subsurfaceAI
Neural Network Inversion of λρ and μρ from Post-Stack Seismic Attributes
Published:
Author: subsurfaceAI

Webinar

Neural Network Inversion of λρ and μρ from Post-Stack Seismic Attributes

Click here to register
Date: Wednesday, October 14th, 2020
Time: 9:00 AM-10:00 AM (MST)

 

Presented By:

Dr. Rongfeng Zhang

Senior Geoscientist
Geomodeling Technology Corp

 

Please join us for a webinar about Neural network inversion of λρ and μρ from post-stack seismic attributes. The workflows are available in AttributeStudio since version 8.4.
λρ and μρ are the results of pre-stack inversion and they are very popular for seismic interpretation. In areas that AVO effects present, λρ and μρ are effective parameters to discriminate lithofacies and fluid types. However, λρ and μρ are not always available in practice for various reasons such as cost, time, data quality, skills etc. We present a new workflow of λρ and μρ inversion from post-stack attributes and well logs through a neural network approach. The neural network inversion workflow provides an alternative way to obtain λρ and μρ attribute volumes in areas where pre-stack inversion results are not available or reliable. Pre-stack inversion gives usP-impedance, S-impedance and density, from which λρ and μρ can be calculated and used for interpreting reservoir facies and fluid distribution. However, λρand μρ are generally not derived from post-stack inversion, which only generates P-impedance. We propose a neural network inversion workflow to predict λρ and μρ from post-stack attributes and well logs.

The workflow consists of five steps:

  1. Calculate λρ and μρ logs from wells that have shear sonic and density logs.
  2. Build a neural network between λρ, μρ logs and other conventional logs such as gamma, compressionalsonic, etc. The network is used to predict λρ and μρ logs in wells that do not have the shear sonic log.
  3. Upscale λρ and μρ logs to seismic resolution
  4. Build another neuralnetwork between upscaled λρ and μρ logs and seismic attributes along the wellpath.
  5. Apply the neural networkbuilt in step 4 to create a seismic stratigraphic grid.

The steps will be demoed in the webinar using AttributeStudio 8.4.1.

Click here to register