Membership renewals open for 2022 - Click here

Industry

Deep learning for seismic interpretation

Thursday, April 16, 2020
0900 (India)
1000 (India)

https://www.knowledgette.com/p/deep-learning-for-seismic-interpretation

 

Format: Virtual Webinar. 45 min. presentation followed by 15 min. Q&A

Please note that two sessions will be given at different dates listed below.

Session 1, Tuesday, April 14, 2020, 10 am to 11 am Beijing time Register Here

Session 2, Thursday, April 16, 2020, 9 am to 10 am IST time (India) Register Here

 

Abstract:

Seismic interpretation involves detecting and extracting structural information, stratigraphic features, and geobodies from seismic images. Although numerous automatic methods have been proposed, seismic interpretation today remains a highly time-consuming task which still requires significant human efforts. The conventional seismic interpretation methods or workflows are not automated or intelligent enough to efficiently or accurately interpret the rapidly increasing seismic data sets, which leaves significantly more data uninterpreted than interpreted.

We improve automatic seismic interpretation by using CNNs (convolutional neural networks) which recently have shown the best performance in detecting and extracting useful image features and objects. One main limitation of applying CNNs in seismic interpretation is the preparation of many training data sets and especially the corresponding geologic labels. Manually labeling geologic features in a seismic image is highly time-consuming and subjective, which often results in incompletely or inaccurately labeled training images. To solve this problem, we propose a workflow to automatically build diverse geologic models with geologically realistic features. Based on these models with known geologic information, we further automatically create numerous synthetic seismic images and the corresponding ground truth of geologic labels to train CNNs for geologic interpretation in field seismic images. Accurate interpretation results in multiple field seismic images show that the proposed workflow simulates realistic and generalized geologic models from which the CNNs effectively learn to recognize real geologic features in field images.

In this lecture, I would like the share you with our research experience on the following topics:

Automatic preparation of training data sets and labels;

CNN for fault detection, fault orientation estimation, and fault surface construction;

CNN for relative geologic time and seismic horizons;

CNN for seismic geobody tracking;

CNN-based multitask learning in seismic interpretation.

Deep learning for seismic interpretation

Tuesday, April 14, 2020
10 AM (Beijing)
11 AM (Beijing)

https://www.knowledgette.com/p/deep-learning-for-seismic-interpretation

 

Format: Virtual Webinar. 45 min. presentation followed by 15 min. Q&A

Please note that two sessions will be given at different dates listed below.

Session 1, Tuesday, April 14, 2020, 10 am to 11 am Beijing time Register Here

Session 2, Thursday, April 16, 2020, 9 am to 10 am IST time (India) Register Here

 

Abstract:

Seismic interpretation involves detecting and extracting structural information, stratigraphic features, and geobodies from seismic images. Although numerous automatic methods have been proposed, seismic interpretation today remains a highly time-consuming task which still requires significant human efforts. The conventional seismic interpretation methods or workflows are not automated or intelligent enough to efficiently or accurately interpret the rapidly increasing seismic data sets, which leaves significantly more data uninterpreted than interpreted.

We improve automatic seismic interpretation by using CNNs (convolutional neural networks) which recently have shown the best performance in detecting and extracting useful image features and objects. One main limitation of applying CNNs in seismic interpretation is the preparation of many training data sets and especially the corresponding geologic labels. Manually labeling geologic features in a seismic image is highly time-consuming and subjective, which often results in incompletely or inaccurately labeled training images. To solve this problem, we propose a workflow to automatically build diverse geologic models with geologically realistic features. Based on these models with known geologic information, we further automatically create numerous synthetic seismic images and the corresponding ground truth of geologic labels to train CNNs for geologic interpretation in field seismic images. Accurate interpretation results in multiple field seismic images show that the proposed workflow simulates realistic and generalized geologic models from which the CNNs effectively learn to recognize real geologic features in field images.

In this lecture, I would like the share you with our research experience on the following topics:

Automatic preparation of training data sets and labels;

CNN for fault detection, fault orientation estimation, and fault surface construction;

CNN for relative geologic time and seismic horizons;

CNN for seismic geobody tracking;

CNN-based multitask learning in seismic interpretation.

SA/NT AGM 2020

Tuesday, March 24, 2020
1800
1900

I would like to advise you of an update on next week's talk and general meeting. Due to the current circumstances, we will now be holding the short AGM virtually via Zoom, and unfortunately will be postponing Dr. Ian Moffat's talk for now. 

 

Date: Tuesday, 24 March

Time: 6:00pm

Location: via Zoom videoconference

 

It would be fantastic to see as many of you as possible on Zoom to round off the previous year, and we will be happy to take any feedback or suggestions for the upcoming year. Instructions for joining the Zoom meeting are below.

