08:30 〜 08:45
[S26-1-01] Shallow crustal heat flow and heat production inversion
Understanding the heat flowing from the Earth is fundamental to understanding the nature of the crust. Previous attempts to analyse the surface heat flow across the Australian continent have typically interpolated the sparse available data points. These studies inherently make the assumption that lateral distance between heat flow determinations is sufficiently short to act as a proxy for changes in the estimated rate. However, the variations apparent in the data suggest that such an assumption cannot hold.
Surface heat flow estimates can be regarded as the sum of heat flowing from several source components. The observation of linear heat flow provinces suggested that, over large distances, the heat production at the surface of the crust provides a first-order constraint on surface heat flow rates. Regardless of the physical mechanisms behind these patterns, this seemed to suggest that the remaining heat flow component, commonly referred to as ‘reduced heat flow', exhibits large wavelengths of variation that are, in general, more amenable to spatial interpolation over large distances.
We approach heat flow inversion as regression problem. The reduced heat flow signal is regressed from surface heat flow determinations, where upper crustal heat production provides a purely-additive noise that masks the coherence of the reduced heat flow signal. We use a transdimensional-tree Markov Chain Monte Carlo model to regress reduced heat flow across the Australian continent. Our prior beliefs about the heat production in the upper crust, informed by surface geochemical measurements, enable the two components of surface heat flow to be probabilistically separated. In this format, the noise modelled in our regression can be recovered and represents the total inferred heat production for the shallow continental crust. This method is applied to the Australian surface heat flow dataset, and the results and their uncertainties are presented.
Surface heat flow estimates can be regarded as the sum of heat flowing from several source components. The observation of linear heat flow provinces suggested that, over large distances, the heat production at the surface of the crust provides a first-order constraint on surface heat flow rates. Regardless of the physical mechanisms behind these patterns, this seemed to suggest that the remaining heat flow component, commonly referred to as ‘reduced heat flow', exhibits large wavelengths of variation that are, in general, more amenable to spatial interpolation over large distances.
We approach heat flow inversion as regression problem. The reduced heat flow signal is regressed from surface heat flow determinations, where upper crustal heat production provides a purely-additive noise that masks the coherence of the reduced heat flow signal. We use a transdimensional-tree Markov Chain Monte Carlo model to regress reduced heat flow across the Australian continent. Our prior beliefs about the heat production in the upper crust, informed by surface geochemical measurements, enable the two components of surface heat flow to be probabilistically separated. In this format, the noise modelled in our regression can be recovered and represents the total inferred heat production for the shallow continental crust. This method is applied to the Australian surface heat flow dataset, and the results and their uncertainties are presented.