National Bushfire Intelligence Capability (NBIC) Terrain release 2.0
Published by CSIRO Data Access Portal
About this dataset
Existing national-scale elevation datasets often contain inconsistencies in spatial resolution, acquisition date, and vertical accuracy, and may include terrain artefacts in densely forested areas where radar-based methods can misrepresent vegetation canopy as ground surface [1]. These limitations can affect applications requiring accurate terrain representation, including bushfire behaviour modelling, hydrological analysis, and environmental assessment. This dataset provides a seamless, continental-scale bare-earth Digital Elevation Model (DEM) of Australia generated through the integration of multiple elevation datasets of varying resolution, date and source. A total of 1,549 open-access lidar-derived DEM datasets (513,620 individual files) were obtained through Geoscience Australia's Elevation Information System (ELVIS) and merged with the Shuttle Radar Topography Mission (SRTM) DEM [1]. All source datasets were standardised to a common coordinate reference system (EPSG:3577) and vertical datum (Australian Vertical Working Surface, AVWS). The final product provides national coverage of Australia at 90 m spatial resolution, matching the grid specification used for the National Bushfire Intelligence Capability (NBIC) Bushfire Fuels Classification 90 m dataset [2]. Processing included metadata extraction, quality ranking of source DEMs, vertical datum transformation, reprojection, DEM mosaicking and application of the DEM blending and smoothing method described in Gallant (2019) [3]. This continental implementation extends previous large-area DEM integration workflows [4] and includes additional procedures for standardised gridding and metadata harmonisation. Dataset quality was assessed using high-precision Permanent Survey Marks (PSM) acquired and cleaned from all redactions. Comparison with 100,463 independent PSM control points showed that the blended DEM achieved a statistically significant improvement in vertical accuracy relative to SRTM, reducing Root Mean Square Error (RMSE) from 3.32% (5.05 m to 4.88 m) and Mean Absolute Error (MAE) by 4.2% (3.32 m to 3.17 m), despite replacement datasets covering only 17% of the DEM extent. Most improvements were seen in high slope and elevation areas. The resulting dataset provides a nationally consistent, gap-free elevation surface suitable for bushfire behaviour modelling, vegetation mapping, hydrological studies, flood risk assessment and other environmental applications.