Joint Simulation of Mine Spoil Uncertainty for Rehabilitation Decision Making

  • R. Dimitrakopoulos
  • S. Mackie

Abstract

Soil attributes, including those in mine spoil heaps, critically affect plant growth during land rehabilitation. Their characterization through a limited number of samples requires quantification of spatial variability, which is then used at various stages throughout the rehabilitation process and assists risk analysis and rehabilitation decision making. Stochastic simulation is a tool used for the quantification of uncertainty. This paper presents the suitability of stochastic simulation for the joint simulation of soil attributes and a computationally efficient method based on: (i) the Minimum/Maximum Autocorrelation Factors (MAF), involving the de-correlation of pertinent variables into spatially non-correlated factors, and (ii) the simulation of MAF and back transformation to the conditional simulations of the correlated variables. MAF factors in point (ii) are simulated using the new generalized sequential Gaussian simulation technique which is substantially more efficient than the traditional sequential simulation methods. The formulated approach is applied to mine spoil data, specifically electrical conductivity and pH, which typically contribute to restricted plant growth on spoils in coal mines. The results of the simulations are used to quantify the risk of exceeding significant thresholds for each variable, thereby identifying problem rehabilitation areas. The case study demonstrates the practical aspects of the method, as well as its use in planning rehabilitation strategies and predictions of future performance of the rehabilitation.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer Science+Business Media B.V. 2008

Authors and Affiliations

  • R. Dimitrakopoulos
    • 1
  • S. Mackie
  1. 1.Department of Mining and Materials EngineeringMcGill UniversityMontreal QcCanada H3A 2A7

Personalised recommendations