Magnetic map-making for advanced applications: Quantitative comparison of frequency dependent features, errors, and uncertainties in gridded magnetic data

Abstract

Listen

Translate article

Existing models of the Earth’s lithospheric magnetic field (e.g.; World Digital Magnetic Anomaly Map (WDMAM-2) and Earth Magnetic Anomaly Grid (EMAG2v3)) are composed of short wavelength information (< 100 km) from near-surface survey data and long wavelength information (> 300 km) from satellite data. In oceanic areas, compilations of ship trackline data provide the near-surface measurements used to construct global-scale gridded maps of the magnetic anomaly field. Although these maps have been used widely for tectonic and geodynamic studies, advanced applications, including complex inversions and the use of magnetics for alternative (to GPS) navigation, require renewed attention as to how gridded maps are made. Data selection, including detection of anomalous tracklines, knowledge of the sampling and power spectra of the potential field, quantification of uncertainty and an accurate representation of the gradients in the estimated field all represent areas of interest for advanced applications. We approach the problem of magnetic map-making for advanced applications by developing a means of quantitative comparison of magnetic data which we apply to each length scale of the underlying magnetic measurements and interpolated grids as a function of potential field frequency (spatial wavelength). Magnetic anomaly maps of the same region generally reproduce comparable long wavelength features. Further comparison of these maps indicates that dominant differences occur at short spatial wavelengths. Coherence analysis provides a technique to make a wavelength dependent comparison quantitatively, which can be used for data selection as well as to measure length-scale dependent errors and uncertainties. Coherence can help to assess if individual tracklines are consistent with the overall dataset and could help determine if these tracklines should be included in a final map product. Such a methodology could help automate or semi-automate the trackline selection and grid generation procedure. An understanding of the uncertainty at different length scales provides important information for the development and tuning of navigation algorithms and can provide an analytical framework for understanding different methods of map construction. In areas with high-quality reference maps this type of analysis can help inform scale dependent uncertainty models.