Assessing mining-related impacts on stream mud loads downstream of Ranger
Fine suspended-sediment (mud) moves through stream systems in pulses or waves generated by rainfall discharge (stream-flow) events. Reliable impact assessment requires knowing how much mud is transported during these pulses (event mud loads). For Ranger mine, two techniques are being developed for impact assessment based on event mud loads measured during a rainfall-runoff event – an experimental contrast and a regression model. The experimental contrast is a Before-After-Control-Impact, paired difference design (BACIP), where downstream and upstream (of a catchment disturbance e.g. Ranger mine) data collection sites are treated as paired sites. The mud load measured downstream of the disturbance is compared to that measured upstream on an event basis to assess impact. The regression model approach is a site-specific relationship fitted between event mud load and corresponding event discharge characteristics. In this case, the event mud load is compared against the discharge (stream flow) characteristics of that rainfall event to assess impact, rather than against the load measured upstream. It is recommended that impact assessment of the mud load within a catchment should comprise a combination of BACIP and regression relationship techniques. Using event mud load data collected upstream (Magela Creek upstream (MCUS) and GCUS (Gulungul Creek upstream)) and downstream (Magela Creek downstream (MCDS) and Gulungul Creek downstream (GCDS)) of Ranger along Magela and Gulungul Creeks respectively, the impact of both mining-related and natural events on fine suspended sediment in these creeks can be assessed.
Figure 1 Monitoring sites in Magela Creek
Trigger levels for impact assessment
The Supervising Scientist uses trigger levels to assess impact. These trigger levels are the 'focus', 'action' and 'limit' and when a measured property e.g. mud load reaches a value equivalent to the trigger levels investigative action is instigated depending on the level i.e. 'focus', 'action' or 'limit'. The value of the trigger levels for a specific contaminant can be determined in the laboratory through testing the effect of the contaminant on the health of aquatic animals and plants. This testing is complex and where such testing has not been done, interim trigger levels can be determined based on the statistical distribution of the baseline concentrations of the measured contaminant.
For BACIP analysis, when mud load data from the upstream (Control (C)) site are subtracted from those at the downstream (Impact (I)) site, a set of paired-site (P) 'difference' values are derived. Statistical percentiles ('trigger' values) of the difference data are derived and event mud load data are compared to these values. BACIP analysis is based on a comparison of differences of log-transformed event mud loads at the upstream and downstream stations along Magela and Gulungul Creeks (Figure 1).
Statistically derived trigger levels associated with the mud load-discharge relationship are used to compare and assess observed event mud load with predicted mud load determined from corresponding discharge characteristics. The most significant discharge characteristics for predicting event mud load within the region are total event runoff and maximum periodic rise in discharge. The form of the regression equation is as follows:
Total mud load = K(QT)a Rib (1)
where QT is total discharge during the mud pulse, Ri is maximum periodic rise in discharge over 6 minutes and a, b and K are fitted parameters.
Event mud load data collected during 2005 were used to establish preliminary trigger values for the event-based BACIP analysis. Given that the log-transformed difference data are a non-normal distribution, trigger levels are the 80th, 95th and 99.7th percentiles of the data, which correspond to the 'focus', 'action' and 'limit' triggers, respectively. Events that lie above the 95th percentile ('action') are investigated to determine the cause of the elevated mud load measured downstream relative to the load upstream.
The event load data for the 2007–08 wet season were compared against the preliminary trigger levels derived from 2005–06 and 2006–07 data for both Magela and Gulungul Creek. Events where mud loads measured downstream are significantly elevated compared to the load upstream (i.e. that lie above the 95th percentile ('action')) are shown in Figure 2.
Figure 2: Hourly rainfall data collected at Jabiru airport during the flood event. The flood height at G8210028, a station along the Magela Creek main channel, is also shown.
Relationship between mud load and discharge characteristics
Event data have been used to fit statistically significant regression relationships between event mud load and corresponding event discharge characteristics for each site of the form given in Equation (1). Given that observed loads are normally distributed around the best-fit line (predicted loads), trigger levels are +1 SD, +2 SD and +3 SD from the 1:1 line. Similar to the BACIP analysis above, these correspond to the 'focus', 'action' and 'limit' triggers, respectively. It is considered that events that lie above the +2 SD line ('action') have a relatively high mud load compared to that predicted using the corresponding event runoff characteristics. Using the regression model approach, an impacted event is one in which:
- the mud load measured downstream of Ranger is significantly elevated compared to the corresponding event discharge characteristics (i.e. lies above the + 2 SD line), and
- the corresponding event mud load measured upstream of Ranger is not significantly elevated compared to the discharge characteristics (i.e. lies within + 2 SD of the fitted relationship).
Figures 3 and 4 show the event mud loads observed during the 2007–08 wet season at Magela and Gulungul Creek, respectively, plotted against the predicted event mud loads using the fitted relationships. The impacted events are also identified in Figures 3 and 4. It is considered that during these events an impact has occurred between the upstream and downstream stations (possibly mining-related) resulting in the elevated mud load observed downstream.
