Current practice for landscape reconstruction following opencast mining relies on topographic reconstruction, adaptive land management and botanical characterisation. Environmental processes may be altered where reconstructed landforms have significant relief. Consequently, environmental outcomes in cases where there is large scale land forming are unpredictable. Moreover, landscape restoration lacks an integrated methodology, and while many mine closures have detailed ecosystem and biodiversity objectives based on natural analogue areas there has been no reliable way to design these objectives into mine landforms. The methods used in landscape restorations to describe reference conditions are based on generalised environmental factors using regional information and incorporating conceptual models. Such models lack the precision and accuracy required to understand and restore hillslope environmental pattern at mine sites.
However, methodological integration and statistical inference models underpinning the spatial inference methods in conservation and landscape ecology, and pedology may be applied to solve this problem. These inference models utilise digital terrain models as the core environmental data incorporating ecological theory to predict biodiversity and species distribution. Also, numerical mass balance models such as water and solute balance, which have been applied to understand environmental processes in landscapes, can be used to assess mine landform design. The objective of the work reported here was to investigate environmental variation, with sufficient accuracy and precision, in natural landscapes to design mature mine landforms and to demonstrate the capacity to predict ecological outcomes. This would extend current best practice - designing mine landforms with predictable hydrological and geotechnical outcomes needed to protect off-site environmental conditions – to the on-site environment after closure.
The specific aims of this thesis were to: (i) evaluate the predictability of ecosystems based on regional ecological mapping: (ii) develop and evaluate quantitative, site specific environmental mapping and natural analogue selection methodology; (iii) evaluate a trial final landform cover (reconstructed soil) using water balance, water chemistry monitoring; (iv) design and evaluate a conceptual mine landform through the assessment of environmental processes in natural analogue areas; and (v) make valid predictions of revegetation outcomes on the conceptual landform. In meeting these aims, links between ecological theory, landscape analysis and the current practice in mine landform design were identified.
The first phase of the thesis involved environmental investigations and surveys of extensive savanna environments on the Tiwi Islands (7320 km-2) and similar environments in the vicinity of Ranger uranium mine (150 km-2) in northern Australia. This first phase, reported in Chapter 3, investigated the reliability of conceptual landscape models used in regional ecological mapping in predicting ecological patterns in terms of vegetation and soil. The Tiwi Islands was selected because of the relatively uniform parent material and its simplified climate. This allowed the study of physiographic control of soil and vegetation patterns. The results identified correlations between vegetation pattern and landform that were confounded by a subjective and complex land unit model of ecosystems. This investigation enabled the development methodological approach to analogue selection and ecological modelling at Ranger uranium mine – a site that will require a restoration approach so as to meet environmental closure objectives.
The second phase is the methodological development – involving an initial reconnaissance, is presented in Chapter 4. This phase was aimed at selecting natural analogue areas for mined land restoration. Environmental pattern recognition involving classification, ordination and network analysis was implemented based on methods of conservation ecology. This led to quantitative landscape model to identify natural analogue areas and design ecosystem surveys. This quantitative landscape model incorporated a grid survey of vegetation and soil variation into a nearby analogue landform that matched the area of mine disturbance. This analogue landform encapsulates the entire ecosystem types observed on rocky substrates in the broader reconnaissance survey. The natural analogue selection incorporated a combination of digital terrain analysis and k-means clustering of primary and secondary terrain variables to classify habitat variation on hillslopes. Landscapes with similar extent to the mine landscape were identified from numerical similarity measures (Bray-Curtis) of fine grained habitat variation and summarised using a dendrogram. The range in hillslope ecosystem types were described from stratified environmental surveys of vegetation and soils along environmental gradients in selected analogue landforms.
The results show that the mapped environmental factors in close correlation with water and sediment distribution were strongly associated with observed vegetation patterns in analogue areas at Ranger uranium mine. Environmental grain size and landform extent concepts were therefore introduced using landscape ecology theory to integrate different scales of environmental variation in a way that provides direct context with the area impacted by mining. Fine-grained environmental terrain attributes that describe runoff, erosion and sediment deposition were derived from a digital elevation model and classified using non-hierarchical multivariate methods to create a habitat class map. Patch analysis was used to aggregate this fine-grained environmental pattern into a grid that matched the scale of the mine landform. The objective was to identify landforms that were similar in extent to the reconstructed mine landscape. Ecosystem support depends on soil as well as geomorphic factors.
An investigation into critical environmental processes, water balance and solute balance, on a waste rock landform at Ranger uranium mine is presented in Chapter 5 to characterise waste rock soils and investigate cover design options that affect environmental support. This involved monitoring of water balance of a reconstructed soil cover on a waste rock landform for four years and the solute loads for two years. A one dimensional water balance model was parameterised and run based on 21 years of rainfall records so as to assess the long-term effects of varying cover thickness and surface compactness on cover performance. The results show that the quality of runoff and seepage water did not improve substantially after two years as large amount of dissolved metal loads persisted. Also, tree roots interacted with the subsoil drainage-limiting layer at one metre below the land surface in just over two years - and thus altering the hydraulic properties of the layer. Further, the results of water balance simulations indicate that increasing the depth to, and thickness of, the drainage-limiting layer would reduce drainage flux. Increasing layer thickness could also limit tree root penetration. It was also found that surface compaction was the most effective means of limiting deep drainage, which contained high concentrations of dissolved metals. However, surface compaction creates an ecological desert. Therefore long-term rehabilitation of the cover will be required to allow water to infiltrate for it to be available for ecosystems. A cover that can store and release sufficient water to support native savanna eucalypt woodland may need to be three to five metres deep, including a drainage limiting layer at depth so as to slow vertical water movement and comprise a well graded mix of hard rock and weathered rock to provide water storage and erosion resistance. The resulting waste rock soils would be similar, morphologically to the gradational, gravelly soils found in natural analogue areas.
