基于GPU并行加速的逆时偏移成像方法

GPU-accelerated reverse time migration and its application

现今逆时偏移的应用面临海量数据存储及计算量巨大的问题。本文应用震源重构方法解决了逆时偏移互相关成像条件需要的海量存储问题,引入GPU Kernel解决逆时偏移计算效率低的问题。在此基础上,本文建立了一套基于GPU加速的叠前逆时偏移成像方法,并成功地应用于四川盆地A复杂区块。

There are many difficulties in reverse time migration applications for nowadays seismic exploration,storage and computations of the mass data may be the top ones.In this paper we use source reconstruction method to reduce mass storage brought by cross-correlation imaging condition, and introduce GPU kernel to improve computation efficiency.And based on these,we build a GPU accelerated reverse time migration method,which successfully applied in Block A,a complex area in Sichuan Basin.