This is the title of an example SEG abstract using ...

39 downloads 0 Views 526KB Size Report
Desmond FitzGerald*. GeoIntrepid Namibia. Intrepid Geophysic. P.O. Box 86685. Suite 110, 3 Male Street. Windhoek, NAMIBIA. Brighton, Victoria, 3186.
Merging FTG Data and the Long Wavelength Problem Rainer Wackerle GeoIntrepid Namibia P.O. Box 86685 Windhoek, NAMIBIA [email protected]

Desmond FitzGerald* Intrepid Geophysic Suite 110, 3 Male Street Brighton, Victoria, 3186 [email protected] and the southern blocks flown and processed by one and the central block by a second contractor. All surveys have a nominal terrain clearance of 100 metres and a traverse line spacing of 1 km with tie lines located at 5 km spacing. The central survey however was amended to include an infill part with line spacing of 500 m. To validate the decision to infill, the contractor was requested to produce the final data with the maximum resolution possible, whilst the other two surveys were processed with a wavelength cut-off equal to the line spacing. This resulted in quite different high frequency patterns between the surveys as demonstrated by the Tzz image of the northern overlap area shown in Figure 2.

Summary Geophysicists have always wanted to move their instruments away from static measurements on the ground to more flexible platforms. Advances in data acquisition and processing technology have allowed the employment of potential field and radiometric acquisition systems from small aircraft to such an extent that nowadays countries like Australia and Namibia have a virtual complete coverage with high resolution airborne magnetic and radiometric data. The individual surveys, each a marvel on its own, have to be merged into one consistent dataset to facilitate regional interpretation and data distribution. The merging process however exposed long wavelength differences between the surveys of a higher order than simple DC shifts. These differences are the result of a changing environment such as the IGRF as well as differences in the applied data processing. Elaborate merging software was developed to obtain one seamless product from millions of line kilometres of airborne data. The latest development in airborne geophysics is Airborne Gravity Gradiometry (AGG) which allows the sampling of the curvature of the gravity field down to wavelengths of a couple of hundred metres. Albeit at a relatively high price, AGG has established itself firmly in the exploration for hydrocarbons and minerals and the first adjacent surveys are now available for merging. When processing AGG data however, be it from full tensor or partial tensor measurements, the relationship between the tensor components should be kept at all times. Merging the tensor components individually from various surveys using the tools available for scalar data and then reassembling the merged components into a final tensor dataset may lead to a violation of the Laplacian nature of the gravity field.

Figure 1

Original Tzz data

Figure 2 overlap

NE shaded image of the original Tzz data – northern

Introduction In 2012 Tullow Oil PLC completed an extensive AGG programme over its licenses in the East African Rift Valley. From the more than 70,000 line kilometres produced three surveys totalling about 58,000 line kilometres were flown with slight overlaps between each other and are merged into one consistent dataset. Figure 1 shows the original delivered Tzz data and gives an overview of the survey layout. A Lockheed Martin FTG instrument was used for all three surveys, with the northern

Merging FTG Data

It should be noted that the surveys were flown under very difficult conditions. Not only the rugged terrain, but also the hot and windy weather conditions proved to be a challenge for the acquisition of airborne gravity data. Full tensor gradient (FTG) data and the vertical gravity (Gz) were delivered at the planned drape surface elevation for the northern and southern survey blocks, whilst the data from the central block were calculated for an idealised drape surface that neither represents what was actually flown nor the planned elevations.

merged grid. The results of this process are presented in Figures 3 and 4 showing Tzz images from the two overlap areas.

New Drape Surface As a first step a new combined drape surface had to be produced from the drape elevations delivered by the contractors. This was achieved by a standard DC merge followed by a wide feathering filter to remove residual mismatches in the overlap areas. The application of a final Gauss deconvolution filter yielded the new drape surface to which the gravity data have to be migrated. Care was taken that the migration to the new drape did not involve downward continuation over large distances since this may result in mathematical instabilities and oscillations.

Figure 3

Tzz after altitude correction – northern overlap

Continue and merge FTG data To prepare the data for the continuation to the new drape new grids were produced from the delivered line data to an identical cell size and to avoid small differences caused by the different gridding algorithms employed by the contractors. The new tensor grids were then transformed into line data format resulting in a line dataset with identical point-to-point and line distances. The altitude correction from the original to the new drape is calculated using a locally fitted truncated Fourier series to the potential. From the potential, on an observation by observation basis, the altitude adjustment to the new drape surface is calculated. The altitude corrected data from all three surveys were then used as input data to calculate the

