Missing sonic log prediction using convolutional long short-term memory
Nam H. Pham, Xinming Wu · 2019
We propose a method to estimate missing sonic logs by using a bidirectional convolutional long short-term memory (bidirectional ConvLSTM) cascaded with a dropout layer and fully connected neural networks (FCNNs). We train the model on 177 wells from mature areas of the UK continental shelf (UKCS). We test the trained model on one blind well from UKCS, two wells from the Volve field in the Norwegian continental shelf (NCS), and one well from the Penobscot field in the Scotian shelf offshore Canada. The method takes into account the rock properties trend and the local shape of logs, and produces accurate prediction of sonic logs from gamma-ray and density logs with an addition of uncertainty estimations and without the need for applying rock-physics model intervally. Presentation Date: Wednesday, September 18, 2019 Session Start Time: 1:50 PM Presentation Time: 1:50 PM Location: 221D Presentation Type: Oral