Characterizing stratigraphic traps using improved waveform classification seismic facies analysis: an example from central Saudi Arabia
The stratigraphic traps in Unayzah formation play an important role in central Saudi Arabia where Aeolian sandstone with good reservoir quality is laterally sealed by up-dip playa siltstone with low porosity and low permeability due to the facies change. The recognition of facies and lithology therefore is a good way to characterize the stratigraphic traps. Seismic waveforms are employed using a supervised artificial neural network (ANN) to classify and identify the seismic facies since they carry multiple information such as amplitude, frequency and phase. Supervised ANNs are proven to be a successful classification technique which is implemented by the multi-layer perceptron (MLP) network with two layers. The network is trained with the facies information in the first layer and is to define and classify the input into several subclasses. When the training is completed, the sub-classes are combined and associated with the targeted facies in the second layer. And then the trained network is applied to the seismic data to predict the facies between well locations. The seismic reflection patterns for recognizing facies include configuration, amplitude, continuity, frequency, and interval velocity. Important lateral variations in reflectivity and reflector continuity provide information which is absent from the individual seismic waveforms and this is blended into the waveform classification map in order to improve the facies recognition.