针对塔中地区碳酸盐岩储层类型多、非均质性强的特点，利用地震波形变化特征，半定量地识别和预测了优质储层的分布。文章论述了该技术的基本原理、实现过程以及关键参数选择的原则。通过井-震标定建立了储层段地震响应的模式，分析了不同品质储层的地震波形特征；利用自组织神经网络逐道进行了目的层段地震道波形聚类；结合钻井标定最终实现了碳酸盐岩储层的半定量化评价。根据预测结果部署了2 口井，钻探结果与预测结果吻合较好。

The carbonate reservoir is characterized by various types and strong inhomogeneity. Seismic waveform variation characteristic is used to semi-quantitatively identify and predict the distribution of high-quality reservoir. The basic principle, realization process and the key parameters of this technology are analyzed. The seismic response pattern is established in reservoir zone by well-seismic calibration, and the seismic waveform characteristics of different reservoirs are analyzed. Seismic trace waveform of target zone is classified by means of self-organizing neural network. The semi-quantitatively evaluation of carbonate reservoir is finally carried out. According to the prediction, two wells have been drilled, and the drilling result is consistent with the prediction result.