Abstract: Based on two phases HJ remote sensing imagery which before and after flood disaster，the recognition technology of farmland damaged by floods caused by rainstorm in Beijing Fangshan district at‘7.21’was studied. First using the method of supervised classification，distribution map of farmland for before and after the flood disaster was acquired，then put the two periods distribution map of farmland overlay，information of farmland changing was extracted，distribution area of the affected fields was added up.Th result showed the area of farmland damaged by floods caused by rainstorm in Beijing Fangshan district at‘7.21’was 4 103.82 hm2，mainly distributed around the basin of the Juma River，the stone river and Jiakuo river.The operational feasibility of identification to farmland damaged by floods combining HJ remote sensing imagery with supervised classification method was verified，which could provide certain theoretical basis for dynamic monitoring issues related.