 

Please consider joining the committee if you have ideas for our society - we are always looking for new faces, and I promise you it's a low time commitment. I've attached the nomination form here, and please feel free to contact me if you have any questions about joining.

 

I look forward to seeing you on Zoom, and apologies for the changes. 

 

INSTRUCTIONS FOR JOINING ZOOM MEETING

 

Join online using this link: https://zoom.us/j/993761357

Meeting ID: 993 761 357

 

If you have not used Zoom before, we recommend joining a few minutes early so you can set up and ensure your video and sound is working properly.

Advances in Marine Seismic Data Acquisition Workshop

Wednesday, December 2, 2020
0800
1900

2-day workshop on 2 - 3 December 2020.

The workshop will bring together all aspects of marine seismic acquisition, highlighting advances in technologies and methodologies. It will focus on the science of data acquisition covering a broad range of topics, from advances in survey design to developments in seismic source, streamer and Ocean Bottom Seismic technology and its configurations, in order to address both exploration and development objectives. Workshop attendees will have a valuable opportunity to discuss among experts the future vision in domains such as machine learning, artificial intelligence and robotization.

Federal AGM

Tuesday, April 7, 2020
TBC
TBC

Please SAVE THE DATE for the evening of April 7th for the Federal AGM, which this year will be held in Adelaide. This will be in the Balcony Room at the Hotel Richmond, with speaker Prof. Graham Heinson, University of Adelaide as guest speaker. More details to follow.

ASEG WA Tech Night - Case Studies from Loupe & Helitem

Thursday, March 12, 2020
1730
1900

Please rsvp before 11th March 2020 to attend Joint Tech Nights organised by ASEG WA branch. The link for registration:

 

https://www.eventbrite.com.au/e/march-2020-joint-tech-night-tickets-97449440901

 

Direction to the venue: Celtic club

 

First Technical Talk: Case Studies from Loupe – New Technology in Portable TEM for Near-Surface Measurements

Presenter: Andrew Duncan, Loupe Geophysics

Abstract:

A portable, broadband TEM system, Loupe, has been developed for the purpose of measuring near-surface electrical conductivity. The system records continuously while walking using a three-component coil receiver mounted on an ergonomic backpack from signals generated by a small (660mm) diameter, multi-turn transmitter loop mounted on a similar backpack.

The Loupe system is designed primarily to measure electrical conductivity in the top 25 metres, previously the charter of frequency-domain EM systems.  Using modern electronics and software, combined with full time-series measurement, we have been able to overcome the interference and calibration issues around measuring near surface conductivity with a broad bandwidth time-domain system.  Sampling at around a half million samples per second and processed to produce a measurement of secondary TEM field every second, the Loupe system provides very high spatial resolution.  Data can be viewed as the operators walk, allowing survey redesign as necessary.

During 2019, trial surveys were conducted with Loupe in a number of near-surface applications including mineral exploration on surface and underground, geological / regolith mapping, study of groundwater around tailings storage facilities and the mapping of structural features in open-cut mines.  We see a wide application for Loupe in mapping seepage both from mine tailings and acid mine drainage.

Loupe has proved to be quite versatile, working in difficult terrain and areas with high electromagnetic interference such as mine sites and urban sites. Special challenges are presented when working on these sites due to power reticulation, vehicle movement and infrastructure.  We will give examples showing data collected in these environments.

During this presentation, we will summarise the Loupe system and show results from several recent surveys.

Bio:

Andrew Duncan is the Managing Director of ElectroMagnetic Imaging Technology Pty Ltd (EMIT), based in Perth, WA. EMIT, which recently celebrated 25 years in business, has developed technology including the SMARTem electrical methods receiver system, Maxwell EM software and the DigiAtlantis borehole magnetometer system for EM. Andrew has a background in the development of technology for electrical geophysics including the development of airborne EM systems and distributed systems for geophysical measurements. Recently, he founded Loupe Geophysics with Greg Street, in order to develop and commercialise a novel, portable TEM system called Loupe. Andrew has interests in signal processing, EM techniques for highly conductive targets and modelling of EM data.

 

 

Second Technical Talk: Helitem2: New Technology in Airborne TEM for Deep and Covered Targets with Western Australia Examples

Presenter: Adam Smiarowski, CGG

Abstract:

Exploration for targets at depth or targets obscured by conductive overburden have historically been a challenge with airborne EM methods. Although modern systems have been improved with greater primary transmitter moments, noise levels from receiver coil motion in the Earth’s ambient field has limited the detection of secondary target signals, especially at late times, and has limited the use of lower base frequencies. The new Helitem² system uses a patented low-noise receiver, a 50% duty cycle square pulse transmitter waveform, and low Tx base frequency, to achieve increased signal detectability for deep and covered targets.

