Derivation of Nested Catchments
and Sub-Catchments for the
Australian Continent

Michael Hutchinson, Janet Stein and John Stein

Introduction

A nested series of drainage basins, catchments and sub-catchments has been determined from the new 9 second Digital Elevation Model of Australia. The catchments and sub-catchments will support comprehensive reporting for the National Land and Water Resources Audit (NLWRA). The new DEM was recently completed by CRES under contact to AUSLIG to overcome significant deficiencies in drainage structure in the first 9 second DEM. By using the new DEM, the catchments and sub-catchments determined by this project respect the true surface drainage structure of the Australian continent down to an approximate scale of 1:250,000.The new DEM and catchments and sub-catchments were determined in consultation with concurrent Audit projects modelling sediment transport and catchment condition. The new catchments and sub-catchments were also required to take into account existing AWRC River Basins and Drainage Divisions, with a view to possible future upgrade of these Basins and Divisions.The new catchments and sub-catchments form a fundamental data library layer, with initial distribution to relevant Land and Water Audit projects. Ongoing custodianship and arrangements for licenced access to the new data layer are being negotiated with AUSLIG by the Audit Management Unit.

Methods

The catchments and sub-catchments have been derived from the flow direction grid associated with the new 9 second Digital Elevation Model for Australia, as produced by CRES for AUSLIG. The DEM and the associated flow direction (aspect) grid were calculated by the ANUDEM program in 44 overlapping tiles corresponding to the standard 1:1 million topographic map series. The catchments and sub-catchments were then derived using standard ARC/Info watershed functions applied to the 9 second flow directions grid, but with several modifications and enhancements. These modifications improved connectivity of sub-catchments and hence their general suitability for reporting of results. These methods and the minimum area thresholds used are briefly described below. A full description of the methods used to determine the flow direction grid and the nested catchments and sub-catchments will be made available on the NLWRA website.

Revision of the Flow Direction Grid

ARC/Info watershed functions were applied to the 9 second flow direction grid after first developing an automatic procedure to revise flow directions in the flow direction grid. The revised flow directions are consistent with the elevations of the associated 9 second DEM, but remove curvilinear features (or "tails"), defined using the original flow direction grid, that were attached to sub-catchments via a single grid cell. These features arose in sub-catchments derived from the unedited flow direction grid mainly because of the limitations in single grid cells acting as seeds for multiple sub-catchments, as defined by standard ARC/Info procedures.The effect of the revised flow directions is shown in Figures 1 & 2. Figure 1 shows sub-catchments with tails defined using a 2.5 km² minimum area criterion on the original flow direction grid. Figure 2 shows sub-catchments without tails defined using a 2.5 km² minimum area criterion on the revised flow direction grid. The basic mechanism of the procedure is to associate tails with the upstream adjoining sub-catchment via the lowest adjoining seed point. This procedure also helped to identify minor deficiencies in the original flow direction grid that arose where neighbouring grid cells had crossing diagonal flow directions and where there were closed loops.

Sink Catchments

Many sink catchments occurred in inland regions with area less than the specified threshold. Each sink catchment was iteratively amalgamated with its lowest neighbouring catchment, via an appropriately low pour point, until the combined area of the catchment was equal to or greater than the specified minimum area threshold.

Coastal Catchments

Coastal areas are dominated by small catchments with seed points having a contributing area less than the minimum area threshold. In existing AWRC drainage basins, these have been associated with the nearest principal catchment, even though they do not flow into this catchment. The same approach was adopted here by iteratively combining small coastal catchments with a neighbouring coastal catchment, providing the neighbouring catchment was not in a different AWRC River Basin. The extent of these small coastal catchments varies with the minimum area threshold, as shown for Tasmania in Figure 3 and Figure 4.

Minimum Area Thresholds

The minimum area thresholds defining the catchments and sub-catchments have been specified in consultation with the CSIRO Land & Water modelling group for the Land and Water Audit, and with representatives from AFFA/BRS, the Murray Darling Basin Commission and the CRC for Freshwater Ecology. The four minimum area thresholds used were 2.5 km², 25 km², 50 km² and 500 km².

The sub-catchments defined at each level form a strict heirarchy because the seeds for catchments defined by a larger minimum area threshold are simply a subset of the seeds for sub-catchments defined by a smaller minimum area threshold.

The 2.5 km² minimum area criterion was selected as a base area unit approximating the minimum catchment area defining the AUSLIG TOPO-250K stream line network. This is the stream line data network used in the construction of the DEM. It is well approximated in Tasmania by those grid cells having a contributing area exceeding 2.5 km².

