基于基追踪去噪的地震反射率反演通过应用l1范数正则化产生稀疏解，但由于正则化项的存在，基追踪去噪获得的解与真实解之间总是存在偏差。为了减小这种偏差，本文通过在基追踪去噪地震反射率反演中引入“残差回加”的策略，提高了解的准确性。“残差回加”策略通过使用一个适中的权衡因子，迭代执行快速迭代软阈值算法（fast iterative soft thresholding algorithm，FISTA）来求解基追踪去噪的目标函数，而每次求解的残差将会不断加回至原始输入。与使用FISTA单次求解相比，结合“残差回加”策略的FISTA求解得到的解更接近真实解。本文将该方法应用到我国南海珠江口盆地珠一坳陷惠西南古近系的文昌组储层，通过高精度的波阻抗反演结果对钙质砂岩和滩坝砂岩进行了有效划分。

Seismic reflectivity inversion based on basis pursuit denoising produces sparse solutions through the application of l1 norm regularization. However, due to the presence of the regularization term, there is always a deviation between the solution obtained by basis pursuit denoising and the true solution. To reduce this deviation, this paper introduces the "residual addition" strategy in the basis pursuit denoising seismic reflectivity inversion, thereby improving the accuracy of the solution. The "residual addition" strategy uses a moderate trade-off factor to iteratively perform the Fast Iterative Soft Thresholding Algorithm (FISTA) to solve the objective function of basis pursuit denoising, and the residuals obtained each time are continuously added back to the original input. Compared with a single solution using FISTA, the solution obtained by FISTA combined with the "residual addition" strategy is closer to the true solution. In this paper, this method is applied to the Wen Chang Formation reservoir in the Huizhou Southwest Tertiary System of the Pearl River Mouth Basin in the South China Sea, and the calcareous sandstone and beach-dam sandstone are effectively divided through high-precision impedance inversion results.