Cellular Automaton Evolutionary Slope and River

 CAESAR is a cellular landscape and river reach evolution model. It allows the user to input a DEM of a river catchment or reach and enter water and sediment fluxes, or rainfall data as a basis for simulations. CAESAR represents a landscape with a mesh of grid cells. For each cell, further values are stored representing hydrological parameters, grainsize, water discharge, vegetation levels etc. Then for every model iteration, these are altered according to a set of rules, loosely grouped into 1) hydraulic routing, 2) fluvial erosion and deposition and 3) slope processes.

CAESAR can be used to simulate the impact of extreme rainfall events, since time series data for actual or simulated rainfall events can be used as input (see CAESAR simulation). This ability is a critical model attribute as the rehabilitated Ranger mine site should securely contain contaminants for long time periods without exposure to the environment. It is likely that one or more extreme rainfall events will occur during that time. Extreme event testing of the proposed design parameters for the constructed landform has assumed greater importance given the possibility of an increase in frequency of intense rainfall periods as a consequence of climate change.

 Location of catchment being modelled

Figure 1: Location of catchment being modelled

Applying CAESAR to the ERA Ranger Mine, Northern Territory

For the purposes of this study, the CAESAR
model was applied to a proposed rehabilitated of the main catchment on the proposed rehabilitated landform mine which drains what is now the tailings storage facility (Figure 1). A 25 metre resolution DEM of the proposed rehabilitated landform was supplied by ERA. Using ESRI ArcGIS™ software with the ArcHydro™ extension, the DEM of the proposed landform was hydrologically corrected and pit-filled and catchment boundaries and drainage lines on the proposed landform derived. The modelled catchment has an area of 921 hectares, and a mean slope of 1.9%.

Unlike Siberia, CAESAR can not presently simulate the effect on the hydrology and erosion of areas of different surface treatment on a landform, hence initial simulations used only the proposed tailings dam sub-catchment (TDC) surface properties. The TDC flows into Magela Creek (Figures 1) draining the rehabilitated tailings dam footprint through a channel between the capped Pit 1 and capped Pit 3. This was the largest catchment with uniform surface treatment on the proposed rehabilitated site.

Landform simulations - Data sources

The main data sources required for CAESAR are rainfall data (mm/hr) and soil particle distribution data. For this study, 21 years of complete hourly rainfall data in the period from 1972-2006 recorded at the nearby Jabiru Airport were used.

To simulate an extreme rainfall event the rainfall record for 2006-2007 was used. During late February and early March 2007 a period of exceptionally heavy rainfall occurred throughout the entire Magela Creek system as a result of a monsoon trough that extended across the Top End of the Northern Territory. Total rainfall over a three-day period between 1700 on 27 February and 1700 on 2 March 2007 at Jabiru airport was 785 mm – the largest three-day rainfall that has been recorded in this region. Figure 2 shows that rainfall was almost continuous during these 72 hours and that within this period there were four much more intense rainfall episodes. The most intense period of rainfall occurred during the morning of 1 March 2007. Maximum rainfall intensities during this period exceeded a 1 in 100 year storm event for durations between six and 72 hours.

 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.

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.

This event resulted in the highest flood levels recorded within the Magela Creek catchment since recording began in 1971. Total runoff in the main Magela Creek channel during a three-day period between 1200 on 28 February and 1200 on 3 March 2007 was greater than the mean annual runoff for the catchment. Peak flood height along the main channel exceeded the previous maximum recorded flood height on 4 February 1980 by more than 25% (Fig 2). The corresponding peak discharges along the main channel were approximately eight times the mean annual flood discharge and more than double the 1-in 100-year flood event discharge.

Model calibration

Sensitivity studies were conducted using observed rainfall to calibrate input parameter values to previously measured erosion rates for both mine disturbed areas and adjacent natural areas. Once parameter values were estimated, simulations were run using the Jabiru rainfall record; surface material size distributions for Ranger waste rock (Evans and Loch 1996); Magela Creek discharge (Q) data at gauging station G8210009 (Figure 2) and estimated sediment discharge (Qs) for Magela Creek based on continuous turbidity data; and stream sediment size distributions of Roberts (1991). The Magela Q and Qs were used as inputs into Magela Creek at the upstream boundary of the DEM to simulate the effects of throughflow in the Magela Creek reach adjacent to the study catchment and to investigate how backflow effects of Magela Creek may effect transport and deposition of sediment leaving the study catchment.

Simulation scenarios

Three simulations were run:

  1. A 21 year period using the observed hourly rainfall record and daily Magela Q and Qs for complete years between 1972 and 2006;
  2. A 21 year period as above with the 2006-07 time series for Q and Qs data inserted after 10years to assess the impact of an extreme event after the landform has reached equilibrium; and
  3. A 21 year period using the 2006-07 rainfall year to replace the first year of the 1972-06 data series used in 1 above. This was to simulate the effects of an extreme event on the landform in the wet season immediately after rehabilitation.