Figure 4 Event-based mud load relationships for GCDS (Left) and GCUS (Right). Event data collected during 2007-08 are plotted against the trigger levels derived using the 2003-07 data. (Note: no mud load data were collected at GCUS during the event on 25 Apr 06.)
Application to impact assessment
A combination of BACIP and regression model techniques should be used for impact assessment on mud load within the Magela Creek catchment. For example, an impacted event initially identified by BACIP should be further tested against the event mud load-discharge relationship fitted for the downstream site. Using such an approach, an impacted event is one with a significantly elevated mud load compared to both the mud load measured upstream and the corresponding event discharge characteristics observed at the downstream site. Data collected within the Magela Creek catchment since 2003 show that there have been three impacted events observed downstream of Ranger (two at MCDS and one at GCDS), and these are discussed briefly below. All three events occurred during the 2005–06 and 2006–07 wet seasons.
Magela Creek – March 2007 flood
This extraordinary rainfall–runoff event, estimated at >>1 in 100 y event caused widespread flooding (Figs. 5 & 6) and geomorphic impact (Fig 7) in the catchment. The standard sediment management controls implemented on the minesite and the drilling area were simply overwhelmed as a result of record rainfall over the region. The subsequent input of poor quality water from the ore stockpiles into the minesite tributaries and the input of fine suspended sediment into Magela Creek from the exploration drilling area during this event are likely to have contributed to the elevated mud concentrations recorded downstream of Ranger.
Figure 5b Gauging station G8210009 along Magela Creek post-flood on 6 March
Figure 5a Gauging station G8210009 along Magela Creek during the flood on 1 March (1m short of peak flood height on 2 March)
Location of the rain gauge at GCDS. The main creek channel is behind the treeline approximately 20m from the rain gauge. Taken the week after the major flood
Figure 7 One of the fourteen landslips that occurred in the upper Magela catchment during the March 2007 extraordinary rainfall runoff event
Magela Creek – Cyclone Monica (25 April 2006)
Remotely sensed data of the Magela Creek catchment was used to assess the impact of the cyclone on tree canopy cover. The data showed that the cyclone had a relatively large impact on the mine site catchments downstream of MCUS compared to the area upstream of MCUS. The upheaval of soil by treefall along the channel banks in this area (as a result of high wind velocities at a time when the soil was saturated towards the end of the wet season) would have been expected to have resulted in increased stream mud concentrations during the event.
It is worth noting that substantial treefall occurred throughout Gulungul Creek catchments as a result of this cyclone. However, unlike the damage in the Magela catchment, field observations confirmed that treefall in this catchment was catchment-wide. The mud load measured at the downstream station along Gulungul Creek was significantly elevated during this event compared to the corresponding event discharge characteristics (Figure 4). Despite the fact that, in this case, turbidity/mud concentration data were not recorded upstream at GCUS because of cyclone damage to the gauging station equipment, it is considered likely that the upstream mud load should also have been elevated relative to event discharge characteristics.
Gulungul Creek – 23 February 2007
Prior to the 2006-07 wet season, construction works commenced to elevate the tailings dam wall for increased storage capacity. Therefore, it is possible that during this event, the relatively intense rainfall over the exposed soil associated with earthworks around the perimeter of the tailings dam, contributed to elevated mud concentrations in the ungauged, mine-impacted tributaries, which originate from the base of the tailings dam wall and enter the main Gulungul Creek channel between GCUS and GCDS. Consequently, the input of mud into these tributaries may have attributed to an elevated mud load recorded at the downstream station.
2007–08 event data
During the 2007–08 wet season, one mud load event measured at MCDS and two at GCDS were elevated above the 'action' trigger level using the BACIP approach (Figure 2). That is, these events had a significantly elevated mud load measured downstream of Ranger compared with that upstream. However, in all three cases, the events were within the 'action' trigger level derived from the regression model approach, which suggests that the event mud loads measured at the downstream stations are reasonable for the corresponding flow conditions. Therefore, these events should not be considered as 'impacted' events.
The event on 4 February 2008 at GCDS, however, was close to the 'action' trigger level derived from the regression model approach (Figure 4). During 2007–08, construction works were continuing at the tailings dam wall and, therefore, similar to the event on 23 February 2007, it is possible that intense rainfall over the exposed soil associated with earthworks around the perimeter of the tailings dam, contributed to elevated mud concentrations in the ungauged, mine-impacted tributaries (which enter the main Gulungul Creek channel downstream of GCUS). Consequently, the input of mud into these tributaries may have attributed to a higher than expected mud load recorded at the downstream station.
Moliere DR & Evans KG 2010. Development of trigger levels to assess catchment disturbance on stream suspended sediment loads in the Magela Creek catchment, Northern Territory, Australia. Geographical Research, DOI: 10.1111/j.1745-5871.2010.00641.x (published online 24 Feb 2010)