Modeling and a series of demonstration surveys compared several helicopter-borne time-domain system configurations, including high-moment halfsine waveforms and low base frequency (15 Hz and 7.5 Hz) 50% duty cycle square waveforms. Using a thin-plate, modelling showed that a low base frequency square pulse will have a significantly larger response than a half sine pulse at standard 30 Hz base frequency for a wide range of target conductances. At early times, the sharper (quicker) turn off of the square wave results in much more high-frequency energy, and therefore better signal, for weakly conductive targets, and better near-surface resolution. At the other extreme, the response from very conductive targets is determined by the area under the transmitter curve, so the low frequency square waves with 16 and 33 ms widths produces more than twice the signal as the half sine.

Demonstration survey line profiles and decay curves over the target and background locations confirmed this modelling for a 400 m deep target and variable overburden. The combination of pulse width, power, and low noise enabled the Helitem2 system to be effective at low base frequencies, where very late time data is beneficial for detecting strong and deep targets. The survey demonstrated that the redesigned Rx suspension system was able to reduce coil motion noise, enabling acquisition of high quality low base frequency data useful for detection of deep targets to very late times. The wide-pulse waveform was effective at energizing a moderately-conductive target, increasing signal level by a factor of 2 above a 6 ms pulse. This will be even more beneficial when exploring for strong conductive targets at depth. Prior to this Rx re-design, noise levels at low base frequencies was too high, and the data was not useful for target detection.

Examples from Western Australian are provided, illustrating data improvements of Helitem2 operating at 12.5Hz, over a previous survey at 25Hz.

Bio:

Adam has been involved with electrical methods for environmental and exploration applications for 15 years. Adam completed an MSc in Geophysics at RMIT University and PhD in Physics and Geology at the University of Toronto.  He has been involved with airborne EM research, both in frequency and time-domain, with CGG MultiPhysics for the past 9 years.

 

 

 

ACT tech night - Case Studies from Loupe – New Tech in Portable TEM for Near-Surface Measurements

Wednesday, March 11, 2020
16:30
1900

Abstract:

A portable, broadband TEM system, Loupe, has been developed for the purpose of measuring the distribution of near-surface electrical conductivity.  The system records continuously using a three-component coil receiver mounted on an ergonomic backpack from signals generated from a small (660mm) diameter transmitter loop mounted on a similar backpack.

The Loupe system is designed to measure primarily in the top 25 metres of the ground, previously the charter of frequency-domain EM systems.  Using modern electronics and software we have been able to overcome many of the problems associated with the broad bandwidth needed to define near surface conductivity with a time domain system.  Sampling at around 500,000 samples per second and processed to produce a measurement of secondary field every second, the Loupe system provides very high spatial resolution.  Data can be viewed as the operators walk, allowing survey redesign as necessary.

During 2019, trial surveys have been conducted with Loupe in a number of near-surface applications including mineral exploration on surface and underground, geological / regolith mapping, study of groundwater around tailings storage facilities and the mapping of structural features in open-cut mines.  We see a wide application for Loupe in mapping seepage both from mine tailings and acid mine drainage.

The Loupe system has proved to be extremely versatile working in difficult terrain and areas with high electromagnetic interference such as mine sites and urban sites. Special challenges are presented when working underground due to power reticulation, vehicle movement, infrastructure and particularly steel mesh reinforcing.  We will give examples showing data collected in these challenging circumstances.

During this presentation, we will summarise the Loupe system and show results from several recent surveys.

Greg is the director of Loupe Geophysics based in WA. He has been working with geophysical systems in mining, groundwater and environmental applications for over 40 years

TAS tech talk - Insights into the 2018 eruption of Kilauea Volcano from ambient seismic noise and the application of seismic noise for imaging and monitoring in mines.

Thursday, February 27, 2020
1730
1900

On Thursday 27 February, Dr Gerrit Olivier will present Insights into the 2018 eruption of Kilauea Volcano from ambient seismic noise and the application of seismic noise for imaging and monitoring in mines.  This is a joint meeting with the Tasmanian branch of the GSA.  As ever, it will be preceded by drinks and nibbles at 5:30 pm in the Earth Science tea room (upstairs from the main entrance), with the main event following at 6 pm in the lecture theatre (back downstairs).  Gerrit’s abstract is also attached.

Gerrit obtained his PhD in Geophysics from the University Grenoble Alpes in France after completing an MSc and BSc in Theoretical Physics at the University of Stellenbosch in South Africa. He is currently a Director and heads up the Applied Geophysics group at the Institute of Mine Seismology – the world’s leading provider of seismic monitoring technologies for mines. He also serves as a Senior Adjunct Researcher at the University of Tasmania and and an Associate Researcher at University Grenoble Alpes in France. He has received awards from the American Geophysical Union and the Institution of Civil Engineers for his research in applying seismic noise interferometry to monitor and image underground mines and tailings dams.

