Abstract

Summary The clastic reservoir in Tahe oil field, west China, is characterized by thin sandstone thickness and low seismic resolution, which brings difficulty to sand description. To solve this problem, we present a seismic resolution enhancement method and a reservoir prediction method to detect thin bed reservoir. We improve seismic resolution based on the matching pursuit (MP) algorithm. The original seismic data is decomposed into superposition of many best-matched wavelets using the MP algorithm. Then we increase the amplitudes of the best-matched wavelets at high frequencies with high signal-noise ratio, while maintaining low-frequency components. Summing these wavelets can reconstruct a new seismic data whose resolution is enhanced. Based on the data with improved resolution, we use a stochastic simulation method to predict the reservoir. A continuous time Markov chain is introduced as vertical prior distributions, which is used to perform stochastic simulation of pseudo wells. Then we calculate the synthetic seismic seismograms for the pseudo-wells, and use a optimal similarity coefficient to find the best-matched pseudo wells to the actual seismic trace. These best-matched pesudo-wells are used to predict lithofacies, porosity and shale content. The above methods are applied to Tahe oil field, and show good and reliable performance.