Innovations of the decade to come
Matt HallOn Monday I posted what I think were the major advances in exploration and reservoir geoscience in the last decade. I wanted to follow up with a look at what might happen next.
As oil and gas become harder to find and develop safely, responsibly, and economically, our tools and data will only continue to improve. In particular, acceptable oil sands and shale gas recovery efficiency demand new ideas and new methods. I hope the next decade will see us making progress in some of these areas, some of them long-lived problems:
- Pre-stack interpretation is still, in my view, niche. Only a few interpreters have the patience and software tolerance to get into finding, loading, QC'ing, and analyzing gathers or even partial stacks. These data can show how amplitude varies with offset, the average distance between the source and the receiver (sometimes expressed as a reflection angle instead, with large angles being equivalent to long offsets). This can help with lithology and especially fluid prediction.
Image links to TIBCO's websiteMulti-variate analysis and prediction will be paramount as interpreters juggle more and richer data sets. As stacks, limited-angle stacks, inversions, volume attributes, and other volumes mont up, it will get harder to make decisions consistent with all the data. Sound probabilistic methods are needed, but also the ability to handle dozens of variables, each with its own error and uncertainties. GIS will play a big role here, as well as powerful analytics tools like TIBCO Spotfire 2011 (shown here). - GIS will get more integrated into our workflows. A lot of the silliness in PowerPoint is really quick GIS hacks. One upshot of this is that somewhat marginal data today, like gravity and magnetics, will get easier to integrate. It will also mean people make and show more beautiful maps, which can only be a good thing!
- Error and uncertainty are such monstrous problems that I am going to mention them again. Since every measurement we make in the subsurface is subject to error and uncertainty, we tend to ignore it (and sometimes abuse it!), handling it later with some sort of gross fudge factor. For example, instead of mapping different possible structures, or trying different velocity models, we simply say, 'Oh, the area could be wrong by 25%', so we arbitrarily run our volumetrics with that, and see if it 'looks right'. Maybe geoscientists think this is OK, but I don't think the VP of Exploration would agree with you!
- Effortlessly stepping across scale will be important if we are to unravel problems like fracturing, heterogeneity, stratigraphic prediction, and time-series phenomena (like production effects). The ability to see the whole picture, zoom in to see the details, and analyse these scales in different ways, is a challenge. We must bridge the Integration Gap.
- Intuitive and fast display manipulation tools. The mouse and keyboard are hopeless interpretation instruments and should be killed as soon as possible. Multi-touch, gestures, tablets: all the pieces are there. Some people are making progress already, for example check out this wonderful teaser from Wacom and open-source OpendTect software.
- Microsoft PowerPoint is a scourge on our productivity as scientists, in my opinion, at least as a documentation tool. Maybe, maybe, it has a rôle to play as a presentation tool. I see people drawing lines on maps, faking scale bars, stretching maps, showing weird-looking equations, and all sorts of other horrors. Try doing a presentation without slides one day; instead turn up with a handout, some maps, some rocks, and a smile, and I guarantee that people will be pleasantly surprised, thank you heartily, and tell everyone how cool your story was. Okay, enough ranting... I will post about this one day.
- Plotting will get easier. So much grief. So many wasted hours.
There should be some resemblance to the unsolved problems in exploration geoscience I posted a few weeks ago. Please let me know what I've missed!
PowerPoint is a trademark of Microsoft Corporation.Spotfire is a trademark of TIBCO. The screenshot of TIBOC Cpotfire 2011 is copyright of TIBCO and used here with permission, for which I am grateful. OpendTect is a trademark of dGB Earth Sciences. The YouTube clip is embedded and copyright of Wacom. I am not connected in any way to any of these companies.
Reader Comments (6)
I love seeing the new tools to intrepret. I alwasy found "drawing" with a mouse clumsy and it nevers just works out exactly how you wanted it to.
Have you seen if there have been any advances on haptic devices - so we can actually touch and feel the data?