The 25 km² minimum area criterion is the agreed minimum area criterion in the Project Brief. The 50 km² minimum area criterion was selected in light of stated needs of the CRC for Freshwater Ecology, University of Canberra in their analysis of catchment and river condition. Sub-catchments for Tasmania are shown in Figure 5 and Figure 6 with minimum area thresholds of 25 km² and 50 km² respectively. Tasmania has a total land area of about 1% of the area of the Australian continent

Catchments for the entire Australian continent, with a minimum area threshold of 500 km², were required by AFFA/BRS. These catchments are shown in Figure 7.

River basins for the Australian continent, corresponding to the standard AWRC River Basins (AUSLIG 1997), are shown in Figure 8. With minor exceptions, these basins match the AWRC Basins to within graphical accuracy. Recommendations on how to correct these anomalies are given in the recommendations below. The derived basins do not recognise AWRC sub-divisions of the Murray-Darling Basin and the Lake Eyre Basin.

Statistics of the Catchments and Sub-Catchments

Summary statistics of the sub-catchments, catchments and river basins are presented in the three tables below. Table 1 shows summary statistics for all sub-catchments, catchments and river basins for the whole continent. The river basins were created by amalgamating all sub-catchments via their lowest pour points.

1. Summary statistics for all final sub-catchments, catchments and river basins.

Table 2 shows summary statistics for all sub-catchments, catchments and river basins excluding the amalgamated small coastal and sink catchments. These statistics represent a more consistent relationship between the minimum area thresholds and the summary catchment statistics.

2. Summary statistics for sub-catchments, catchments and river basins excluding amalgamated small coastal catchments and amalgamated sink catchments.

Table 3 shows summary statistics for sub-catchments and catchments that arose from just the small coastal sub-catchments.

3. Summary statistics for the amalgamated small coastal sub-catchments.

Format of the Catchments and Sub-Catchments

The nested sub-catchments have been supplied in the form of a single ARC/Info integer grid with an associated value attribute table classifying the sub-catchments into four nested levels as defined by the four successive minimum area criteria and the river basins. The spatial extent of the sub-catchment grid is defined by the primary 9 second flow direction grid. The nested sub-catchments lie between the primary 9 second flow direction grid and the AWRC drainage basins, as calculated from the 9 second DEM. Polygons bounding the catchments and sub-catchments have also been calculated.

Key findings

A key conclusion of the asscociated Audit modelling groups is that the new DEM, and its associated catchments and sub-catchments, have yielded a much improved definition of drainage structure across the continent. This has directly facilitated improved continent-wide modelling of sediment transport and catchment condition.

The new catchments, when amalgamated, show good, but not complete, agreement with the existing AWRC River Basins and Drainage Divisions. The discrepancies reveal some shortcomings in the AWRC drainage boundaries, particularly in the Western Division.

The discrepancies also reveal occasional deficiencies in the supporting DEM, where closer inspection revealed insufficient source topographic data used to calculate the DEM. There were also some deficiencies in sub-catchment determination in areas with divergent flow. A further (and final) revision of the 9 second DEM, using selected additional source topographic data, would be required to enable a comprehensive and accurate revision of the AWRC River Basins and Drainage Divisions.

Applications

The immediate application of the new catchments and sub-catchments has been direct support of Land and Water Audit Projects modelling sediment transport and catchment condition. The catchments and sub-catchments are then to be used to provide appropriate units for reporting by relevant Land and Water Audit Projects. The catchments and sub-catchments thus form a fundamental data library layer, with eventual distribution to the public domain anticipated.

The catchments and sub-catchments have not been explicitly related to known lakes and reservoirs. These data were not included in the source topographic data for the DEM, and it was difficult to find an appropriate strategy to incorporate the very large numbers of small and/or ephemeral lakes and reservoirs that crossed sub-catchment divides.

Recommendations

The new 9 second DEM, and the associated catchments and sub-catchments determined by this project, respect the true surface drainage structure of the Australian continent at this resolution, which is approximately equivalent to a scale of 1:250,000. These products therefore form a secure basis for supporting and reporting both current and future projects assessing land and water conditions across the continent at this scale.

Remaining discrepancies between the derived catchment divisions and the AWRC River Basins and Drainage Divisions should be resolved in two ways:

• Firstly, the AWRC catchment divisions in the Western Drainage Division should be revised. The new catchment analysis for the Western Division is in substantial agreement with the earlier coarser resolution (0.025 degree) analysis by Hutchinson and Dowling (1991) who also found substantial disagreement with the existing AWRC boundaries in the Western Division.

• Secondly, the 9 second DEM should be slightly revised by obtaining additional source topographic data in the few areas where drainage structure is not well defined by the DEM, as evidenced by minor discrepancies with AWRC River Basins. This would enhance the utility of the DEM and its associated catchments and sub-catchments for continental and regional assessments. The very large numbers of small and/or ephemeral lakes and reservoirs that crossed sub-catchment divides could also be incorporated into the source data for a revised 9 second DEM to further enhance its compatibility with known drainage features. The sub-catchment definition procedure should also be enhanced to allow for areas with divergent flow.

References