Mapping Curie depth in western Canada from a wavelet analysis of magnetic anomaly data

The Curie depth (zC) is the point at which crustal rocks lose their magnetization, which occurs at a temperature of 580 °C for magnetite. Curie depth estimates therefore provide quantitative constraints on geotherms. This depth corresponds to the depth to the bottom of a magnetized layer and can be estimated from the spectral analysis of magnetic anomaly data. Due to the long wavelengths involved in its estimation, mapping the spatial variations in zC becomes a spatio-spectral localization problem. The wavelet transform using a combination of Morlet wavelets arranged in a `fan' geometry is particularly well suited to study spatial variations in the power-spectral density of magnetic anomaly data. In this study, we test the recovery of zC using the wavelet transform by synthesizing magnetic anomaly data using models of crustal magnetization. Our models are characterized by three parameters: the depth to the top (zt) and bottom (zb, or zC) of the magnetized layer, and a fractal dimension for 2-D crustal magnetization (β). Extensive tests are carried out using uniform parameters and spatially-variable zt and/or zb. We show that: 1) correcting for the fractal dimension β is essential to recover the correct layer parameters; 2) estimates of zb are biased toward low values if zb and zt are inverted simultaneously; and 3) the wavelet transform is able to retrieve the patterns of the crustal magnetization models when zt is fixed to its correct value. This method is then applied to magnetic anomaly data in western Canada, in a region that encompasses the Canadian Cordillera and parts of the Canadian Shield. The Cordillera is characterized by high heat flow and Moho temperatures in contrast to much lower values in the Shield, which suggests that Curie depth is likely to vary widely across the study area.