2.1.1.1 Observed data

2.1.1.1.1 Physical geography

Digital elevation model

The digital elevation model (DEM) was obtained from the 3 arc-second (~90 m resolution grid cell) Shuttle Radar Topography Mission (SRTM) (Farr et al., 2007). Dual radar antennae acquired interferometric synthetic aperture radar (InSAR), using phase difference measurements derived from the two radar images to measure topography (originally acquired onboard the NASA Space Shuttle Endeavour during its mission between 11 and 22 February 2000, when it measured the Earth’s surface elevation between 60° N and 56° S latitudes). The positional accuracy (often represented as X and Y) of the SRTM data are in the order of 10 m, as reported by Smith and Sandwell (2003, Section 4.3) and Rodriguez et al. (2006, p. 257). For Australia, these data were processed according to Gallant et al. (2011) and the levational accuracy (often represented as Z) of the SRTM DEM compared to 1198 permanent survey mark (PSM) data points had a mean error of –0.539 m. The absolute accuracy of the DEM was 14.54 m at the 95th percentile with a root mean square error (RMSE) of 7.029 m in open, flat terrain. Ninety-nine percent of points are within a height difference of less than 29.97 m (Gallant et al., 2011, p. 63–64).

Surface watercourses

Surface watercourses were defined using the GeoData Topo 250K Series 3 Topographic Data – a vector representation of the major features appearing on 1:250,000 scale NATMAP topographic maps published by Geoscience Australia (2006). Using the hydrology theme from this dataset, major and minor watercourses are identified and both, as appropriate, used to describe the surface hydrology of the Gloucester subregion. Surface water basins or catchments are defined using the Australian Hydrological Geospatial Fabric (Geofabric), a specialised geographic information system published by the Bureau of Meteorology (2012). The Geofabric registers the topology between important hydrological features such as rivers, water bodies, aquifers and monitoring points, and information about surface water basins and catchments.

Physiographic classes

The physiographic classes were obtained from the Australian Soil Resource Information System (ASRIS) (Pain et al., 2011). The following description is derived from this reference. These classes are based on a visual interpretation of landforms as expressed on the SRTM DEM. Apart from its descriptive role, a map of physiographic regions provides a regional system of reference for geomorphological and related physical geographical accounts. Through the groupings of physiographic regional characteristics at different levels, the action of underlying controls (for instance geology or climate) may be made apparent. These data have an Australia-wide coverage.

Soil classes

Soils classes used the Australian Soil Classification (ASC) system which is a product from ASRIS (2011). National soil data was provided by the Australian Collaborative Land Evaluation Program (ACLEP), endorsed through the National Committee on Soil and Terrain (NCST) (ACLEP, 2014). These data have an Australia-wide coverage and are underpinned by a collation of the best available nationally consistent soils data and information. Usually these data are the most dominant soil in a polygon, not the only soil in a polygon.

Pre-European vegetation

Pre-European (1788) vegetation data was sourced from Carnahan (1976) and Australian Survey and Land Information Group (AUSLIG, 1990). The following description is derived from these references. A reconstruction of natural vegetation of Australia is shown as it probably would have been in the 1780s. Generally, the minimum mapping unit is 30,000 ha, but in some cases smaller areas of significant vegetation, such as rainforest, are also mapped. Attribute information includes growth form of the tallest and lower stratum, foliage cover of the tallest stratum and the dominant floristic type. These data are provided at a map scale of 1:5 million and have an Australia-wide coverage.

Current vegetation

Current major vegetation types were obtained from the National Vegetation Information System (NVIS), a comprehensive data system that provides information about the extent and distribution of vegetation types in Australian landscapes published by SEWPaC (2012). The following description is derived from this reference. This dataset (v4.1) contains the latest summary information (November 2012) about Australia’s present (extant) native vegetation, which has been classified into major vegetation groups (MVGs) and major vegetation subgroups (MVSs). Many state and territory vegetation mapping agencies supplied new information to the NVIS v4.1 from 2009 to 2011, however for NSW, NVIS data was only partially updated from 2001 to 2009, with extensive areas of 1997 data remaining from earlier versions. NVIS v4.1 identifies 85 MVSs summarising the type and distribution of Australia's native vegetation. The classification contains an emphasis on the structural and floristic composition of the dominant stratum (as with MVGs), but with additional types identified according to typical shrub or ground layers occurring with a dominant tree or shrub stratum. In a mapping sense, the subgroups reflect the dominant vegetation occurring in a map unit from a mix of several vegetation types. Less-dominant vegetation groups which are also present in the map unit are not shown. It is in Albers equal area projection with a 100 m resolution (1 ha) grid size.

