How to combine layer based & voxel based gravity modelling for frontier basin exploration

GaudPouliquen

In the early stage of exploration, seismic coverage is often sparse or nonexistent. Analyzing gravity anomalies can help unravel the basin architecture, sediment thickness, and delineate structural trends and crustal domains, all necessary to assess the hydrocarbon bearing potential of a basin.

In this example we demonstrate how public domain gravity data inversion, carried out with limited constraints, can reveal a meaningful, basin-scale preliminary density model of the subsurface. We take on a scarcely understood area offshore, East Africa, and combine 2D and 3D modelling, using both GM-SYS and VOXI. In a challenging economic environment, this low cost modelling approach can prove extremely useful to explore frontier basins and outline the regional structural and tectonic framework.

Some of the benefits are that it relies on public domain data, which can be easily integrated directly in Oasis montaj through Seeker. It makes the most of Geosoft's modelling capabilities, by combining both our layered-Earth based and voxel-based modeler to produce a finely tuned density model of the prospective basins. Low-cost! Since it's using public domain data.



Figure 1- Sandwell version 23 gravity data, Free-air anomaly. Green polygon shows extent of GM-SYS 3D model. Black polygon: AOI of VOXI inversions. Grey lines: location of GM-SYS Profile model.

Workflow:

• Use Seeker to download version 23 of Sandwell and Smith gravity model (Sandwell et al., 2014) and the EMAG2 data.
  
• In this example, we use Gz, and not the VGG (Gzz), because we're interested in density variations within the sediments but also in the upper and lower crust. If you're mostly interested in density variations in the sedimentary section, and in the shorter wavelength of the gravity anomalies, we recommend that you use the calculated Gzz instead (see MacLeod et al., 2016).
  
• We upward continued the Free-Air Gz anomaly to 1 km above Mean Sea Level.
  
• With the aim to build a 3D density model, we integrate within our project the Moho depth from Crust 1.0 (Laske et al., 2013, link to CRUST 1.0), and sediment thickness from NOAA (Whittaker et al., 2013), as well as the Sandwell bathymetry offshore, merged with the SRTM data onshore. This can be easily done using Seeker.
  
• Using these horizons, we built a layer-based 3D density model of the subsurface, including seafloor, basement, lower crust and Moho, using Geosoft GM-SYS 3D. The starting lower crust grid was created by extrapolating the 2D models and using a % of the crystalline crustal thickness. Then adjusted it to the 3D model using a long wavelength filtered inversion after adjusting the Moho.
  
• In parallel to the 3D modelling, we model the gravity anomaly along a few selected lines (figure 1). On two of these lines, we have seismic sections to guide our modelling. This allow us to check and tie our initial 3D model, and also to calibrate the depth of the lower crust horizon (figure 2).
  
• Using GM-SYS 3D and FFT calculations allows a rapid calculation of the model's gravity anomaly.
  
• We then export this starting 3D density voxel from GM-SYS 3D (figure 3).
  
• We prepare the data for VOXI: first, since VOXI offers the Cartesian Cut Cell (CCC) method to represent our bathymetry, we forward calculate the gravity effect of the water-sediments contact, and subtract it from our Free Air anomaly.
  
• We want to use VOXI to refine our density model within the upper crustal section, sediment, and upper crust. So we also remove the contribution of the lower crust and the Moho layers, which we have modeled in GM-SYS 3D. We end up with a residual Gz anomaly, which should reflect the contribution of the sediments and upper crust only.
  
• We amend the densities of the starting model exported from Geosoft to reflect these corrections (we can simply subtract the density we've used for the sediment-water contrast, and fix the density to zero below the upper crust). This model is set as both our starting model and parameter model in VOXI.
  
• We set up upper and lower bounds to limit the range of variations of the densities in the sediments and the upper crust.
  
• Finally, we run the inversion and get a 3D distribution of density variations. We can add up the reduction density back to this model to obtain a final density model over the area (Figure 4).
  

This 3D density model can be compared back to the 2D gravity model by extracting section and using them as overlay in GM-SYS 2D. Or horizontal sections can be extracted from the voxel to help the interpretation. This is a versatile workflow which can be adapted to the resolution of the input gravity data (e.g. use Gzz and concentrate the inversion on the sedimentary section only). However as with every inversion the more constraints you can add, the more robust your inversion results will be.



Figure 2 - GM-SYS 2D Gravity model across the Obbia basin (Obbia model in Figure 1)

Some tips:
You can use either VOXI or GM-SYS 3D to reduce your Free-Air anomaly from the water-sediment interface. Using VOXI you'll take advantage of the CCC approach and get more accurate calculation. If you use VOXI:

• Ensure that the size of your mesh adequately reflect the bathymetry. You might need to refine the vertical cell size of your model.
  
• When forward modeling in VOXI, padding is a key parameter to consider. Early on in your project, when you download data such as the depth of the Moho from the CRUST 1.0 model, always consider a larger area than your AOI, when available (and if not, carefully extrapolate your surfaces). GM-SYS 3D and VOXI handle padding differently.
  
• Remember that when setting up an inversion in VOXI, if you use constraints, such as upper and lower bound, starting model, or parameter model, these need to be consistent in term of densities. For example if your input gravity data is a Bouguer anomaly, remember to take this into account when you build your starting model and parameter model (i.e. reduce the densities from the Bouguer reduction densities in your starting model).
  
• GM-SYS 3D allows draped calculation, this can be useful if your work with an onshore-offshore project. And since 9.0, we use the GPU to speed up the calculation.
  

Figure 3 - starting model extracted from GM-SYS 3D



Figure 4 - Final density model after VOXI inversion

For more information about this method, join us for the keynote presentation at the 2017 CGS/SEG International Geophysical Conference

GaudPouliquen

References
Laske, G., Masters., G., Ma, Z. and Pasyanos, M., [2013], Update on CRUST1.0 - A 1-degree Global Model of Earth's Crust, Geophys. Res. Abstracts, 15, Abstract EGU2013-2658.
MacLeod, I. N., G. Connard, and A. Johnson. "Inversion of Public Domain Gravity Gradient Data in the Gulf of Mexico." 78th EAGE Conference and Exhibition 2016-Workshops. 2016.
Makris, J., Papoulia, J., McPherson, S. and Warner, L., [2012], November. Mapping of sediments and crust offshore Kenya, east Africa: a wide aperture refraction/reflection survey. In 2012 SEG Annual Meeting. Society of Exploration Geophysicists.
Sandwell D.T., Muller R.D., Smith W.H.F., Garcia E., Francis R., [2014], New global marine gravity model from CryoSat-2 and Jason-1 reveals buried tectonic structure, Science, 346, 65-67.
Whittaker, Joanne, Alexey Goncharov, Simon Williams, R. Dietmar Müller, German Leitchenkov (2013) Global sediment thickness dataset updated for the Australian-Antarctic Southern Ocean, Geochemistry, Geophysics, Geosystems

Applied Learning
Geosoft VOXI Earth Modelling
Geosoft Seeker
GM-SYS Profile Modelling
GM-SYS 3D Modelling