The Hidden Geodetic Bill in a 'Harmonised' Seismic MegaSurvey

What TGS Has Actually Taken On

TGS has signed an agreement with Equatorial Guinea’s Ministry of Hydrocarbon and Mining Development to build a large-scale offshore multi-client seismic dataset, the first phase of a planned nationwide MegaSurvey across the country’s offshore basins. The opening phase covers post-stack reprocessing of approximately 27,273 km of 2D and around 35,000 km² of 3D data, with completion targeted for the third quarter of 2026. The full vision reaches roughly 46,343 line kilometres of 2D and more than 59,000 km² of 3D, spanning the Rio del Rey and Rio Muni basins. David Hajovsky, Executive Vice President, Multi-Client, described the result as a basin-wide regional screening tool supporting prospect identification, ranking and work-commitment planning, with TGS integrating and reprocessing legacy datasets into what it calls a consistent regional framework.

The press language is about imaging technology and prospectivity. What it does not mention, and what actually decides whether the word “harmonised” survives contact with the data, is the geodetic and positioning reconciliation underneath. That is the subject we examine here. The phrase that should concentrate the mind is “post-stack reprocessing” of “legacy datasets”: this is a merge of decades of pre-existing surveys, not a clean-sheet acquisition. The positioning is already baked into the traces. You inherit it, and you cannot re-navigate it.

Why the Coordinate Reference System Decides the Product

A MegaSurvey is sold to many buyers, each of whom ties it to their own wells, their own licence-block coordinates and their own depth-conversion models. The commercial promise of a “harmonised and seamless” regional product rests on a single assumption: that every contributing survey has been brought onto one coordinate reference system (CRS) with the correct datum transformations applied. Get that wrong and the error does not stay with TGS. It propagates to every customer, who ties wells to the wrong place and ranks prospects against a structural picture that is distorted by positioning rather than geology.

The exposure is sharpest at boundaries. Licence blocks in any offshore jurisdiction are defined by the regulator in a specific geodetic datum. If the survey CRS and the block-definition datum are not reconciled, a prospect can plot on the wrong side of a boundary. A datum mismatch in the Gulf of Guinea is not a rounding error: a WGS72-to-WGS84 difference is on the order of 10 to 17 metres, and a poorly documented local datum shift can run to hundreds of metres. At regional screening scale those numbers look small, but at the point a client commits a well or negotiates a block they are not small at all.

There is a second cost that never appears in the announcement: the forensic effort to recover the geodetic provenance of each legacy survey. Field navigation files, datum statements and bin-grid definitions from older campaigns are frequently missing, contradictory, or recorded in formats that have since fallen out of use. Reconstructing them is slow, specialist work, and it sits on the critical path to a Q3 2026 delivery whether or not anyone budgeted for it.

What Drives a Defensible Regional Framework

The first decision is the project CRS. A nationwide footprint that reaches toward a UTM zone edge forces a choice between a single project projection carried consistently across the whole area and a zone-by-zone approach that introduces scale-factor and grid-convergence discontinuities at the joins. For a regional product intended to read as one continuous dataset, a single documented projected CRS is usually the right call, with the scale factor managed deliberately at the margins rather than discovered later as an apparent mistie.

The second driver is metadata recovery and verification. Every contributing 2D line and 3D survey carries a native CRS, a datum and an acquisition-era positioning quality. Transit and Doppler satellite navigation from the earliest campaigns delivers absolute accuracy at the hundreds-of-metres level. GPS acquired during Selective Availability sits around the hundred-metre mark. Differential GNSS brings that to single metres, and modern multi-constellation GNSS with inertial aiding reaches decimetres. When a 1980s 2D line crosses a recent 3D survey, the older line’s absolute positioning uncertainty can be an order of magnitude worse. The apparent offset at that intersection is positioning, not structure, and the framework has to be able to tell the two apart.

The third driver is the post-stack reality. Because the data is already stacked, the common-midpoint bin positions are fixed. Harmonisation is therefore a coordinate-transformation, re-gridding and signal-matching exercise, not a re-binning of field positions. You can transform header coordinates onto the project CRS, re-grid onto a common output bin grid, and match phase, amplitude and spectra across vintages. You cannot retrospectively improve the navigation that produced the original bins. That bounds what “harmonised” can honestly mean, and a clear-eyed buyer should understand the difference.

