The REHEATFUNQ Python package helps to work with the (residual) scatter of surface heat flow even in small regions. REHEATFUNQ uses a stochastic model for regional aggregate heat flow distributions (RAHFD), that is, the collected set of heat flow measurements within a region marginalized to the heat flow dimension. The stochastic model is used in a Bayesian analysis that
(1) yields a posterior estimate of the RAHFD which captures the range of heat flow within the analysis region, and
(2) quantifies the magnitude of a surface heat flow anomaly within the region, for instance through the generating frictional power.
The stochastic model underlying REHEATFUNQ views heat flow data, uniformly sampled across the region of interest, as a random variable. A gamma distribution is used as a model for this random variable and information from the global data set of Lucazeau (2019) is introduced by means of a conjugate prior (Miller, 1980). The detailed science behind the model is described in Ziebarth et al. (202X).
The analysis by Ziebarth et al. (202X) can be reproduced through the Jupyter notebooks contained in the subdirectory “jupyter/REHEATFUNQ/”. The location specified in the map below covers the region to which REHEAFUNQ is applied in this analysis.
REHEATFUNQ is a Python package that uses a compiled Cython/C++ backend. Compiling REHEATFUNQ requires the Meson build system and a number of scientific libraries and Python packages (and their dependencies) that are listed in the documentation.
A Docker image “reheatfunq” is provided as an alternative means of installation. The Docker image comes in two flavors, specified in “Dockerfile” and “Dockerfile-stable”. The former is based on the current “python:slim” image and downloads further dependencies through the Debian package manager, leading to a short image generation time. The latter bootstraps the REHEATFUNQ dependencies from source, aiming to create a reproducible model. To do so, “Dockerfile-stable” depends on the sources contained in “vendor-1.3.3.tar.xz”. If you plan to build the stable image, download both “REHEATFUNQ-1.3.3.tar.gz” and “vendor-1.3.3.tar.xz”, and see the README contained in the latter. Later versions of the “REHEATFUNQ” archive are compatible with the latest “vendor” archive.
A quickstart introduction and the API documentation can be found in the linked documentation.
In Haeger et al. (2022), we created a three dimensional model of the temperature distribution and the geothermal heat flow of the Antarctic lithosphere as well as a new model of the lithosphere-asthenosphere boundary (LAB). The models were obtained in a three-step approach: First, we calculate the initial temperature distribution in the upper mantle by iteratively combining seismic tomography (An et al., 2015; Schaeffer & Lebedev, 2013) and gravity data (Förste et al., 2014; Scheinert et al., 2016) considering composition and density variations self-consistently (Haeger et al., 2019). Second, we define the lithosphere-asthenosphere boundary in a thermal sense based on the resulting geotherm by assuming it corresponds to the 1300°C isotherm. Third, we solve the steady-state heat equation to obtain the temperature distribution and the geothermal heat flow in the lithosphere.
One crucial yet still largely unknown factor in the model is the parametrization of the crust. In order to overcome this, we calculated thermal models for a range of crustal properties that are described in detail in Haeger et al. (2022) and the related supplementary material. Here, we only share the conductive temperature and the geothermal heat flow model for the preferred model (n° 29 in the supplementary) in binary netCDF files. Additionally, we present the depth to LAB and surface and mantle heat flow maps, the latter represents the heat flow at the depth of the Moho discontinuity (Haeger et al., 2019) as .txt ascii tables. As a measure of uncertainty of the preferred surface heat flow model, the standard deviation of all calculated models is additionally given. The models are presented in polar stereographic projections with true scale at 71° South (Snyder, 1987) and span ±3700 km with a 10 km spacing in x- and y-direction, respectively. For the netCDF files, the depth ranges from the bedrock surface (BedMachine, Morlighem et al., 2020) which is defined as the 0 level to the LAB in a 1 km spacing. The depths to the Moho and the LAB are given relative to sea level.