The only thing I sort of disagree with you is on plotting and I assume you mean paper. It has always been the bain of my existence and I figure I don't want to even bother unless I have to. The new "paper" should be screens that can display like paper but aren't - like the little glass plates you see in all the sci fi flicks these days.
My two bits
Stay dry
@Doris: Thanks for the great comment. I think haptic is over, at least for now. I like the idea of feeling data textures, maybe hearing too, but I think vision is much higher bandwidth and may be more intuitive too. Certainly the first haptic fad is over!
And plotting, yeah, I know... I felt bad about writing it because I used to be quite vocal about paper-free... but then I worked in Exploration on 2D data again and even dual 24" inch screens just don't work anymore. Drawing on paper is so easy and comfortable... but then the result is frustratingly static and un-digital. This is one of the reasons the re-emergence of tablets and styluses is interesting. Multi-touch + ultra-high-res = win? I hope.
Cheers!
That youtube vid is a funny one: Matt is interested in it because of the cool Wacom tablet and drawing on the seismic: I'm interested in it because of the automated sequence strat. [Full disclosure: Renee Bourque is my co-author on a recent AAPG conference paper.
Automated seismic picking is my bet for this-decade innovation: we still don't use it enough (since in my view it's not really mature and that user-friendly). Let the computer do the grunt work and give the geos time to do the thinking - especially about the possible uncertainties which, as Matt points out in his post, we don't do enough of. See Hoyes & Cheret (2011) in this month's TLE for A review of “global” interpretation methods for automated 3D horizon picking.
@Richie: Thank you for the comment... I agree, automated is one to watch. If someone can balance fast, intuitive autonomy with meaningful, intuitive judgement, then I think it will overcome the general skepticism and/or fear many interpreters and chiefs have for it today. The three efforts I've seen (Schlumberger's Ant Tracking, Landmark's Atomic Meshing, and OpendTect's SSIS) all seemed to require either very good data or a lot of fiddling, or both. I always feel like these tools try to eliminate uncertainty (by smoothing the data, or by picking a best fit) instead of trying to represent it --- for me that would be a big step.
I took the liberty of adding links to your comment. Thanks for the references!
In typical seismic interpretation projects we end up with a couple of mapped horizons and an over-simplified geologic model. In the process we reduce terra-bytes of useful seismic information to a mere mega-bytes of interpreted data. With automated horizon tracking tools it is possible to extract much more geology from the seismic data. With a dense set of auto-tracked horizons we can see more geology by slicing through the data in a geologically sound way; we can use these horizons to guide well correlations; to flatten data (Wheeler transformations); to obtain a better insight in the depositional environment; to interpret systems tracts; and to increase our chance of finding stratigraphic traps. The biggest benefit of having a dense set of horizons is however, in geologic model building and related low frequency model building to improve seismic inversions. By interpolating well data along the dense set of horizons we can for the first time ever generate detailed geologic models that are consistent with seismic measurements. Better models will lead to improved reservoir management and better economic decisions.
The first tools to auto-track hundreds of horizons are now coming to the market. Different algorithms and work flows are emerging. Which one will come out on top remains to be seen but what is certain, at least in my mind, is that this is game-changing technology. Therefore, I agree with Richie and Matt to watch out for automated tracking.
I think integration of different types of data will continue to be compulsory (and challenging), particularly as we move between time and depth data sets, different projections, or mix temporal (e.g. production histories, time lapse seismic, etc.) and static data (e.g. openhole logs, 2D reconnaisance seismic grids, etc.); let alone the (literally) hundreds of attributes we can now generate on 3D seismic data. The more data we integrate, the more robust interpretation(s) we yield. Matt's comment about "appreciating paper" when he worked on a large 2D exploration seismic grid resonates with me as well....... we somehow haven't developed technology as good as the human eye (combined with paper) for zooming quickly between regional and local scale features; at the same time doing all that incredible pattern recognition that our eye-brain does. In response to Doris' comments; another cool way to look at seismic data is the Steve Lynch SeisScape/terrain presntation style ( http://www.cseg.ca/publications/recorder/2000/10oct/oct00-ancient-evenings.pdf )...