Abstract: Areas at a low exploration level have drawn increasing attention as future contributors to reserves growth.However,they are facing many geophysical challenges.Lufeng 22 subsag is such an area due to few drilled wells and insufficient geological data.In this case,it is difficult to build an accurate low-frequency inversion model using traditional logging data or stacking velocity.Moreover,affected by ghost reflections,low- and high-frequency waves in marine seismic data are suppressed.As a result,the bandwidth of seismic data is decreased,thus reducing the authenticity and accuracy of inversion results.To address these problems,this paper firstly obtained seismic data with broader bandwidth using broadband processing technology.Then,it built a low-frequency model for areas without well control using colored inversion combined with a high-precision velocity field obtained through tomographic imaging.Based on this,this paper predicted the distribution of source rocks and reservoirs using the extended elastic impedance inversion method.This technology was applied to the Lufeng 22 subsag,enabling the successful prediction of the distribution of high-quality source rocks and favorable reservoir areas.Thick layers of middle-deep lacustrine-facies source rocks as well as oil and gas have been discovered during the drilling of the first exploration well in the subsag,which started the exploration in the new area.This study indicates that this technology effectively improves the reliability of seismic inversion of middle-deep layers using broadband data and can well identify lithology utilizing low-frequency information,thus serving as an effective technology for the lithology prediction of areas at a low exploration level.