Figure 4

Tzz after altitude correction – southern overlap

Whereas the data from the central and the southern surveys merge almost perfectly there is a definite level shift between the central and the northern surveys. Level shifts do not only occur in the Tzz data but, with variable amplitudes, in all tensor components. The vertical components have the largest differences due to the fact that most of the energy of the gravitational field is contained in the vertical direction. Similar to a DC shift, which can be applied to correct scalar potential field data, a fraction of the average tensor of the entire grid can be added to a tensor grid without changing the relation of the tensor components or any potential modelling results. This was applied to the data from the northern block resulting in better consistency along the suture line to the central block. A tensor smoothing filter was finally applied to the merged grid to remove residual mismatches. The Tzz component of the final merged FTG tensor grid shown in Figure 5 can be compared to the original data in Figure 1. Clearly visible is the noise reduction effect in the central block as well as a slightly improved signal to noise ratio in the northern and southern blocks. Vertical gravity data Despite the fact that full tensor data contain intrinsic information about the source body geometry and that the measured gradients should be the preferred input data into semi-automatic interpretation techniques working on gradients such as Euler Deconvolution or Multi Scale Edge detection, most interpreters will only use Gz and Tzz data. Equally, only a very few modelling packages take advantage of or can handle full tensor data. It is hence important to investigate the vertical Gravity derived from FTG data. Due to their very nature, gradiometry measurements have difficulties in capturing long

Merging FTG Data

wavelengths accurately and the lack of accuracy may well be propagated into the Gz data. The Gz data delivered by the contractors were calculated as vertical integration of the Tzz data for the northern and southern blocks and from an equivalent source model in the central block. Both methods, but especially the vertical integration, may result in a misrepresentation of long wavelength components of the gravitational field. To prepare the delivered Gz data for merging, new grids were produced from the delivered line data using identical gridding algorithms for all three surveys. The grids were then transformed into line data format with identical point-to-point and line distances. To continue the data from Figure 5 Final merged Tzz data the delivered to the new drape a variable continuation filter was applied. Then, new grids were produced from the continued data and were merged into one consistent output grid using DC shifts based on the data in areas of overlap. As expected, the available tools to merge scalar potential field data do produce a reasonable merged grid as shown on the left in Figure 6. The question however remains as to how reliable the long wavelength components are captured by the AGG measurements. The large size of the merged data – Figure 6 covers an area of about 170 by 480 km – facilitates long wavelength comparison. An ideal dataset for such a comparison is the global DNSC08 satellite gravity model published by the Danish National Space Centre which is shown in Figure 6 on the right. Whilst the long wavelength components of vertical gravity derived from the AGG measurements generally conforms well to the satellite data the two do differ in the south-eastern part of the survey: The prominent low indicated by the satellite data just south of where the single lines of the central block meet the

Figure 6

The final merged Gz data on the left are compared to the data from the DNSC08 global gravity model

southern block is much subdued in the airborne data. This scenario continues south along almost the entire eastern arm of the southern block. The very deep low along the western arm of the southern block on the other hand is clearly evident in both, satellite and airborne data. Conforming airborne to satellite data It was because of these differences that it was decided to conform the airborne gravity data to the DNSC08 satellite data in order to provide an alternative vertical gravity dataset. The conforming process involves a low-pass filter applied to the satellite data and an identical high-pass filter applied to the airborne data. The filtered grids are then combined to produce the conformed grid containing the long wavelengths from the satellite data and the high frequency features from the airborne survey. To avoid unnecessary errors the satellite gravity data were also variable continued to the new drape. As cut-off wavelength for the filters a value of 50 kilometers was chosen. This uncommonly large value was deemed necessary to avoid suppressing the anomalies produced by the obvious volcanic plugs present along the

Merging FTG Data

eastern arm of the southern survey block. The image in Figure 8 demonstrates the effect of the conforming exercise when compared to the merged Gz data shown in Figure 6. Yet another possibility to get an alternative vertical gravity dataset is to integrate the merged FTG grid from all vertical components as opposed to only using the Tzz. The result of the integration exercise is shown in Figure 9. Unfortunately, the long wavelength components differ from both, the merged Gz and the Gz grid conformed to the satellite data, and therefor do increase the Gz ambiguity.

be modified to maintain the Laplacian relationship of the tensor components which should be kept over the entire processing chain. It was shown that FTG data from three surveys in the East African Rift Valley with a combined size of about 58,000 line kilometers and produced by two different contractors could be merged into an (almost) seamless combined dataset. The merging process not only minimizes level shifts between the surveys but is further believed to have improved the signal to noise ratio of the original data. One problem however remains and that is the consistency of long wavelength components between the merged surveys. Long wavelengths are best expressed by the vertical gravity (Gz) field, which also is the most familiar to interpreters. Gz data can be derived from tensor data by integration and through an equivalent source model. Both methods however may introduce long wavelength artifacts into the data. Global satellite derived gravity models can be used as control surface or to conform/correct the long wavelength components of the airborne data. The question remains however which of the various Gz datasets to trust for interpretation and modeling. It is the opinion of the author that more research is required in this field to eliminate the long wavelength ambiguity. Acknowledgements GeoIntrepid wishes to thank Tullow Oil PLC for the permission to use their data for this publication.

Figure 8 Merges Gz data conformed to DNSC08 global gravity model

Figure 9 Gz data integrated from vertical tensor components

Conclusions Merging of millions of line kilometers of airborne scalar potential filed data is done routinely on a regional or continental base since many countries are in the possession of large amounts of airborne data. The latest developments in airborne gravity focus on various AGG systems and the number of AGG surveys is steadily increasing to an extent that the first overlapping datasets are available for merging. The proven methods to merge scalar data however have to

Suggest Documents