Rapid 3D inversion of gravity and gravity gradient data

to test geologic hypotheses

Leonardo UiedaValéria C. F. Barbosa

Observatório NacionalRio de Janeiro, Brazil

GGHS2012S7-044

Forward modeling

Observed

Predicted

Model

✔ Control

✔ Prior information

✔ Speed

✗ Tedious

✗ Gravity + gradients

✗ 3D

Geophysical inversion

Inversion

Observed data

3D model

Prior information

Regularization:● Damping● Smoothness

Inversion

Observed data

3D model

✔ Automatic fit

✔ Gravity + gradients

✔ 3D

✗ Control

✗ Prior information

✗ Speed

Source http://xkcd.com/747

Something in the middle

Planting anomalous densities

Uieda and Barbosa (2012), Geophysics

Observed data

Mesh

Zero density contrast

Seeds (user specified)

Predicted data

Neighbors

The best

New predicted data

The best

New predicted data

Fit!

Seeds = Skeleton

Source http://www.sciencebuzz.org

Inversion Body

Source http://dinocrisis.wikia.com

✔ Control

✔ Prior information

✔ Speed

✔ Automatic fit

✔ Gravity + Gradients

✔ 3D (only need skeleton)

Seeds

Example applications

Redenção granite

Outcrop

Oliveira et al. (2008)

Seed

Density contrast = -0.09 g.cm-3

Quadrilátero Ferrífero

Complex geology and topography

Complex geology and topography

Iron ore deposits

Seeds

Hypothesis testing

Registro do Araguaiaintrusion

Registro do araguaia

● Not outcropping

● Alkaline intrusion

● Density contrast ≈0.3 g.cm-3

After Dutra et al.

(2012)

After Dutra et al.

(2012)

3D gravity inversion (Dutra et al., 2012)

Bottom = 18 km

Hypothesis 1

18 km

Hypothesis 1

Hypothesis 1X

Hypothesis 2

18 km 5 km

Hypothesis 2

Hypothesis 2X

Hypothesis 3

5 km

Outcropping

Hypothesis 3

Hypothesis 3X

Hypothesis 4

5 km

Density 0.5 g.cm-3

Not outcropping

Not outcropping

Hypothesis 4

Hypothesis 4V

Conclusion

● 3D forward modeling = difficult

● Traditional 3D inversion = not flexible

● Planting densities = fast + flexible