A method for filtering noise from seismic data is disclosed. The method includes: receiving at least one seismic data set; computing a value of at least one subset of a plurality of subsets of the at least one seismic data set; computing quartiles and an upper moustache (M) for the plurality of subsets; comparing the upper moustache (M) to the value of the at least one subset; and one of discarding and retaining the at least one subset as filtered seismic data. A system and computer program product for filtering noise from seismic data is also disclosed.