Skip to content
Open Data Finder
data.gov.au

Machine-learning infilled radiometric grids of Australia - potassium (K), thorium (Th), and uranium (U), 2026

Published by Geoscience Australia Data

About this dataset

This dataset provides spatially continuous national grids of potassium (K), thorium (Th), and uranium (U) across Australia at a nominal 100 m cell size in the GDA2020 / Australian Albers projected coordinate system (EPSG:9473). The grids are derived from airborne radiometric survey data combined with modelled infill in areas where observations are absent. Measured airborne radiometric data were compiled from Geoscience Australia’s Radiometric Map of Australia (radmap v4) and associated survey datasets, with additional recent survey coverage incorporated for Tasmania. Where observations are available, the grids retain measured values. In areas without coverage, values were estimated using a machine-learning approach based on a convolutional neural network (CNN) combined with residual correction, producing spatially consistent predictions. The final products are provided on a 100 m grid to enable consistent national coverage; however, the spatial density and effective resolution of the underlying airborne survey data vary across regions. As a result, the grids represent a combination of measured and modelled values and should be interpreted accordingly. In infilled areas, values are derived from statistical relationships with environmental predictor variables (e.g. geology, terrain, and vegetation) and may therefore introduce dependencies if used in analyses involving similar inputs.

Machine-learning infilled radiometric grids of Australia - potassium (K), thorium (Th),… · Open Data Finder