# Cross-well machine learning prediction of sonic logs in Newfoundland and Labrador

## TL;DR

This paper presents a machine learning approach to predict sonic logs from non-sonic data in offshore wells in Newfoundland and Labrador, reducing costs and improving field planning.

## Contribution

The study introduces a leakage-free, features-only strategy for cross-well sonic log prediction using tree-based models and depth-aware feature engineering.

## Key findings

- Tuned XGBoost achieved high performance with R² = 0.895 in cross-well prediction.
- Random Forest performed competitively, while BiLSTM underperformed on the dataset.
- Leakage control and feature engineering were key to achieving strong results.

## Abstract

Predicting compressional slowness (DTCO) from non-sonic logs can reduce acquisition cost, fill data gaps, and support field planning. We evaluate blind cross-well DTCO prediction on two offshore Newfoundland & Labrador wells using a strictly leakage-free, features-only strategy: causal lag windows are built from past non-sonic logs and all sonic/sonic-derived channels are excluded. The pipeline includes deterministic depth conditioning, relative-depth features, multi-scale depth derivatives, rank-aggregated feature selection, and time-aware validation on the training well. We compare three model families: Random Forest (RF), Extreme Gradient Boosting (XGBoost), and a BiLSTM. In this setting, tuned XGBoost with the top 20 predictors and a 10-sample lag attains blind cross-well performance of \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$R^2=0.895$$\end{document}, MAE\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$=11.38~\mu \mathrm {s/m}$$\end{document}, RMSE\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$=15.12~\mu \mathrm {s/m}$$\end{document} when trained on Well 1 and tested on Well 2; the reverse direction is lower, indicating inter-well distribution shift. RF performs competitively in several configurations, whereas BiLSTM underperforms on these data. Overall, rigorous leakage control, depth-aware feature engineering, and principled feature selection are key drivers of performance, and tree-based ensembles provide strong, data-efficient baselines for cross-well pseudo-sonic prediction.