Seeing the unseeable

Improved ways to gather and interpret geophysical data were highlighted at OTC. Bruce Nichols reports.

Rocky Roden, consulting geophysicist for Geophysical Insights, showed an OTC session how to combine principal component analysis (PCA) with self-organizing maps (SOM) to derive more useful information from multi-attribute data in searching for oil and gas. Image from Rocky Roden.

The oil industry’s relentless drive to see the unseeable deep in the Earth was on full display at the 2015 Offshore Technology Conference, with technical session presenters offering new ways to gather data and interpret it to reduce risk in exploration and development.

The conference featured a full day of sessions on improved tools and techniques, a morning devoted to advanced and integrated geophysical interpretation and an afternoon discussing the pitfalls and promises of broadband seismic.

In the morning session, CEO Michael Zhdanov and chief geophysicist Masashi Endo of TechnoImaging and Johan Mattsson, research and development manager at Petroleum Geo-Services (PGS), promoted towed-streamer electromagnetic data-gathering by showing a way to accomplish 3D inversion of the data – the step necessary to create a usable model (OTC-25751-MS).

“We apply the concept of moving sensitivity domain, originally developed for airborne EM surveys, to the interpretation of marine EM survey data. This makes it possible to invert the entire towed-streamer EM surveys with no approximations into high-resolution 3D geoelectrical seabottom models,” they said.

In the afternoon, geoscientists Peter Mesdag and M.D. Schakel of CGG Geosoftware touted simultaneous AVO (amplitude versus offset) inversion of broadband seismic data and careful use of quality controls derived from well logs to improve reservoir characterization with wider spectrum data (OTC-25668).

“Of critical importance to inversion processes is wavelet estimation,” they said. “Broadband, with its extended low frequency range, demands longer wavelets and therefore longer wavelet design windows. Innovative techniques are necessary to estimate the low frequency components of these wavelets while retaining relevance to the reservoir in the high frequency band.”

Towed-streamer eases electromagnetic sensing

Zhdanov told OE that oil companies are slowly adopting the use of controlled-source electromagnetic (CSEM) data to characterize hydrocarbon targets, but the approach is still not standard, even though EM is better than seismic for discriminating between hydrocarbon-filled and water-filled reservoirs.

“The range of resistivity variation within the reservoir rocks is much greater than the range of variation of seismic velocities,” Zhdanov said. “As a result, EM data can distinguish between a hydrocarbon-filled reservoir and a saline-water filled reservoir based on differing resistivities, while the seismic data have difficulties in solving this problem.”

Until recently, node CSEM systems with fixed seabottom receivers have been used in the marine environment. In that situation, only the source is mobile. PGS’ towed-streamer system is more efficient and cheaper because both the source and the receiver are attached to the vessels, and the receiver nodes don’t have to be deployed on the seabottom, Zhdanov said.

The moving platform geometry of the towed-streamer EM system enables EM data to be acquired over very large areas in both frontier and mature basins for higher production rates and lower costs compared to conventional node CSEM methods. A towed-stream system is also capable of simultaneous seismic and electromagnetic data acquisition, Zhdanov said.

“The resistivity data, in combination with seismic and well log data, offer the possibility to determine the hydrocarbon content of the reservoir with increased probability of accuracy, and there are several published case studies that show the improved success rate of exploration using EM methods,” Zhdanov said.

Five steps boost inversion accuracy

Mesdag, CGG’s technical product manager for deterministic reservoir characterization, discussed using seismic data from a variable depth streamer to characterize offshore reservoirs for development drilling and then to monitor their changes as production proceeds.

Broadband seismic is adding value to inversion technology, but quality controls (QC) on the data and interpretation of the data remain key, he said.

“You still need well control. That has not been completely banished by broadband inversion,” Mesdag told OE.

Secondly, great care is necessary in carrying out wavelet estimation because wavelets are so long in the low-frequency range, much longer than well logs, that logs are insufficient to provide quality control over wavelet estimation.

“There’s a work flow we’ve designed and implemented,” Mesdag said. “We can have accurate wavelets. It involves a couple of steps. You can’t use an off-the-shelf program to get that. The steps are outlined in our paper.”

Mesdag listed the five steps in the CGG approach: Data loading and QC, conventional well-tie, conventional wavelet estimation, broadband wavelet estimation and broadband well-tie and inversion.

New math aids quantitative interpretation

Senior research geoscientist Ehsan Zabihi Naeini and senior geoscientist Ben Hardy of Ikon Science offered a way to extract wavelets from broadband seismic to produce more accurate seismic inversions to support interpretive estimates of reservoir content and volume (OTC-25749-MS).

“We have an upcoming solution to the main problem of working with broadband data in quantitative interpretation, and that is the fitting of the low frequencies into the wavelet extraction,” Hardy told OE, adding that the company plans to make the system available commercially around January 2016.

“The challenge is that broadband’s ultra-low frequencies relate to wavelengths that far exceed the length of typical well sections. Ehsan uses what he modestly refers to as ‘mathematical tricks’ to get the low frequency information from seismic, and he mixes it with well log information to extract the broadband wavelet,” Hardy said.

“Away from the well, you no longer know what Earth impedances are, but you still have seismic, and if you’ve extracted the correct wavelet, you can solve for the Earth’s response in the form of impedances. You mathematically transform the entire seismic volume into an impedance volume – which you can then use to create fluid, lithology and porosity volumes, which geoscientists refer to as reservoir characterization,” Hardy said.

PCA and SOM help sift big data

Rocky Roden, consulting geophysicist for Geophysical Insights as well as president and chief geophysicist of Rocky Ridge Resources, presented a way to sift through the huge quantities of data now available due to improvements in interpretation and processing technology (OTC-25718-MS).

His method combines principal component analysis (PCA) with self-organizing maps (SOM) “to derive more useful information from multi-attribute data in the search for oil and gas.”

The first step is to choose seismic attributes for the self-organizing map, and Roden said principal component analysis appears to be useful for this purpose. PCA is a linear mathematical technique that reduces a large set of variables to a small set that still represents the most important variations in the large set.

Then, self-organizing mapping is used to achieve non-linear cluster analysis and pattern recognition to “identify patterns in the data that can relate to desired geologic characteristics.”

The key to all this is the increasing power of computers, and harnessing that power along with PCA and SOM to characterize exploration and development targets. This could represent the next generation of advanced interpretation, Roden said.