Abstract:

Insights in to the 2018 eruption of Kilauea Volcano from ambient seismic noise and the
application of seismic noise for imaging and monitoring in mines

Ambient seismic noise interferometry is a method that enables seismologists to extract useful
information from faint background seismic noise. The method can be used to image the
subsurface with high resolution and/or monitor time-lapse changes in seismic velocity with
high accuracy in nearly any environment, without the need for active sources or earthquakes.
In this presentation, I will show how applying seismic noise interferometry has helped us
gain valuable insights in to 2018 eruption of Kilauea volcano. The 2018 Kilauea eruption was
a complex event that included deformation and eruption at the summit and along the East Rift
Zone. The eruption lasted three months and emitted around 800 million cubic meters of lava,
destroying more than 700 homes in the process. We used ambient seismic noise
interferometry to measure time-lapse changes in seismic velocity of the volcanic edifice prior
to the eruption. Our results show that 10 days before the eruption, there is a clear change in
the response of the seismic velocities to applied pressure. We also applied ambient seismic
noise tomography to image the state of the volcano after the eruption. The results of this
study will have implications for forecasting volcanic eruptions and our understanding of the
behaviour of volcanoes leading up to major eruptions. Finally, I show how the methods we
applied to Kilauea volcano are currently being used by the Institute of Mine Seismology to
monitor underground mines and tailings dams, while also being used as a cost effective and
environmentally friendly method for mineral exploration.

TAS tech talk - How the Canadian Metal Earth project strives to improve our knowledge of mineralization: an overview of multidisciplinary geophysical methods.

Tuesday, February 25, 2020
1300
1500

Just before he leaves us for the second time, Dr Esmaeil Eshaghi will tell us of How the Canadian Metal Earth project strives to improve our knowledge of mineralization: an overview of multidisciplinary geophysical methods.  Esi’s insider perspective of this huge effort getting underway to improve Canadian discovery rates will be at 1300 on Tuesday 25 February in the CODES conference room (UTas Sandy Bay).  An abstract is attached.  Members are invited to join the speaker and president for lunch at the Uni Staff Club afterwards.  Please RSVP to me or Matt Cracknell (tassecretary@aseg.org.au) if you’d like to come.

As many of you will well remember, Esi was in the airborne geophysics group at the Geological Survey of Iran before completing his PhD in regional potential field modelling at UTas a few years ago and proceeding to a postdoc stint at Laurentian University in Sudbury.  Currently he is with Thomson Aviation as a geophysicist.

Abstract:

How the Canadian Metal Earth project strives to improve our knowledge of mineralization:
an overview of multidisciplinary geophysical methods

Since 2005, there was a marked increase of costs to explore new economical mineral occurrences,
while the success of discovery of new deposits has diminished. Metal Earth, led by the Mineral
Exploration Research Centre at the Harquail School of Earth Sciences, Laurentian University,
Canada, is a seven-year $104 million applied research project with the main goal of improving the
understanding of mineral endowment in Precambrian greenstone belts. An essential part of this
project is to use geological observations and geophysical data to define crust to mantle scale
differences between transects that cross metal endowed and lesser endowed Archean greenstone
belts to define key mechanisms responsible for the genesis of base and precious metal deposits.
For this purpose, high resolution datasets are acquired and combined with existing data to create a
multidisciplinary database containing different geological and geophysical information (e.g.
geological observations, targeted mapping, seismic, gravity, magnetic survey, electromagnetic,
and petrophysics).
In this talk, I preview multidisciplinary geophysical data acquisition, processing and initial models
across different transects, both minerally enriched and less endowed, and provide an overview of
findings so far. It includes a brief discussion of seismic, magnetotelluric and potential field data
(i.e. gravity and magnetic) data and models along some transects as well as petrophysical
characterisations within Abitibi Greenstone Belt. The initial models indicate some distinct
differences between minerally prospective and less-prospective areas at depth. The findings are
promising in terms of identifying new components contributing to mineral deposition.

ASEG WA Tech Night - SEG 2020 Honorary Lecturer Lisa Gavin

Wednesday, February 12, 2020
17:30
19:00

 

 

Please rsvp before 11th February 2020 to attend 2020 1st Tech Nights organised by the ASEG WA branch. The link for registration:

 

https://www.eventbrite.com.au/e/aseg-wa-february-tech-night-2020-tickets-90110126815

 

Direction to the venue: Celtic club

The handout for the event is available here.

Pages