Seam Seismic Detection And Comprehensive Forecast Of The Hole Reservoir

Fracture is the main reservoir and migrating channel of petroleum, but it is difficult to detect fracture strip by a single subject or method. Especially in carbonate formation, owing to the intricate interstice, the variational size, shape of cave, and the different character of filling, it is almost impossible to get all-around cognition by a single way. Under the circumstance, comprehensively prediction by multi-subject, multi-method and multi-parameter is the way which can reduce multiplicity to predict caves in carbonate formation.The paper discusses the probability of fracture detection by seismic method firstly, and then analyses seismic reflected wave character of fracture-cave stratum. At last, the paper validate that dynamical characteristic of seismic wave is more sensitive to fracture than kinetic characteristic. Based on that theory, several seismic attributes, seismic impedance inversion, multi- scale edge detection and edge-preserving, and coherence algorithms are adopted to detect fracture in AA zone(clastic reservoir) and BB zone(carbonate reservoir). Analyzing the effect of different methods, we summarize the differences between clastic reservoir and carbonate reservoir detecting. That is, gradient class attributes are adapted to carbonate reservoir and non- gradient class attributes are adapted to clastic reservoir.Three kinds of methods are advanced in this paper. First, the paper describes artificial reservoir forecasting, which is to regulate and remedy seismic detecting result based on profound understanding of geologic data, log data and exploitation data, and then gets seismic forecasting map which accords with geologic rule and very consistent with log data. Apart from the method above, integrative fracture detection by confidence limit analysis and integrative fracture detection by Neural Network are also advanced as well. The former is linear method and the latter is linear method.The paper has the following innovation. First, developing and improving the supporting method of multi- scale edge detection. That is, multi-scale edge-preserving. Secondly, comparing effects of different methods in carbonate reservoir and clastic reservoir, methods adapted to different kind of reservoir are advanced. Thirdly, integrative fracture detection by confidence limit analysis, which makes use of correlation between well log fracture interpretation and seismic fracture detection to weight detection from different methods in order to improve the detection reliabilities, is advanced to forecast reservoir. In this way, excellent methods adapted to objective are selected. In addition, facing the problem that structure of Neural Network lacks of reasonable guidance, the paper adopts genetic algorithms to optimize net structure first, and then uses Neural Network to predict reservoir.Detection and forecast of reservoir is an important direction in geophysical development. The paper lists a clear clue to detect and forecast carbonate reservoir and clastic reservoir. However, some detail (such as selection of sample, fusion of qualitative data and quantitative data) need to be deep research.