9 second substrate rasters for continental modelling of Australia's biodiversity patterns

These data provide consistent rasterised layers of edaphic (physical and chemical conditions of the soil) and land surface physiography (landform and geomorphology) variables hypothesised to explain spatial patterns in biological diversity at continental scales for immediate use with statistical modelling tools. These data are intended to be used along with a similarly compiled and spatially standardised set of climatic layers. Consistent "stacks" of raster variables are needed for spatially-explicit biodiversity modelling using tools such as MAXENT or Generalised Dissimilarity Modelling (GDM). Full details of each dataset, with a list of data sources and bibliography, are provided in a table as part of the data collection. Additional information provided with the 1km gridded raster is relevant to some these data and provided here also. Each dataset will need to be separately cited. These data have also been made available for use in the Atlas of Living Australia's Spatial Portal. \nLineage: Rasterised environmental attributes representing soil conditions, landform and geomorphology were compiled from a wide range of data sources. Each dataset was pre-processed to match a 9-second spatial data/nodata mask for Australia derived from the V3, 9-second digital elevation model (Hutchinson et al. 2008; ANU and Geoscience Australia). The layers were expanded to fill voids (NODATA values) and more fully capture the land surface of Australia. This is not the ideal way to account for these areas, but adequate pending more thorough updates in the future. Layers were expanded using the DATA option of the FOCALMEAN command with a CIRCLE expand radius of 2 cells. The focal mean command expands data values into nodata areas using the mean value of nearby cells. The focal mean command changes all the values in the original grid, so the cell values in the expanded areas are put back onto the original grids. The 2-cell expansion is applied iteratively until all NODATA values or voids are filled, and an expanded coastline mask is applied to limit the coast expansion. For ordered data derived from soil map classifications, the NEAREST value was used. \n\nData were subsequently masked to consistently define data/nodata values across all layers in the stack (dataset MASK0). They are supplied in ArcGIS floating-grid formats in the GDA94 geographic reference system. \n\nThe filling of lakes and voids may have implications for communication or interpretation of subsequent terrestrial biodiversity models using these data, where taxa are geographically predicted to occur where they would not be expected to be found. Outputs should therefore be masked to reinstate coastline and lakes, as needed.