Logo Universitat Bremen
Die Inhalte dieser Seite sind leider nicht auf Deutsch verfügbar.
Seitenpfad:  

Workshop on machine learning approaches for automating geological correlation

Introduction

When?

30 November – 2 December 2022, 09:00 – 13:00

Where?

MARUM, seminar room 2070 & online

Why? 

This workshop explores different numerical techniques that allow for automating the correlation of multiple geologic depth- or time-series. The automatization of geologic correlation becomes increasingly important because several national governments now require oil-and-gas companies to make their wireline data publicly available after the expiration of a moratorium period. This data provides a largely untapped resource for scientific geologic investigation. The main advantage of the industrial logs is the right spatial density, potentially allowing a complete depiction of a basin’s geologic history. Their main disadvantage is poor age control: typically, only very rough constraints on the age of the geological layers penetrated by the industrial borehole exist. This disadvantage can be eliminated by correlating the industrial wireline logs with scientific wireline logs in the area. The geological age of scientific drill holes is typically accurately and precisely known, which means that correlation can serve to convey these geologic ages to the industrial wireline logs. To date, correlation of multiple well logs is still largely carried out manually. This is a time consuming and painstaking process, with repetitive micro-decisions to be made by the user.

What?

During this workshop, we will test three different numerical tools for automated stratigraphic correlation and apply them to industrial and scientific data from offshore Australia (case study). The objective is to identify the most effective approach and to test the extent to which automated stratigraphic correlation can help in generating a dense spatial network of correlated wireline logs.

For whom?

The workshop is open to everyone interested in stratigraphy, machine learning, dynamical time warping. Please sign up by sending an email to: [Bitte aktivieren Sie Javascript] or [Bitte aktivieren Sie Javascript].

Costs?

The workshop is free of charge thanks to the financial support of the MARUM Cluster of Excellence.

Workshop programm

30.11.2022

09:00 – 09:30

Introduction

David De Vleeschouwer

09:30 – 10:10

MyDTW – Principles and implementation

Heiko Pälike

10:10 – 10:30

Coffee Break

 

10:30 – 11:10

R packagedtw– Principles and implementation

Toni Giorgino

11:10 – 12:00

ChronoLog – Principles

Zoltan Sylvester

12:00 – 12:30

Setting up ChronoLog for beta testing

Zoltan Sylvester

13:00 – 14:00

Lunch at Haus am Walde (10 minutes walk)

All in-person participants

01.12.2022

09:00 – 10:30

Application of MyDTW and dtw (R-package) to Australian data

D. De Vleeschouwer & H. Pälike & Toni Giorgino

10:30 – 11:00

Coffee Break

 

11:00 – 11:30

ERC Starting grant 'Mind the Gap'

 Emilia Jarochowska

11:30 – 13:00

Application of ChronoLog to Australian data

D. De Vleeschouwer & Z. Sylvester

02.12.2022

09:00 – 10:30

Break-out in three groups (ChronoLog / MyDTW / dtw) to identify strengths and weaknesses of individual approaches.

10:30 – 11:00

Coffee break

11:00 – 12:30

Each group presents the advantages/disadvantages of three different tested approaches.

Group discussion on which tool is most useful.