The 6-year erosion field study described in this thesis is unique since no similar study has been conducted on coalmine rehabilitation in Australia for this length of time, and under natural rainfall conditions. The study describes the impact of vegetative cover on runoff and sediment loss, from a range of topsoil and spoils material, and a range of slope gradients. In addition the study also provided a unique opportunity to study and compare the nature of erodibility for individual rainstorm events and for very different materials using empirical and an erosion/deposition process models.
The study was conducted on 3 coalmines, Curragh, Goonyella/Riverside and Oaky Creek where topsoil and spoil material were representative of material found on central Queensland coalfields. Runoff and erosion and runoff rates were measured from 3 slope gradients (approximately 1 0, 20, 30%) where pasture and trees treatments were applied to topsoil and spoil plots (20 m x 5m). On the 20% slope a topsoil and spoil plot was left bare and are used in the thesis to determine field derived erodibilities.
A review of literature showed that topsoil and spoil material has very different physical and chemical characteristics, and these characteristics determine the relative success of vegetative growth. In general, spoil material is often, neutral to extremely alkaline (pH 6.4 - 9.6), saline to highly saline (EC 1 :5, 0.2 to 4.0 dS/m), highly sodic (ESP 25 to 50), and more deficient in N and available P than topsoil. On sodic spoils hard dispersive crusts form and result in reduced infiltration, increased runoff and erosion, and limited vegetative growth.
A review of erosion studies showed larger plot scale experiments conducted under natural rainfall better capture the effect of processes not evident or difficult to express at smaller scales, such as rilling, channelised flow and piping. All the methodologies reviewed have a role in understanding and quantifying erosion processes at different scales, detail and complexity. The great value that field plots offer is the ability to explore different rehabilitation treatments, and derive soil and spoil erodibility parameters over a wide range of rainfall and runoff erosivities.
A significant decrease in sediment concentration and sediment loss resulted from high percentage pasture cover (50%) that protected the surface from rainfall detachment, runoff entrainment, and overland flow velocities. By the end of the study there was up to 1200 t/ha less sediment loss from pasture plots compared to plots left bare. Once buffel grass (Cenchrus ciliaris) colonised soil plots there were negligible differences in soil erosion between slope gradients. In contrast, rhodes grass (Chloris gayana) only sparsely established on spoil. Where spoil formed a crust infiltration rate declined and resulted in poor vegetative growth and unacceptable large runoff and erosion rates throughout the study (>200t/ha/year). The field study found the greatest on-site soil erosion risk occurred before pasture cover established, when a large surface area of soil (>0.5 plot area) was exposed to rainfall and overland flow. It is suggested that a minimum surface cover of 50% is required to produce stable landforms on mine-rehabilitated slopes of up to 20%. On steeper slopes greater surface cover is required.
The Revised Universal Soil Loss Equation (RUSLE), and the Modified Universal Soil Loss Equation (MUSLE) empirical models, and the Griffith University Erosion System Template (GUEST) erosion/deposition process model are used to derive a single erodibility parameter for soil and spoil material at the 3 coalmines sites. A comparison between erodibility derived by the three models is conducted, along with a comparison of field and laboratory flume empirically derived erodibilities.
The study indeed showed that field derived RUSLE and MUSLE erodibity K values decreased exponentially with very large rainfall erosivity (EI30), and rainfall/runoff erosivity (W) indices. A decline in erodibility suggests sediment detachment limited processes are operating under large rainfall and runoff conditions that are not fully described by the empirical models. The decline in erodibility is attributed to a lack of soil and spoil disturbance, surface crusting and consolidation, and in some instancies due to progressive rock armouring over the 6-year study.
The study confirmed that when runoff is accounted for in MUSLE there is a closer relationship between observed and predicted event sediment loss than the RUSLE derived erodibilities, indicated by an improved model coefficiency E. When average RUSLE and MUSLE K values are derived to account for the variability in event erodibility there is a good agreement in observed and predicted average sediment losses for both models. It is suggested that RUSLE would be more readily used to determine sediment loss from mine rehabilitation hill slopes since there is less reliance on hydrological inputs, and given rainfall erosivity is a readily available parameter, and the field study has provided erodibility K values for a broad range of soil and spoil materials.
Just a 2-fold difference was found between lowest and highest derived average RUSLE and MUSLE K values. Whereas, a relationship between soil loss and vegetative cover (C-factor) showed a 100-fold reduction in sediment loss when >50% vegetative cover was attained. Highlighting the importance of predicting the suitability of soil and spoil material for successful establishment of surface vegetative cover is as equally important as the ability to predict erodibility. A comparison between laboratory and field derived RUSLE K values showed little promise, although there was a good agreement between MUSLE K values for the 3 spoil material.
The GUEST parameter β described soil and spoil erodibility where either sediment transport (β = 1 ), or sediment source limiting (β < 1 ) processes were operating. The GUEST derived erodibility parameters had less variability than the empirical derived parameters. However, rock armouring was a notable process that limited the supply of sediment available for rainfall and flow detachment and transport and reduced event β values during the study.
A comparison between the empirical and process based models showed a good relationship between MUSLE and GUEST erodibility parameters, where β is related to average MUSLE K by the equation
β = 1.741KMUSLE + 0.6836 r2 = 0.70
*the above equation cannot be accurately replicated from the original thesis. The K in KMUSLE has a macron ( ¯ ) above it.
Although there is a close relationship between MUSLE and GUEST erodibility parameters, the inclusion of a sediment depositability (φ) term in GUEST provided a greater understanding of erosion processes, and the source and fate of eroded sediment than either of the empirical models K factors are able to. For example, the Goonyella/Riverside spoil RUSLE and MUSLE K factors were up to half that of Curragh soil and spoil, whereas with GUEST erodibilities were similar. Although GUEST erodibilities were similar there were large differences in sediment depositability and hence transportability, with Goonyella/Riverside spoil producing a greater proportion of fine sediment, and low settling velocity (60% < 0.002 m/s) and effective depositability (φe) than Curragh soil and spoil. The example serves to illustrate the greater potential that erosion/deposition models such as GUEST have to predict the transport and fate of sediment to the wider environment than catchment models that simply rely on a constant sediment delivery ratio.
Despite mining companies and other agricultural practices undertaking rehabilitation and conservation practices, the impact of these systems on offsite sediment yields and downstream water quality are largely unknown and poorly quantified. Consequently, in order to assess the impact on sediment water quality, which is strongly related to the sediment size distribution, and also assess the fate of sediments generated, more research should be focussed on the development of multi-class erosion/deposition models.