Inversion of Z-Axis Tipper Electromagnetic (Z-TEM) Data The UBC Geophysical Inversion Facility Elliot Holtham and Douglas Oldenburg
Outline Introduction Transfer functions and Z-TEM data Conducting block example Synthetic inversion Field data inversion Conclusions Questions / Discussion
Z-TEM Technique Z-TEM technique uses natural source fields similar to the magnetotelluric (MT) technique Vertical magnetic fields due to natural sources are recorded over the survey area
Z-TEM Technique Data relates the vertical magnetic field to the horizontal field at some fixed grounded reference station Reference station compensates for unknown source field amplitude Large areas can be surveyed quickly and economically Promising technique to find large scale structures at depth
Transfer Functions Transfer functions relate the vertical magnetic field measured above the earth to the horizontal magnetic field at some fixed reference station Two unknowns and only one equation Source polarization is assumed to be random Use measurements from independent source polarizations
Computation of Transfer Functions for a Conducting Block Source electric field into the page Resulting charge buildup leads to a secondary current Conducting block
Transfer Function Computation for Synthetic Example Secondary electric field Resulting transfer functions
Inversion of Z-TEM Data Inversion algorithm has been implemented where : Regularization parameter Q: Projection matrix u: Fields : Observed data : Model and Reference model W d, W : Data error, model weighting Minimize = d + m
Inversion of Z-TEM Data Minimize = d + m Gauss-Newton method
Synthetic Inversion Example Data computed at 1, 3.2, 5.6, 10, 18, 32 Hz Reference Station: (-3000, -3000, 0)m Data collected at a constant height of 100m Data collected over an area of 2500 x 2500m 10m data spacing and 50m line spacing
Synthetic Inversion 1 Hz Observed data Predicted data Misfits
Synthetic Inversion 3.2 Hz Misfits Observed data Predicted data
Synthetic Inversion 5.6 Hz Misfits Observed data Predicted data
Synthetic Inversion 10 Hz Misfits Observed data Predicted data
Synthetic Inversion 18 Hz Misfits Observed data Predicted data
Synthetic Inversion 32 Hz Misfits Observed data Predicted data
Misfit Target misfit: 1.2E+06 Final misfit: 1.39E+06 Misfit
Synthetic Inversion Model Depth 0 m True Model Inverted Model
Synthetic Inversion Model Depth -500 m True Model Inverted Model
Synthetic Inversion Model Depth m True Model Inverted Model
Synthetic Inversion Model Depth m True Model Inverted Model
Synthetic Inversion Model Depth m True Model Inverted Model
Synthetic Inversion Model Northing m True Model Inverted Model
Synthetic Inversion Model Northing -500 m True Model Inverted Model
Synthetic Inversion Model Northing 0 m True Model Inverted Model
Bingham Canyon Field Data Inversion Bingham Canyon site is 50 km west of Salt Lake city. Z-TEM data acquired in the winter of 2008 Rugged topography 30, 45, 90, 180, 360 Hz, real and imaginary components 1000 m flight line spacing line-km of data 96 x 92 x 95 cell mesh
Field Inversion Work flow
Error Assignment The range of the data is determined by sorting each data (ie. real Tzx, Imag Tzx, Real Tzy, Imag Tzy) The range is set to be the difference between the 90 th and 10 th percentile data The assigned standard deviation is then a small fraction of this range. Initially C=0.125 for all frequencies.
Error Assignment Invert each frequency of the data with the assigned standard deviations. Adjust the constant C, until the final misfit achieves the target misfit Misfit
Error Assignment Inverting and adjusting the errors on each frequency separately gives the correct weighting of each frequency Ensures that no frequency dominates inversion 45 Hz misfit – scale factor Hz misfit – scale factor 1.90
Final Misfit Curve Target Misfit: 1.52 E+06 Final Misfit: 1.58 E+06 Misfit
30 Hz Misfits Observed data Predicted data
45 Hz Misfits Observed data Predicted data
90 Hz Misfits Observed data Predicted data
180 Hz Misfits Observed data Predicted data
360 Hz Misfits Observed data Predicted data
Cut-offs (Below 1600m) Conductors > 0. 1 S/m Resistors < S/m
Comparison of Inverted Model with Geology (Surface resistors and conductors) Inverted Model Geologic Model
Comparison of Inverted Model with Geology (Surface resistors and conductors) Inverted Model Geologic Model
Comparison of Inverted Model with Geology (Surface resistors and conductors) Inverted Model Geologic Model
Comparison of Inverted Model with Geology (resistors and conductors below 1600m) Inverted Model Geologic Model
Conclusions Z-TEM data can be forward modeled Inversion algorithm exists for inverting Z-TEM data Inversion yields encouraging results on a synthetic model Z-TEM technique has been applied to a field dataset and yields good results Shows promise to find large scale structures at depth
Acknowledgements Michael Zang and Exploration Syndicate, Inc. Thank You