Large scale Laterally Constrained Inversion of over steeping IP decays in AEM
Large scale Laterally Constrained Inversion of over steeping IP decays in AEM
Tuesday, 11 June 2019
Davie West Lobby (Florida Atlantic University)
Abstract:
The induced polarization (IP) response in airborne EM data has been within the last few years a subject of numerous papers. Improved signal to noise ration of the AEM systems results in data where IP can be detected and therefore, there has been an extensive software development for inversion of such data using dispersive earth parameters.
A typical IP response, for example due to disseminated mineralization, usually leads to presence of negative transients in the late time gates. However, the IP response can also be caused by shallower layers such as clays where the chargeability is not high enough to reverse the sign above the noise level. We refer to this class of IP affected curves as over steeping decays. Such data can be still inverted with a nondispersive earth model, though leading to very high resistivities.
In this contribution, we analysed a survey in Howard River region in NT, Australia. The data were collected in July 2017 by Geoscience Australia when over 2000 km of lines were surveyed with the SkyTEM312 system. After a standard processing of the raw data all negative transients were removed as well as any coupled data. A standard inversion suggested some very resistive deeper layers, however, in some parts of the survey the inversion had high data residuals as it was not able to fit the data within their standard deviation.
Next, we inverted the same data using the airborne IP inversions implemented in the AarhusInv inversion code (Aarhus University). In the part with over steeping decays the models now show chargeable, relatively shallow layers, and the data residuals have dropped to be within the standard deviation of the data (in most cases). In the remaining part, where the data could be well fitted with the standard inversion the resistivity model has not changed significantly. In the Figure a 2D map of the data residuals for the resistivity only inversion is plotted (left) next to the inversion including IP parameters. Both inversions were executed as laterally constrained inversions with the same discretization.
We argue that over steeping decay data are common in many airborne EM surveys and a standard inversion can lead to dubious resistivity models. Here we show that even for large data we can proceed with an IP inversion that can reveal if shallow chargeable layers are present and if the resistivity models need to be altered.
A typical IP response, for example due to disseminated mineralization, usually leads to presence of negative transients in the late time gates. However, the IP response can also be caused by shallower layers such as clays where the chargeability is not high enough to reverse the sign above the noise level. We refer to this class of IP affected curves as over steeping decays. Such data can be still inverted with a nondispersive earth model, though leading to very high resistivities.
In this contribution, we analysed a survey in Howard River region in NT, Australia. The data were collected in July 2017 by Geoscience Australia when over 2000 km of lines were surveyed with the SkyTEM312 system. After a standard processing of the raw data all negative transients were removed as well as any coupled data. A standard inversion suggested some very resistive deeper layers, however, in some parts of the survey the inversion had high data residuals as it was not able to fit the data within their standard deviation.
Next, we inverted the same data using the airborne IP inversions implemented in the AarhusInv inversion code (Aarhus University). In the part with over steeping decays the models now show chargeable, relatively shallow layers, and the data residuals have dropped to be within the standard deviation of the data (in most cases). In the remaining part, where the data could be well fitted with the standard inversion the resistivity model has not changed significantly. In the Figure a 2D map of the data residuals for the resistivity only inversion is plotted (left) next to the inversion including IP parameters. Both inversions were executed as laterally constrained inversions with the same discretization.
We argue that over steeping decay data are common in many airborne EM surveys and a standard inversion can lead to dubious resistivity models. Here we show that even for large data we can proceed with an IP inversion that can reveal if shallow chargeable layers are present and if the resistivity models need to be altered.