 Denudation rates of the TDC and  Magela reach simulated using CAESAR with various rainfall inputs

Figure 3: Denudation rates of the TDC and Magela reach simulated using CAESAR with various rainfall inputs

Annual denudations rate for each simulation are shown in Figure 3. The simulations show that there is a high level of sediment loss in the initial years as the new landform finds equilibrium. This is the phase of catchment conditioning simulated by CAESAR as fine sediment is removed from the catchment, drainage lines are incised, particle size distribution of the surface material is adjusted and vegetation grows, leaving coarser material in the thalweg of drainage lines.

Simulations show that the catchment is conditioned in about 5 years. Similar observations of catchment conditioning or surface armouring have been observed in field erosion studies at mine sites and for natural terrains in the Alligator Rivers Region (Moliere et al., 2002). After this initial 5 year phase of catchment conditioning all three simulations have similar outputs with denudation rates oscillating between positive (erosion) and negative (deposition) with annual medians of 0.01 mmy-1 to 0.07 mmy-1, and annual averages of 0.17 – 0.21 mmy-1, and ranges of -0.08 mmy-1 to 0.94 mmy-1. Negative rates indicate that Magela sediment input from upstream is greater than that exported and that deposition has occurred within the catchment.

Previous studies in the area give a range of denudation rates for waste rock of -2 mmy-1 to 7 mmy-1 with a median of 0.04 mmy-1 and for natural areas, denudation rates of -1.25 mmy-1 to 1.25 mmy-1 with a median of 0.018 mmy-1 (Erskine and Saynor 2000). The simulated denudation rates compare well with published erosion rates, since input parameter values were calibrated to the range of known erosion rates, and the outputs from the model indicate that physical processes in the catchment were adequately simulated by the model. The initial large oscillations are also to be expected as any system in disequilibrium usually oscillates to the steady state condition of a successfully rehabilitated mine. However, further research is required to confirm the magnitude of these oscillation.

 CAESAR output showing (a) sediment pulse entering the TDC; (b) entering Magela Creek; and (c) being dispersed along the channel to the outlet

Figure 4: CAESAR output showing (a) sediment pulse entering the TDC;
(b) entering Magela Creek; and (c) being dispersed along the channel to the outlet

Initial application of the 2006-07 rainfall time series (simulation 3 above) is sufficient to flush all of the mobile sediment from the catchment in the first year giving a denudation rate of 9 mmy-1, with little further conditioning taking place in the following years. If, instead, this extreme event occurs after 10 years (simulation 2), a large pulse of sediment is exported with a denudation rate of 5.58 mmy-1 with erosion rates returning to ‘normal’ after a year.

This is only an initial assessment of the effects of the impacts of an extreme rainfall event on the TDC. It should be noted that although this study includes the Magela Creek reach and surrounds, the model currently treats all areas of the catchment as being covered with waste rock material. It does not account for the different morphologies, sand bed channel or surface condition and vegetation abundance of the riparian zone and adjacent woodlands. However, the results to date indicate that CAESAR has very good potential for simulating the effects of extreme events on landform stability and sediment export from the mine site to natural areas of the catchment. The CAESAR interface allows the user to visualise how a sediment pulse moves through a system. The series in Figure 4 shows a pulse moving along the channel between pits 1 and 3, entering the main Magela channel, and then being dispersed downstream along the channel. Below you can see a simulation of sediment movement as simulated by the CAESAR model.

Flash video showing CAESAR computer simulation of sediment pulse moving through the system.

Summary and conclusions

Soil erosion and landscape evolution models offer the ability to better understand erosion and hillslope processes that will occur on rehabilitated mine landforms and allow better design and management to reduce the potential impact of these processes. Numerical modelling is the only quantitative method with which predictions can be made about landuse and climate change variability on catchment processes. It is important that these models and modelling procedures be evaluated, and that model results be validated against field data.

The model simulation results reported here compare well with independent field data determined for the region. A major advantage of CAESAR is that it uses data that may be already available eg hourly rainfall and particle size distribution, making it easier to run. However, because the outputs from CAESAR can be produced hourly, a long time is required to run simulations for simulated long time periods due to the computer processor power required. However, CAESAR has not yet been fully tested for this application of assessing the stability of post-mining landforms and deeper understanding of the catchment conditioning phase and how this relates to reality is required. In particular, the capability to simulate variation in surface treatments across a catchment is required before this model can adequately simulate the behaviour of a rehabilitated mine site in the context of its surroundings is being developed.


Erskine, W.D. and Saynor, M.J. 2000. Assessment of the off-site geomorphic impacts of uranium mining on Magela Creek, Northern Territory, Australia. Supervising Scientist Report 156, Supervising Scientist, Darwin NT.

Evans, K.G. and Loch, R.J.  1996. Using the RUSLE to identify factors controlling erosion of mine soils.  Land Degradation and Development, 7:267-277

Moliere D.R., Boggs G.S., Evans K.G., Saynor, M.J. and Erskine,W.D. 2002. Baseline hydrology characteristics of the Ngarradj catchment, Northern Territory, Supervising Scientist Report 172, Supervising Scientist, Darwin NT, 2002.

Roberts, R.G. 1991. Sediment budgets and Quaternary history of the Magela Creek catchment, tropical northern Australia. PhD Thesis, Department of Geography, University of Wollongong.