Abstract:The polynomial zero-offset trace fitting is superior to horizontal stacking in many aspects,but inferior to the latter in suppressing multiple waves.By analysing fitting errors,the rejection fitting method can reject some raw seismic data to produce a satisfactory result.When there are strong multiple waves in near traces,the rejection may incur mistakes;this is because both fitting method and stacking method are based on the least square method,which makes the fitting result be on the side of big deviation data.Thus L1 norm minimization method is introduced to accomplish polynomial zero-offset data fitting because it is stabler than least square method.The theoretical and real data results show that L1 norm minimization method retains the advantages of fitting mehtods and attenuates multiple waves desirably.