.　Finally,　the　UAVs　are　deployed　to　the　specified　location　with　the　goal　of　minimizing　deployment　delays.　The　simulation　results　show　that　the　proposed　algorithm　can　achieve　a　small　number　of　UAVs　and　a　low　deployment　delay.　©　2020　IEEE.