National Bushfire Intelligence Capability (NBIC) Terrain release 1.0

Terrain characteristics have a large influence on bushfire behaviour, affecting the rate of advancement of fire, as well as fuel accumulation and dryness. Terrain slope and aspect datasets are derived from a Digital Elevation Model (DEM) raster and are both important input to fire behaviour models. Those fire behaviour models form the core of the National Bushfire Intelligence Capability (NBIC) workflow, used to produce nationally consistent and locally relevant bushfire hazard and risk information. While national elevation dataset exist, those are often tailored to support other applications such as hydrology, artefacts only relevant to modelling bushfires are overlooked.\n\nHere we provide a nationally consistent, high resolution and terrain datasets, developed specifically to support fire behaviour models. This terrain dataset was developed by hybridising two smoothed DEM derived from the Shuttle Radar Topography Mission (SRTM). This allowed to represent the full Australian continent while maintaining an optimal spatial resolution of approximately 20m. The resulting terrain elevation dataset was then post-process to account for water bodies, before being converted to terrain slope and aspect using Sobel operators. These products were independently assessed by their authors and found elevation accuracy of 9.8m for 90% of the Australian continent.\n\nThis terrain dataset was instrumental in delivering the NBIC products, by providing a consistent yet locally relevant representation of terrain characteristics. In addition, this dataset can support future improvements towards terrain elevation dataset, for example by hybridising nationally consistent terrain information with locally available LiDAR dataset.\n\nFurther information about NBIC is available at https://research.csiro.au/nbic/home/data/terrain/\nLineage: DEM products are widely used operationally and in scientific research and are freely available. The Shuttle Radar Topographic Mission (SRTM) was heralded as a ‘breakthrough’ in improved global coverage and significantly improved spatial resolution [1]. \nProcessing steps\n•\tTwo SRTM products, Version 1 of the 1 second (~30m) data [2] and Version 3 of the 9 second (~250m) data [3]. Both data are the Smoothed Digital Elevation Model (DEM-S).\n•\tThe 1 second data does not cover the full extent of the Torres Strait in the north or Dirk Hartog Island in the west. To capture the topographical slope for these remote islands we use the 9 second data. \n•\tEach of the DEM datasets was pre-processed to coincident resolution and projection before being combined and deriving slope and aspect. A value of 0 m (mean sea level) was assigned to areas where no data exists as these areas are typically where water exists.\n•\tFor the purposes of conforming to National Bushfire Intelligence Capability (NBIC) Stage 1 data standards, datasets were resampled to 20m resolution using bilinear resampling. \n•\tSlope was then generated from the blended DEM using a Sobel operator [5] and smoothed with a 5 by 5 circular kernel window.\n•\tAspect is evaluated using a 3 by 3 window where the Sobel for the difference in x and y directions and converted to degrees North.\nLimitations\n•\tThe DEM-S dataset was subject to independent post-production validation and testing. An assessment of heights against discrete spatial data (Permanent Survey Marks) and local datasets of heights produced from several sources. These evaluations yielded a height accuracy of 9.8 metres for 90% of the data. Comparison of the DEM-S to the Tablelands Regional Council Contour Data by Gallant [4] highlights the particular issue of significant height error in highly vegetated areas and lakes with a reported difference in these types of areas being between 8 to 40 metres. Height error for the 9 second SRTM DEM product is similar to those of the 1 second product. \n•\tWhile the DEM-S dataset produces a significant improvement in the representation of bare ground, artifacts still exist. Known artifacts for the SRTM DEM-S are detailed in Gallant, 2011 [2]. The most difficult of these artifacts being the removal of tree offsets [4].\n[1] J. J. Van Zyl, “The shuttle radar topography mission (SRTM): A breakthrough in remote sensing of topography,” Acta Astronaut., vol. 48, no. 5–12, pp. 559–565, 2001.\n[2] J. C. Gallant, T. I. Dowling, A. M. Read, N. Wilson, P. Tickle, and C. Inskeep, “1 second SRTM Derived Products User Guide,” 2011.\n[3] M. F. Hutchinson, J. L. Stein, J. A. Stein, H. Anderson, and P. K. Tickle, GEODATA 9 second DEM and D8: Digital Elevation Model Version 3 and Flow Direction Grid 2008, Record DEM. Canberra: Geoscience Australia, 2008.\n[4] J. C. Gallant, a. M. Read, and T. I. Dowling, “Removal of Tree Offsets From Srtm and Other Digital Surface Models,” ISPRS - Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci., vol. XXXIX-B4, no. September, pp. 275–280, 2012.\n[5] L. G. Richards, “Machine perception of three-dimensional solids,” in Optical and electro-optical information processing, Cambridge, Massachusetts: MIT Press, 1965, pp. 159–197.