Skip to content

mjhay/neem_sonic_model

master
Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
This repo has Python 2.7 code for the anisotropic ice sonic velocity model.

Python package dependencies:
tensorflow
numpy
scipy
matplotlib




 Example plot output of inferred fabric eigenvalues with depth is given by 'eigenvalues.pdf'.

Required input data:
Thin-section eigenvalues (n_ts x 3 matrix)
Thin-section depth (n_tsx1 matrix, not vector)
Observed p-wave velocities (n_vel x 1 matrix) or observed P,Sv,S (n_vel x 3 matrix)
Velocity depths (n_vel x 1 matrix)

Example usage with synthetic data is in 'synth_sess.py'. This repository contains synthetic velocity and eigenvalue data in the corresponding CSV files. The file 'synth_eig_true.csv' contains "true" synthetic eigenvalues that the model attempts to recover. The file 'synth_eigs_ts.csv' contains the eigenvalues sampled at 100 thin-section depths, but with added noise, to simulate sampling error in real thin sections. The file 'vels.csv' contrains "true" synthetic velocities with depth, generated from the forward velocity model from the "true" eigenvalues. Low spatial-frequency noise is added to this to produce "vels_with_corruption.csv," which is the model input.

The model is fit by first finding a smooth fit to thin section eigenvalues using Gaussian process regression. Then, velocities are modeled to the smooth fit. A smooth estimate of velocity mismatch between measured and modeled (from the thin sections) P-wave data is then found, and used to produce corrected velocities under the assumption that the eigenvalue-derived velocities are unbiased. The remaining mismatch is then regressed out by inverting the velocity model to produce corrected eigenvalues.

About

code for neem sonic data assimilation

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages