2.1.1.1 Observed data

2.1.1.1.1 Physical geography

Digital elevation model

A digital elevation model (DEM) is needed in both the groundwater model and the landscape water balance model (Australian Water Resources Assessment (AWRA) landscape model, AWRA-L) to define surface topography for representing hydraulic gradients, flow directions and defining contributing areas. The DEM (Bioregional Assessment Programme, Dataset 1) was obtained from 3-second (~90 m) resolution data from the Shuttle Radar Topography Mission (SRTM), flown in February 2000 (Farr et al., 2007). The technique, known as interferometric synthetic aperture radar, uses two radar antennas, positioned 60 m apart, to acquire pairs of images of Earth, from which phase difference measurements can be calculated to compute elevations at Earth’s surface. The positional accuracy (x, y) of the SRTM data are in the order of 10 m (Smith and Sandwell, 2003; Rodriguez et al., 2006). For Australia, these data were processed according to Gallant et al. (2011) and the elevational accuracy (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).

The groundwater modelling domain extends offshore, beyond the area covered by the 3-second DEM. To extend the digital elevation model to include the topography of the ocean floor, the 9‑second Australian Bathymetry and Topography Grid produced by Geoscience Australia (2009) (Dataset 2) was resampled to 3 seconds and stitched onto the SRTM DEM. This has been generated from a number of different datasets, including single beam and multi-beam bathymetry data, laser airborne depth sounder data, satellite measurements and various topographic and DEM data. Each dataset was gridded at 9 seconds without smoothing, and overlayed with the most accurate grid on top and the least accurate grid at the bottom.

The combined grid of the groundwater modelling domain for the Hunter subregion is a derived product of the Bioregional Assessment Programme (Dataset 3).

Vegetation height

The groundwater model uses information derived from vegetation height to differentiate between deep-rooted and shallow-rooted vegetation. While there can be enormous variability in the relationship between vegetation height and root depth due to site-specific factors such as soil depth, nutrient availability and water availability, observation data indicates that trees generally have deeper root systems than shrubs and herbaceous plants. For example, Canadell et al. (1996) compiled and analysed a dataset containing 290 observations of maximum rooting depth from major terrestrial biomes across the globe and found that the mean and standard error of maximum rooting depths for trees, shrubs and herbaceous plants were 7.0±1.2 m, 5.1±0.8 m and 2.6±0.1 m, respectively.

Vegetation height data for the groundwater modelling domain were obtained from a global 1 km grid resolution vegetation height dataset (Caltech/JPL, Dataset 4). This dataset was derived from measurements made by a satellite-based light detection and ranging system (lidar) between 20 May 2005 and 23 June 2005 using the Geoscience Laser Altimeter System (GLAS) aboard ICESat (Ice, Cloud and Land Elevation Satellite). Using a regression tree approach, Simard et al. (2011) modelled overstorey vegetation height globally 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.

The model uses this vegetation height surface to define a spatially-variable evapotranspiration extinction depth (i.e. the depth below which water cannot be extracted from the soil via plant roots) that is used in scaling potential evapotranspiration (PET) to actual evapotranspiration (AET). The evapotranspiration extinction depth parameter is a fixed parameter in the groundwater modelling. Specification of this parameter is likely to have a moderate impact on model predictions (see Section 2.6.2.8 of companion product 2.6.2 for the Hunter subregion (Herron et al., 2018)).

Land use

The AWRA-R river model needs details of irrigated areas and crop types along each section of the river network in order to distribute non-spatial irrigation diversion data appropriately in the model. Land use data were clipped using reach boundaries to derive irrigated areas and crop types. Land use data for the Hunter subregion were obtained from the Catchment Scale Land Use Management (CLUM) raster surface compiled November 2012 (ABARES, 2012) (Dataset 5). 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 November 2012 dataset is a combined 50 m raster for Australia, with edge-matching errors corrected for NSW (for which there were no new data provided compared to the previous version). The land use surface is based on data collected between 1997 and 2009, from sources ranging from 1:25,000 to 1:250,000 in scale. Section 2.1.4.2.3 contains the specific details for representing areas of irrigation in the river modelling.

2.1.1.1.2 Climate

For retrospective climate analysis the following variables are required for hydrological modelling of the Hunter subregion: (i) precipitation (P), (ii) maximum and minimum air temperature (Tmax and Tmin, respectively) and (iii) net radiation (Rn). Gridded climate surfaces (Bureau of Meteorology, Dataset 6) have been used in the calculation of PET and catchment runoff. This dataset contains daily data for the aforementioned climate variables for the whole Australian continent at a 0.05 degree (or ~5 km) grid cell resolution from 1 January 1900.

Precipitation

Daily precipitation grids (Bureau of Meteorology, Dataset 6) are generated by the Bureau of Meteorology using optimal geostatistics techniques, which take account of elevation, to interpolate daily and monthly station P totals between isolated stations (Jones et al., 2009). Daily time step data were used as input to surface water modelling, with groundwater models using monthly input data. Given that precipitation is the most spatially discontinuous meteorological process, it is the on-ground observation network that has the highest 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 precipitation 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 precipitation 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). These errors will have little impact on prediction of ACRD changes, which are calculated as the difference between the CRDP and baseline futures and use the same rainfall time series. However, the errors can impact on model calibrations, as the quality of the input data is fundamental to obtaining an acceptable calibration.

Temperature

Daily maximum temperature (Tmax) and minimum temperature (Tmin) grids (Bureau of Meteorology, Dataset 6) are generated by the Bureau of Meteorology using optimal geostatistics techniques that take elevation into account (the environmental lapse rate) to interpolate daily extremes of air temperature measured at isolated stations (Jones et al., 2009). 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).

Solar radiation

Daily solar radiation (Rn) data are available from 1900 onwards as part of the Bureau of Meteorology gridded climate surfaces for Australia (Bureau of Meteorology, Dataset 6). The dataset comprises two distinct periods: post-1982, daily solar radiation values are based on observations from ground-based and satellite instruments; prior to 1982, the daily values are based on the long-term climatologies from the post-1982 period. This means, for example, that the solar radiation on 1 January is the same for every year from 1900 to 1981 and reflects the average solar radiation on 1 January in the years since 1981. Uncertainties in the solar radiation data arise from the effect of cloud cover (~5%) and water vapour in the atmosphere (~2%). Comparisons with ground-based measurements (made with pyranometers) indicate that satellite methods tend to slightly over estimate the radiant exposure in wet, cloudy conditions and to under estimate in dry conditions (Bureau of Meteorology, 2016).

Last updated:
18 January 2019
Thumbnail of the Hunter subregion

Product Finalisation date

2018

ASSESSMENT