Current land cover

Land cover was derived from MODIS (or Moderate Resolution Imaging Spectroradiometer) satellite imagery. Specifically the Normalised Difference Vegetation Index (NDVI) (a simple graphical indicator that can be used to analyse remote sensing measurements) is rescaled to percent green vegetation cover and this is temporally filtered into the persistent and recurrent components (Donohue et al., 2009a). The MODIS NDVI imagery has a spatial resolution of 250 m, has global coverage, and is available monthly from February 2000 onwards, with the persistent-recurrent processing being performed Australia-wide. The accuracy of this is likely to be in the order of 5 to 10% (Donohue, 2014, pers. comm.).

Current vegetation height

Vegetation height was measured using a satellite based light detection and ranging system (lidar) between 20 May 2005 to 23 June 2005 using the Geoscience Laser Altimeter System (GLAS) aboard ICESat (Ice, Cloud and Land Elevation Satellite). Using a regression tree approach to model canopy height, Simard et al. (2011) were able to globally model overstorey vegetation height at 1 km spatial resolution with a vertical RMSE of 4.4 m and coefficient of determination (r2) of 0.7 when compared against 59 flux-tower field observations globally.

2.1.1.1.2 Human geography

For human geography the main datasets used are: (i) population density and (ii) land use management. A brief outline of each follows.

Population density

Human population information for the Gloucester subregion was derived from the 2011 Australian Census (ABS, 2011). An estimate of 5000 people living in the subregion was determined by intersecting the subregion boundary with the 2011 Australian Census ‘mesh blocks boundaries’ and population counts. The accuracy of this is likely to be in the order of 1 to 2%. This is as the Gloucester subregion boundary does not exactly match the 2011 Australian Census mesh blocks so intersection was needed away from the dense population centres of Gloucester and Stroud to provide this estimate. This estimate is as accurate as can be performed.

Land use management

Catchment Scale Land Use Management (CLUM) compiled November 2012 (data ranges from 1997 to 2009, scale ranges from 1:25,000 to 1:250,000) was obtained from the Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES, 2012). The most current catchment scale land use dataset for Australia has been compiled using nationally agreed land use mapping principles and procedures of the Australian Land Use and Management (ALUM) Classification version 7. The land use datasets were collected as part of state and territory mapping programs and the Australian Collaborative Land Use and Management Program (ACLUMP). The updated dataset is a combined 50 m raster for Australia, with edge-matching errors corrected for NSW (for which there was no new data provided compared to the previous version).

2.1.1.1.3 Climate (retrospective)

For retrospective climate analysis the following variables were analysed: (i) precipitation (P), (ii) maximum and minimum air temperature (Tmax and Tmin, respectively), (iii) vapour pressure (VP), (iv) net radiation (Rn), (v) wind speed, and (vi) potential evapotranspiration (PET). All these variables are national Australia-wide grids with a 0.05 degree (or ~5 km) grid cell resolution at a daily time step. They come from various sources and have different start dates. They are briefly dealt with in turn in the following paragraphs.

Precipitation

Daily and monthly P grids are available from 1900 onwards and were generated by the Bureau of Meteorology (Jones et al., 2009) by using optimal geostatistical techniques, taking elevation into account, to interpolate daily and monthly station P totals measured at isolated stations. Daily time-step data are used as input to surface water modelling (see companion product 2.6.1 for the Gloucester subregion (Zhang et al., 2018)), with groundwater models using monthly input data (see companion product 2.6.2 for the Gloucester subregion (Peeters et al., 2018)). Given that precipitation is the most spatially discontinuous meteorological process, it is the on-ground observation network that has the high spatial density of observations (Jones et al., 2009, Figure 2). Jones et al. (2009) fully cross-validated the estimates for the seven years from 2001 to 2007 by randomly deleting 5% of the stations in the network, performing an analysis using the remaining 95% of station observations and then calculating the analysis errors for the omitted stations. Between 2001 and 2007, the Australia-wide mean daily P was 1.8 mm/day with a RMSE of 3.1 mm/day (Jones et al., 2009, Table 3b). This represents a relative error of 172% (calculated as RMSE/mean), although absolute differences may be small. For 2001 to 2007, the Australia-wide mean monthly P was 54.3 mm/month with a RMSE of 21.2 mm/month (Jones et al., 2009, Table 3a). This represents a relative error of 39% (calculated as RMSE/mean).

