Parameters

The Supplementary Leaflet group of type: parameters provides a dedicated place for additional critical metadata related to processing and inversion. For example, an inverted models group may need information about the inversion such as the software, settings used, parameter values, and any post-processing applied.

The “parameters” group can take on whatever shape and content the user desires, as long as the type: parameters attribute is included.

 1inversion_parameters:
 2    dataset_attrs:
 3        type: parameters
 4        method: electromagnetic, time domain
 5        instrument: 30Hz Tempest
 6        mode: airborne
 7        property: electrical conductivity
 8
 9    variables:
10        software: GALEISBSTDEM 1-D time-domain deterministic inversion software
11        software_reference: "Brodie, R. C., 2015, GALEISBSTDEM: A deterministic algorithm for 1D sample by sample inversion of time-domain AEM data – theoretical details, accessed May 1, 2020, at https://github.com/GeoscienceAustralia/ga-aem/blob/master/docs/GALEISBSTDEM%20Inversion%20Algorithm%20Theoretical%20Details%20.pdf."
12        description: Inversions were done using a multilayered smooth model formulation in which 30 layer thicknesses were fixed and layer conductivities were solved for. Horizontal (X) and vertical (Z) components of the data were inverted separately. A vertical conductivity smoothing constraint, alpha_s = 1000, was applied. The inversion reference model used a half-space conductivity of 0.04 Siemens per meter (S/m) with a standard deviation of 1 S/m. The relative importance of the reference conductivity model, alpha_c, was set to 1.0. The horizontal and vertical separation between transmitter and receiver was given a lateral and vertical standard deviation constraint of 0.5 meters (m) in the reference model. The receiver pitch was also included with a 0.5 m standard deviation. These steps were repeated using the same inversion parameters but for reference models of higher (0.2 S/m) and lower (0.008 S/m) conductivity representing lower (5 Ohm-m) and higher (125) resistivity, respectively. A number of inversions were conducted with various homogenous prior model values, and constraints on resistivity. The final model parameters described above were selected because they best represent the physical understanding of the system and minimized data misfit. Final inverted resistivity values for each layer, layer thicknesses, and the uncertainty associated with these values can be found in the model dataset.
13        doi_calculation: The depth of investigation (DOI) for each model location was calculated using the difference between the low and high reference conductivity model results. Using the approach from Oldenburg and Li (1999), models from the low and high reference inversions were divided and rescaled producing a metric of their similarity. Models were similar where constrained by the data (shallow depths) and diverge back to their distinct reference model values when no longer constrained by the data. Therefore, the DOI was calculated as the threshold below which models were no longer informed by the data.
14        phid_cut: Individual models with a data misfit, "PhiD", less than or equal to 1.5 were accepted for final outputs and products. A new channel, "ACCEPT_FLAG" was added to the data file representing this misfit cutoff, with 0 = rejected models and 1 = accepted models.