Detecting shallow gas hazards with full tensor gravity

Aug. 1, 1998
A high frequency, low-density anomaly was present in the measured gradient data. [24,974 bytes] Modeling suggested that a 500 ft by 500 ft by 200 ft cube, with its base at 1,000 ft, was a good solution, convincing proof of a shallow gas hazard. [24,988 bytes]

Anomolies showing up in high-res tests

John Prutzman
Bell Geospace
The first successful moving platform gravity gradiometer has made high-resolution full tensor gradient (FTG) gravity data readily available for the first time. The instrument produces data that inherently contain more high frequency information than is possible to obtain with any gravimeter. Therefore, resol ution is significantly improved with gradient data.

Initial use of the FTG data suggests that gradients have resolution approaching that of seismic data under certain circumstances. The Bell Geospace moving platform gradiometer has the additional property that, at depth less than one kilometer, the signal-to-noise ratio is at least an order of magnitude better than that of gravity data. This characteristic suggests the possibility that FTG data may be useful for shallow hazard detection and delineation. An example from the Gulf of Mexico shows how this has been done.

During the interpretation work for an exploration/exploitation target, Bell Geospace interpreters noticed that a high frequency, low-density anomaly was present in the measured gradient data. The measured gravity signal over the same area fails to detect the anomaly that is very evident on the gradient data.

The high frequency anomaly appears monochromatic suggesting that it is caused by a single low-density mass anomaly and not a composite effect from multiple layers. As a rule of thumb, laterally limited mass anomalies give rise to vertical gradient signals that have spatial wavelengths at the surface (wavelengths measured horizontally) that are twice the depth (the Peters Rule, after L. J. Peters).

This is not always true. Shallow events can sometimes give rise to long spatial wavelengths. Fortunately, the reverse is not true since deep horizons never create high frequency anomalies. Therefore, we can make an estimate of the depth of the low-density anomaly by finding the half wavelength, which in this case is about 1,000 ft. This is too shallow an event for any of the interpreted layers to be causing the anomaly. No deeper layer geometry can cause such high frequency. Apparently, the gradiometer has detected a shallow gas hazard.

Hazard modeling

The obvious next step is to model this shallow gas hazard. At this point, some assumptions must be made about the expected density of the gas charged sediments. Drilling information from the area suggests that normal density at the estimated depth of 1,000 ft is 2.0 g/cc.

A number of values for the gas charged sediments were selected from a range of 1.30 g/cc to 1.65 g/cc. Since this study was not being done as a hazard survey, only simple cube shapes were used for testing having densities selected from the range of interest. Depths were allowed to vary approximately +500 ft from the expected 1,000-ft depth and the cube size was adjusted as needed.

While gravity models tend to be non-unique, gradient data seems to be a lot less so because of the increased resolution and the fact that five independent tensor components must to be matched. Only one of the five gradients is shown in the figures although all five were available during the study. The added constraints limit the number of geologically plausible models that can be found.

The modeling results suggested that a 500-ft by 500-ft by 200-ft cube, with its base at 1,000 ft, worked well if a density of 1.45 g/cc were selected. This model is shown in an accompany figure with the predicted gradient results shown as a line overlaying the measured gradient data shown as dots.

The results are convincing proof that a shallow gas hazard exists. Further work could be done on the geometry of this hazard to improve the match but this was not done since the primary objective was deeper targets. However, these results were compared to the existing shallow hazard survey for the area.

Surprisingly, this gas hazard had not been detected by the normal shallow hazard survey suggesting that the anomaly does not have a strong acoustic contrast or perhaps was simply overlooked in the original hazard survey work. Obviously, this feature is large enough to be a serious hazard to drilling if it were penetrated unexpectedly.

The detected anomaly has a signal strength of at least 12 E?tv?s, which is very large. The Bell gradiometer has a resolution in the 0.5-1.0 E?tv?s range, suggesting a much smaller anomaly or much less density contrast could be detected. Since high frequencies can be used to indicate anomaly depths, the FTG data could be depth sliced by frequency to create a very detailed density distribution map of the near surface.

Such very high-resolution work may be possible to depths of 5,000 ft or more based on sensitivity tests done to date. Below this depth, resolution would be comparable to seismic data and could delineate over-pressured zones that often appear as bright spots in the seismic data.


This study exploits the very high signal-to-noise ratio and high frequency sensitivity of the Bell gradiometer at shallow depths to detect a shallow gas hazard. Measured FTG data shows a distinct high frequency anomaly generated by a definite low-density zone.

Modeling work indicates that it is caused by a shallow gas hazard detected by the gradiometer but missed in the normal shallow hazard surveys. This suggests that FTG data could be used for detailed, high-resolution mapping of the near surface density regime prior to drilling. Sensitivity tests suggest FTG data could detect deeper over-pressure zones which often appear as bright spots on the seismic data.

Copyright 1998 Oil & Gas Journal. All Rights Reserved.