Poster: Uncertainties on Palæogeographic Reconstructions: Effects of Interpolation Methods
Abstract
Poster presented during the American Geophysical Union (AGU) 2025 Annual Meeting in New Orleans (LA, USA), within Session "GP43A - Frontiers in Paleogeography". Palæogeography ― the geography of the Earth in deep-time ― can be derived from different methods, among which thermochonometric data inversion (in general at local or regional scale), computer-generated topography from global plate tectonic reconstruction (see TopoChronia1), or topography inferred from lithofacies deposits. In any cases, all palæogeographic maps stem from the interpolation of spatially irregularly-distributed data points (or nodes). Here, we assess the performance of seven methods from five providers to interpolate global topographic maps based on irregularly-distributed vector points (nodes) derived from the Panalesis plate tectonic model. We sample the elevation for each node to a reference Digital Elevation Model of the present-day (ETopo2022) and compare the 204 output maps to this very reference both in terms of absolute accuracy and terrain variations accuracy. Some open-source solutions from QGIS and associated libraries perform better than ArcGIS solutions. We find in particular that the Triangulated Irregular Network best meets our requirements for palæogeography interpolation. Errors are systematically the same across all methods, with different amplitudes, and caused by the distribution of the input nodes, that are either too dense in some areas and too far apart in other areas. Our input nodes are however sufficiently well distributed for rendering quick terrain variations. A minimal uncertainty threshold of approximately 2.2% is estimated. 1,https://github.com/florianfranz/topo_chronia