Air temperature

Daily Tmax and Tmin grids are available from 1900 onwards and were generated by the Bureau of Meteorology (Jones et al., 2009) by using optimal geostatistics techniques, taking elevation into account (the environmental lapse rate), to interpolate daily extremes of air temperature measured at isolated stations. The mean daily Tmax and mean daily Tmin for Australia between 2001 and 2007 were 24.9 and 12.8 °C with RMSE statistics of 1.2 and 1.7 °C, respectively (Jones et al., 2009, Table 2b). These represent relative errors of 5 and 13%, respectively (calculated as RMSE/mean). The mean monthly Tmax and mean monthly Tmin for all Australia between 2001 and 2007 were 24.9 and 12.7 °C with RMSE statistics of 0.7 and 1.0 °C, respectively (Jones et al., 2009, Table 2a). These represent relative errors of 3 and 8%, respectively (calculated as RMSE/mean).

Vapour pressure

Daily VP data, also generated by the Bureau of Meteorology (Jones et al., 2009), are available from 1950 onwards and are recorded at two times of the day, 9 am and 3 pm, both local times. The same optimal geostatistics techniques (as used for P, Tmax and Tmin above) were used to spatially interpolate VP measurements made at the isolated stations. Between 2001 and 2007, the mean daily VP Australia-wide was 13.7 hPa at 9 am and 13.1 hPa at 3 pm, with RMSE statistics of 1.8 and 2.5 hPa, respectively (Jones et al., 2009, Table 4b). These represent relative errors of 13 and 19%, respectively (calculated as RMSE/mean). Between 2001 and 2007, the mean monthly VP Australia-wide was 13.7 hPa at 9 am and 13.1 hPa at 3 pm, with RMSE statistics of 1.1 and 1.7 hPa, respectively (Jones et al., 2009, Table 4a). These represent relative errors of 8 and 13%, respectively (calculated as RMSE/mean).

Net radiation

Daily Rn is generated by CSIRO Land and Water using a combination of gridded meteorological data and satellite data (Donohue et al., 2010). This is available from 1982 onwards, due to use of satellite based albedo (the colour of the land surface, defining how much sunlight is reflected) in the outgoing shortwave radiation calculations. The incoming shortwave and longwave components have been validated and at a monthly time step have RMSE values of 18 and 9 W/m2 (Donohue et al., 2009b, Figure 5b and Figure 5d, respectively). The outgoing shortwave and longwave components utilise time series remotely sensed imagery, and thus capture the true dynamics of the land surface (as opposed to other methods that use long-term climatologies).

Wind speed

Daily mean wind speed is also generated by CSIRO Land and Water (McVicar et al., 2008) from 1975 onwards using daily wind-run observations made at the Bureau of Meteorology network of anemometers. These are quality controlled and then used as input to a tri-variate thin-plate spline as a function of longitude, latitude and distance inland from the coast (McVicar et al., 2008). Importantly, these grids capture the ‘stilling’ process (declining wind speeds) that has been observed at many terrestrial locations across the globe which is partly responsible for reducing rates of evaporative demand (Donohue et al., 2010; McVicar et al., 2012a). The RMSE of monthly wind speed is 0.32 m/s (Donohue et al., 2009b, Figure 2e).

Potential evapotranspiration

PET, a measure of the ‘drying power’ of the atmosphere, is calculated using the fully physically based Penman formulation and hence uses all of the previously mentioned meteorological variables. It is calculated per Donohue et al. (2010) and is available from 1982 onwards. Being a ‘potential’ means that direct validation of PET is not possible, however, when assessing trends of this physically based PET formulation with other PET forms, Donohue et al. (2010) showed that the Penman formulation was most optimally able to respond in a complementary manner to monthly P trends (Donohue et al., 2010, Table 4).

In summary, all the data sources mentioned in this section provide the best gridded estimates of retrospective climate data available for the Gloucester subregion.

2.1.1.1.4 Climate (prospective)

For prospective climate analysis, Post et al. (2012) assessed changes in P and PET using output from 15 GCMs (global climate models) and reported changes for large basins such as the Manning River and Karuah River. Specifically they used GCMs from the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC, 2007, hereafter referred to as IPCC AR4) and used the IPCC A1B global warming scenario output to transform historical daily climate records to provide future daily climate projections of P and PET that can be used in a rainfall-runoff model. Compared to the global mean temperature in 1990, this scenario indicates a global temperature that is 1 °C higher in 2030 and 2 °C higher in 2070. This scenario is based upon: (i) very rapid economic growth, (ii) with global populations peaking mid-century and declining thereafter, and (iii) the rapid introduction of new and more efficient technologies with a balance across all energy sources (IPCC, 2007).

Last updated:
26 October 2018
Thumbnail of the Gloucester subregion

Product Finalisation date

2018