The governing standards are well established. The EPSG Geodetic Parameter Dataset, maintained by IOGP, is the reference for CRS and transformation definitions, and IOGP’s geomatics guidance notes (the 373 series) set out correct use. Positioning data exchanges through the IOGP P-formats: P1/90 and the later P1/11 for processed positioning (navigation), P2/94 and P2/11 for raw positioning data, and P6/11 for bin-grid definition. Trace data carried in SEG-Y Rev2 can hold an explicit CRS in its extended headers, where earlier revisions could not. These are the yardsticks against which the merge should be audited.

Where Regional Merges Come Apart

1. Trusting the SEG-Y header at face value

SEG-Y Rev0 and Rev1 carry coordinates without a formal CRS, and the coordinate units and scalar live in specific header bytes that are routinely set wrong or left ambiguous. Headers in arc-seconds get read as metres, or a coordinate scalar is ignored, and the survey plots tens of kilometres from where it belongs. Every contributing dataset’s header coordinates must be checked against its independent navigation, not assumed correct because the data loaded without complaint.

2. Assuming “WGS84” means one realisation

A survey labelled WGS84 may have been positioned through a regional DGPS reference frame with its own offset, or through an early realisation that differs from the current one at the decimetre-to-metre level. The label is a starting point for investigation, not a conclusion. Where the metadata says WGS84 but the field reality was a local or regional frame, the merge inherits an undocumented shift that surfaces later as a stubborn, basin-wide mistie nobody can explain.

3. Defaulting to a null datum transformation

The most common single failure is applying an identity transformation between datums that are genuinely different. WGS72-era data merged into a WGS84 framework with no transformation will sit consistently displaced from the modern surveys. The discipline is to apply the documented transformation with a stated accuracy from the EPSG dataset, and to record the transformation code used for every survey, rather than letting the software assume the datums coincide.

4. Reading a positioning mistie as a geological one

A regional 2D dataset of this scale has thousands of line intersections. At each, position and phase both have to tie. Treating an offset as a fault or a structural feature when it is actually the absolute positioning error of an older line is how spurious leads enter a screening product. The reverse error is just as damaging: forcing a tie that masks a real positioning problem and bends the geology to fit.

5. Ignoring the bin-grid definition when re-gridding

A 3D bin grid carries its own origin, bin dimensions and rotation, and merging surveys onto a common output grid without reconciling each native grid against the project CRS produces local smearing and apparent shifts at survey boundaries. Merging surveys onto a common output grid without reconciling each native grid against the project CRS produces local smearing and apparent shifts at survey boundaries. The bin-grid definition in P6/11 is part of the geodetic record, not a processing afterthought, and it has to be carried and checked for every 3D vintage in the merge.

Specifying the Geodetic Control Before You Commit

For a buyer evaluating this product, or for any operator commissioning a regional merge, the controls below are concrete and testable. They cost far less to demand up front than to retrofit after a client has tied a well.

• Require a geodetic register for every contributing survey. No dataset enters the merge without its source CRS (with EPSG code), datum, the datum transformation applied to the project CRS (with the EPSG transformation code and its stated accuracy), the bin-grid definition (P6/11) and the navigation format (P1/90 or P1/11). Treat a missing entry as a hold point, not a detail.

• Fix and document one project CRS. Define a single projected CRS for the national framework, with EPSG codes for the geographic CRS, the projection and every transformation, and carry that definition in SEG-Y Rev2 headers so it travels with the data to every client.

• Set a mistie tolerance and QC every 2D intersection against it. Tie positional misties to a fraction of the bin dimension, on the order of a quarter bin, and route every intersection that exceeds the threshold to a forensic review that separates positioning from phase before any geological interpretation is allowed to stand.

• Verify each SEG-Y against its independent navigation. Confirm the coordinate scalar and units in the header bytes, and prove that the header coordinates plot where the P1 navigation says they should, for a sampled set from every vintage.

• Reject the null transformation as a default. Where WGS72-era or local-datum data exists, apply the documented transformation with its accuracy, and record the transformation code in the deliverable so any client can reproduce it.

• Reconcile against the regulator’s licence-block datum. Confirm that prospects fall on the correct side of block boundaries in the datum the Ministry uses to define them, not only in the project CRS.

• Commission an independent positioning and geodetic audit before commercial release. Audit against IOGP’s geomatics guidance and the EPSG dataset, and require the geodetic register and mistie report as named deliverables alongside the imaging products.

The imaging technology is real and it matters. But on a legacy merge of this size, the imaging is not the binding constraint on whether the product is genuinely continuous. The geodetic control is. A buyer who asks for the geodetic register and the mistie report, and reads them, learns more about the integrity of an Equatorial Guinea MegaSurvey than any statement about the latest algorithm will tell them.

---

Based on: TGS to Create Major Offshore Seismic Dataset for Equatorial Guinea