1 of 28

How open tech can help explore�the digital subsurface

matt@agilegeoscience.com

ageo.co/digex19

2 of 28

ageo.co/digex19

Python for geoscientists, Calgary, 2016

Agile*

3 of 28

ageo.co/digex19

Hackathons

Training

18

3

2

2

3

classes

4 of 28

ageo.co/digex19

©️ Cameron Watt, YouTube

Giant python, Limpopo, South Africa

scientific Python speedbump

I am an ectothermal squamate, hence I am basking in Sol's infrared radiation, ca 200 THz

5 of 28

Karpatne et al. (2017)

Karpatne, Anuj, et al. (2017). Machine Learning for the Geosciences: Challenges and Opportunities.

arXiv:1711.04708v1. (In review for IEEE Transactions on Knowledge and Data Engineering.)

  • Objects with amorphous boundaries.
  • Spatiotemporal structure.
  • High dimensionality.
  • Heterogeneity in space and time.
  • Rare phenomena.
  • Multiresolution data.
  • Noise, incompleteness and uncertainty in data.
  • Small sample size.
  • Paucity of ground truth.

6 of 28

Open source today

ageo.co/digex19

7 of 28

Adapted from Jake VanderPlas 2015

2.7M

7.2M

13M

19k

5k

3k

8 of 28

github.com/agile-geoscience

welly, striplog, bruges

9 of 28

devitoproject.org

TLE tutorials, 2018

10 of 28

SimPEG

11 of 28

GEOLOGY

18

GEOPHYSICS

55

https://en.wikipedia.org/wiki/List_of_free_geology_software

https://en.wikipedia.org/wiki/Comparison_of_free_geophysics_software

12 of 28

Open source vs closed source

Robust support and docs

Training available

Clear license terms

Familiar to users

Patchy support and docs

Ad hoc training

License zoo

Sounds weird

Hard to innovate

Expensive to scale

Inelastic

Black box

Flexible tooling

Lower cost of ownership

Better interoperability

Transparent codebase

​

13 of 28

Open source vs closed source

Robust support and docs

Training available

Clear license terms

Familiar to users

Patchy support and docs

Ad hoc training

License zoo

Sounds weird

Hard to innovate

Expensive to scale

Inelastic

Black box

Flexible tooling

Lower cost of ownership

Better interoperability

Transparent codebase

​

  • Free and open
  • Easy installation
  • Multi-platform
  • Compatible licenses
  • Uniform interface
  • Flexible I/O
  • Quality assurance
  • Support
  • Documentation
  • Training

​

Best of both

14 of 28

Key features of the project

Subsurface science and engineering.

Python and C. Friendly towards Julia, R, JS.

​

Oil & gas, mines, geothermal, water, research.

Support, training, and consulting. Grants? Consortia?

15 of 28

Technologies

Years to decades, $M

  • Research programs
  • Consortiums
  • Start-ups

Months to years, $k

  • Unconferences
  • Hackathons
  • Sprints

Smart people, collaboration, deep skills, perseverance, money

16 of 28

Software Underground is a Slack group for geoscientists who use computers in their work or research.

Go to softwareunderground.org to join in!

17 of 28

ageo.co/digex19

Unsolved problems unsession, Calgary, 2013

© Agile, licensed CC-BY

18 of 28

ageo.co/digex19

1st Geophysics Hackathon, Houston, 2013

© Agile, licensed CC-BY

19 of 28

ageo.co/digex19

Geophysics Sprint, Santa Ana, California, October 2018

20 of 28

TRANSFORM

21 of 28

TRANSFORM

5 to 11 May Rouen, France

22 of 28

We are launching a new era in scientific computing for the subsurface.

At its core is a new open-source software stack, created, maintained, and implemented by a community of scientists and organizations passionate about its potential.

Purpose

23 of 28

Day 1: Warm up�Share skills and tools. Plan the week.

Days 2 & 3: Work out�Share our research, discuss projects, and collaborate on roadmaps.

Days 4 & 5: Sprint�Contribute code and documentation to the open subsurface stack.

Plan

24 of 28

We will collaboratively...

Explore the new landscape: digital-first culture, open data, open source, and lean, agile project definition.

Build skills, strategy, and code around a new open geoscience stack to help power our journey through this landscape.

Activities

25 of 28

World map of the existing open source tools for subsurface.

Gap map: where are the holes? Which are the priorities?

Road map for helping navigate the next steps. Iterate as we go.

Start on the core: we can get a long way in 2 days.

Outcomes

26 of 28

Château de Rosay, near Rouen, northern France.

This is not your average conference centre.

This is a place for doing extraordinary things.

Location

27 of 28

5–11 May 2019.

Participants will arrive on Sunday 5 May and leave the following Saturday.

The main event runs from Monday 6 May to Friday 10 May.

Each day will be packed.

Timing

28 of 28

ageo.co/digex19

ageo.co/transform

matt@agilescientific.com