Hydrologic Reference Stations

Introduction

Hydrologic Reference Stations (HRS) was released to the public in June 2013 with available data to December 2012. It was updated in October 2015, with data to December 2014. Updating of the service is generally done regularly.

The 2020 update includes:

• Revision of station selection guidelines;
• A review of existing locations to ensure these locations continue to satisfy HRS selection guidelines;
• Search and include new stations that satisfy selection guidelines;
• Quality control and update of gap-filled continuous data to February 2019;
• Recalculation of trend analyses;
• Update of website with new products, functionality and information.

Inclusion of new stations

There are about 4,800 streamflow gauging stations across Australia. Many of these stations previously had insufficient duration of data or unavailability of quality checked data for inclusion in the HRS, but may now be available for consideration with sufficiently long streamflow records of quality checked data.

A set of 780 stations were initially selected from these 4,800 stations across Australia, using selection guidelines to identify unregulated and unimpaired catchments with sufficient length of record (Zhang et al., 2013). In search of new suitable stations for HRS, these 780 unregulated and unimpaired catchments, widely spread across Australia were identified, after performing strict quality assurance and quality control, including quality code check for daily streamflow dataset. A review of these catchments identified those that satisfy the additional HRS selection guidelines of minimum of 30 years of continuous data with less than 5% of missing data. 175 of the existing HRS sites are found in these 780 catchments and an additional 340 gauges that initially satisfied the HRS selection guidelines were chosen for further investigations and testing.

Measurement uncertainty

For 171 existing HRS gauging stations, McMahon and Peel (2019) assessed the uncertainty in streamflow considering the gauged range of the stage–discharge relationship for 622 rating curves (for these gauges). They found that estimated flow volumes beyond the gauged range of the rating curve occurred and leads to measurement uncertainty. In the current HRS update (Fig.1), this uncertainty is provided as the percentage of flow volume flowing above the recorded maximum gauged discharge.

Gap-filling and quality control

In HRS, sites with generally less than 5% missing data across the entire record are considered. A quality-assurance, quality control (QA/QC) process was applied to observed time-series of daily streamflow from each gauging station. This process identified and removed erroneous data values such as negative and extreme values, and periods of long linear interpolation. The process of detection and removal was automated and then checked manually.

The GR4J model (Perrin et al., 2003) was adopted to infill missing data. As part of this process, a simple error correction procedure was used to ensure the initial and final estimated flows matched the adjacent observed values. This was done through linearly interpolating the start and end of the infilled period to the observed flows.

Revision of station selection guidelines

A quantitative evaluation of gap-filled data accuracy was assessed by Zhang and Post (2018). That study, in considering 217 unregulated catchments spread across Australia, clearly showed that, examining trends at the annual scale, gap-filled data are most accurate when the missing data make up less than 5 % of the total record. Results clearly indicated that gap filling of Australian streamflow data using hydrological models is very reasonable when the missing data is less than 10%, with only a small number of catchments showing a large trend bias when the missing data is up to 20%. Even when the percentage of gap-filled data is within these limits, the percentage of infilled volume may be large when most of the missing data represent high flow volumes, and result in higher uncertainty in the data. Though the percentages of days of gap-filling suggested by Zhang and Post (2018) are not directly comparable to the percentage of flow volume that is infilled, these percentages provide an upper threshold of potentially acceptable uncertainty of 10%.

Therefore, in assessing the adequacy of the stations for inclusion in HRS update, two further factors are jointly considered:

1. the percentage of flow volume included as infilled data
2. the percentage of flow volume above the maximum gauged discharge.

Therefore, along with the current HRS guidelines, an additional criterion of a threshold maximum of 10% for flow volume of infilled data and a restriction in extrapolated data to a maximum of 25% have been included for selection of stations for the update.

Review of the existing 222 stations

Decommissioned stations

There are several stations with limited or no additional data from 2009 onwards. These stations were decommissioned by the data providing agencies. These are 12 stations in all – 5 from Northern Territory, 2 from Victoria, 2 from Western Australia, 1 from New South Wales and 2 from South Australia. Out of these 12 stations, 4 didn't pass the additional selection guidelines. Data QA/QC and trend analyses are still provided for the remaining 8 sites. However, station status is shown in "Station Facts" as 'Decommissioned' in the website.

Stations excluded

There are stations that were included in HRS but may no longer satisfy the selection guidelines due to introduction of regulation structures upstream or due to some of the additional selection guidelines.

Investigation on the river network of the catchment on Reedy Creek at Wangaratta North (403209) in 2018 showed that the contributing area of flow through this station varied widely between 368 km² to 5506 km² due to several bifurcation points in the river network. Poor quality of data at Trephina Creek at Trephina Gorge (G0060005) in Northern Territory is evident with 91% of the data currently considered missing due to gauge levels exceeding the ratings table. These two stations do not satisfy the HRS selection guidelines and have been excluded.

Information on the gauging used for developing rating curves for most of the 222 stations previously chosen were unavailable at the time of initial selection in 2012. This information is now available, and all 222 stations were evaluated against the additional selection criteria (threshold maximum of 10% for flow volume of infilled data and a restriction in extrapolated data to a maximum of 25%). 171 existing stations and 8 decommissioned stations met these thresholds. Therefore, 41 existing stations have been excluded on this basis.

A total of 43 stations from the existing 222 have been removed from HRS in this update.

Final number of stations

Out of the 340 new gauges that initially satisfied the HRS selection guidelines, 288 satisfied the additional criteria (threshold maximum of 10% for flow volume of infilled data and a restriction in extrapolated data to a maximum of 25%). Thus, the total number of streamflow gauging stations has increased to 467 (Table 1).

Table 1: Total number of stations

Station features	Selection guidelines	Numbers
Existing HRS stations decommissioned	Passed	8
Existing HRS stations operational and data up to February 2019	Passed	171
Existing HRS stations operational and decommissioned	Failed	43
New gauging stations operational and data up to February 2019	Passed	288
Total HRS stations	Passed	467

Figure 1: Snapshot of the updated HRS web portal

Trend analyses

The non-parametric Mann-Kendall (MK) trend test was used, as it was done before, to detect the direction and significance of the monotonic trend. The most popular non-parametric technique for estimating a linear trend namely, the Sen slope (Sen, 1968) was used to represent the magnitude of the trend as shown in Fig.1.

Web portal content

The content of the web portal was updated along with trend analyses products and tables:

• A summary of the 2020 update was included in the Introduction page;
• Three items in FAQ page were updated in accordance with Australian Water Information Dictionary;
• The 'Sen slope' trend analysis was included in the Methods page;
• Reference and Papers page was updated to include recent publications;
• Contact details of the data owners were updated in the Copyright page;
• Feature Stations were updated to include the recent data;
• Selection Guidelines page was updated to include the additional criteria incorporated in this update.