An adaptive multi-objective framework for the scheduling of environmental flow management alternatives using ant colony optimization

Abstract

Rivers and their adjacent wetlands and floodplains worldwide have been altered or have vanished as a result of river regulation and development (such as dams, locks and weirs), as well as water over-allocation. In recent years, environmental flow management has been suggested as a means to mitigate these negative impacts. One approach in order to do this is through the scheduling of environmental flow management alternatives (EFMAs), such as reservoir releases and the operation of wetland regulators. However, this is not an easy task for the following reasons: (i) there are generally many wetlands and floodplains in any particular river system, all containing a wide range of biota that have different flow requirements; (ii) there is generally limited water allocated for environmental purposes, since there are multiple users (e.g. irrigation, domestic), all competing for the same water source; (iii) the schedules are generally developed over multiple years; and (iv) there are multiple competing objectives and constraints that need to be considered. This problem therefore lends itself to be formulated as an optimization problem, where the aim is to maximise the ecological integrity of the system, while also considering humans needs and the constraints of the system. In this thesis, a generic adaptive multi-objective optimization framework for determining the optimal schedule of EFMAs for rivers and their associated wetlands and floodplains is developed and tested. In order to achieve this, ant colony optimization algorithms are selected, since they can take into account the conditional dependencies and sequential nature of the scheduling problem explicitly. This is possible, as the solution space can be represented by a graph structure that can be adjusted dynamically based on the choices made at previous points in the decision graph, thereby reducing the size of the decision space and increasing the proportion of feasible solutions. This is not possible when most other metaheuristics are used. In addition to this, the framework is adaptive and able to incorporate forecasts of environmental water allocation, such that the environmental water can be used most efficiently in order to maximize ecological response. The major research contributions are presented in three journal publications. Firstly, the initial single-objective formulation of the optimisation framework, which incorporates the temporal dependencies associated with the scheduling of EFMAs is presented and validated using a hypothetical case study. The framework is then extended to incorporate multiple objectives and applied to a river section in the South Australian River Murray, so that the trade-off between the ecological response and environmental water allocation can be examined. Finally the framework is further extended to incorporate adaptive features by using forecasts of environmental water allocation in the development of EFMA schedules, as well as an additional objective which aims to minimise the number of differences of EFMA schedules developed at subsequent time steps. Thus the framework provides valuable insight to managers into the EFMA scheduling problem, as it can be applied to investigate a wide variety of problems, such as investigating the likely ecological benefit gained from an increase in environmental allocation, the impact of system constraints on ecological response and the potential advantages of investment in additional infrastructure.