The study then shifted from mined land back to a selected natural analogue landscape at Ranger mine in Chapter 6. The fine grained variation in terrain attributes is described to support a landform design that allowed for mine plan estimates of waste rock volumes and pit void volumes. A process of developing and evaluating the landform design was put forward, in the case of Ranger, that begins with key stakeholder consultation, followed by an independent scientific validation using published landform evolution and integrated, surface-groundwater water balance modelling. The natural analogue and draft final landforms were compared in terms of terrain attributes, landform evolution and eco-hydrological processes to identify where improvements could be required. The results of the independent design reviews are contained in confidential reports to Ranger mine and in conference proceedings that are referenced in Chapter 6. Independent validation will be a key element of an ecological landform design process and the application of published eco-hydrological and landform evolution models at the Ranger mine case study site are presented as an example of current best practice. Also, detailed assessment was made of environmental variation and soil and geomorphic range in the selected analogue landscape to support the landform design process with the mining department.
Ecological modelling of the distributions of framework species in the reconstructed landscape is proposed as an additional assessment tool in this thesis to validate an ecological landform design methodology. To this end, a detailed environmental survey is presented in Chapter 6 of the soils and vegetation in a selected natural analogue area of Ranger mine to identify common and abundant plant species and their distribution in a similar landscape context to the mined land. This work supported ecological modelling of species distributions in reconstructed and natural landscapes in the following chapter.
The results of species distribution models for reconstructed and natural landscapes at the Ranger mine site are reported in Chapter 7. The aim was to predict the distribution of common and abundant native woodland species across a landscape comprising a sculpted, post mining landform within a natural landscape. Species distribution models were developed from observations of species presence-absence at 102 sites in the grid survey of the natural analogue area that was reported in Chapter 6. Issues related to optimising predictor selection and the range of environmental support were investigated by introducing survey sites from the broad area reconnaissance survey reported in Chapter 4. Added to these are the published species abundance data from an independent regional biodiversity survey of rocky, well drained eucalypt woodlands, used as analogues of mined land. Plant species responses to continuous and discrete measures of environmental variation were then analysed using multivariate detrended correspondence analysis and canonical correspondence analysis to select independent variables and assess the relative merits of abundance versus presence absence observations of species. Then, generalised additive statistical methods were used to predict species distributions from primary and secondary terrain variables across the natural analogue area and a reconstructed post-mining landform. This analysis was completed with an assessment of the effect that survey support has on model formulation and accuracy. The scale of the mine landscape was found to provide important context for the stratified environmental surveys needed to support predictive modelling. Extending the geographic range of survey support did not improve model performance, while survey sites remote from the mine introduced some degree of spatial autocorrelation that could reduce the prediction accuracy of species distributions in the mine landscape. Further work is needed to address uncommon species or species with highly constrained environmental ranges and aspects of landform cover design and land management that affect woodland type and vigour.
The combined studies reported in this thesis show that the predictability of mine land restorations is dependent on the landscape models used to characterise the natural analogue areas. It is demonstrated that conceptual ecological models developed for regional land resources survey, commonly used to select natural analogue areas, are subjective, complex and unreliable predictors of vegetation and soil patterns in hillslope environments at particular sites. It was recognised that environmental patterns are subject to terrain and hillslope environmental variation across an extensive areas. The landform model for selecting natural analogues was refined by introducing grain size and ecological extent concepts, used to describe ecological scale in landscape ecology, to address these effects. These refined concepts were adapted to define environmental variation in the context of natural analogue selection for mining restoration, rather than home range habitat conditions for native animals as was their original purpose. It is demonstrated here that the grain size and extent of environmental variation in the natural landscape can be used to select natural analogue landforms, develop ecological design criteria and design field surveys that support the capacity to predict the distributions of common and abundant woodland species in a reconstructed landscape.
In conclusion, it is worth noting that an integrated ecological approach to landscape design can be applied to closure planning at mine sites where cultural and ecological objectives are critical to the success of the mine rehabilitation. Furthermore final landform trials could be used to support a restoration approach — providing an understanding of the interactions between critical physical and ecological processes in the soil layers and environmental processes at catchment scales. The accuracy of the inferences made is dependent on the understanding of hydrological processes in natural and constructed landforms. However, the natural analogue approach provides a clear landscape context for these trials. In a world where species extinction resulting from habitat loss is one of the most important global ecological issues, mine rehabilitation offers unique experimental opportunities to develop capability in ecosystem rehabilitation.