Multi-criteria decision support for the revitalisation of river floodplains

Ecological revitalisation of river floodplains has become a very actual issue worldwide. It has been recognised that floodplains have the potential to become ecologically very productive areas inhabited by many valuable and rare species. Floodplains also play an important role in regional and even in continental context. Namely, beyond their own ecological values floodplains are also important as far as migration and spreading of species are concerned.Only very few natural floodplains have been left along the major regulated rivers of Europe. Thus, revitalisation of the existing ones has a special importance. Some of the revitalisation objectives can be achieved by means of administrative measures but there are several problems, which need active engineering intervention. These problems to be remedied are the results of adverse changes in hydrological conditions, which are the most influencing abiotic factors for floodplain ecosystems. The general problem of desiccation for example has been caused by the decreasing water levels in the rivers, following widespread river training since the 19 thcentury.Thus, the key to ecological revitalisation of river floodplains is improving the water regime. This is proposed to be implemented by transfonning the water conveyance infrastructure in an appropriate manner. Measures like enlarging and/or narrowing floodplain-channels, erecting summer dikes, installing and operating sluices are envisaged for this purpose. Because of the requirements of flood control, navigation and land uses, and also because of conflicts between the envisaged ecological objectives, floodplain revitalisation is inevitably a decision problem with conflicting multiple objectives and multiple stakeholders forming a Decision Making Group (DMG). The task of the DMG is to identify a compromise solution, with satisfactory achievements of all objectives.This dissertation has been conceived as to develop a computer based Decision Support System (DSS) that assists the DMG in searching for compromise solutions for the ecological revitalisation of river floodplains. The search takes place in a discrete decision space where the alternative solutions are specific combinations of water control measures mentioned above.The proposed DSS is based on the Evolutionary SEquential Multi-Objective Problem Solving (ESEMOPS) method developed by Bogardi & Sutanto [1994]. ESEMOPS actually solves a series of surrogate optimisation problems. These surrogate problems seek a minimum of the deviations between the actual aspiration levels of the DMG regarding the individual objectives and their respective achievements through optimisation. It is important to emphasise that surrogate optimisation problems are not substitutes for an explicit expression of the DMG's preference function. By comparing and analysing the solutions of these problems the DMG only gains information concerning the set of feasible solutions, and learns about the interrelationships (trade-offs) between the applied evaluation criteria. These interrelationships are estimated by alternatively treating some objectives as constraints thus enforcing their achievement. On the basis of intermediate results, the DMG modifies its preference structure (by modifying the aspiration levels) and repeats the surrogate optimisation processes. This iterative-interactive search goes on until the DMG declares a certain solution as the best compromise solution for the given problem.Surrogate optimisation problems are solved with the help of a Genetic Algorithms (GA) based search technique. The major advantage of GA is that it relies exclusively on information already available in the input (population of alternative solutions), and output (surrogate objective values) of criteria evaluations throughout the search towards the optimum of the actual surrogate problem.During each GA run a high number of alternative solutions has to be evaluated according to the selected criteria. Criteria evaluation is carried out with the help of a complex modelling system. The most important module of this system is a hydrodynamic model, which enables to simulate the water regime on the floodplain. The applied model is called FOK, which has been developed by the author. FOK belongs to the family of unsteady cell-type floodplain models. Its most pronounced feature is the extremely high simulation speed, which is required by the DSS in order to complete the GA searches within reasonable time limits. The high calculation speed has been achieved with the help of adaptive time step and model configuration control and with that of applying pre-calculated 'delivery functions' for flow calculations in the floodplain-channels. FOK is also enabled to cope with supercritical flow conditions, which may frequently occur in such channels.Output of the F0K model (water level time series on the floodplain) are input for ecological models. which help to estimate the achievements of ecological criteria. Ecological models embedded into the DSS are ranging from the physiologically based MEGAPLANT model [Scheffer et al., 1993] simulating the growth of water plants to habitat evaluation models for fish and for waders. These habitat models have been formulated by, adopting existing habitat modelling approaches to the unsteady hydrological conditions of floodplains.The DSS has been tested on a case study problem. The selected area is a sub-system of the Danube riparian Gemenc floodplain in Hungary. The identified ecological objectives prescribe the improvement of conditions for typical alluvial wet ecosystems. Special attention is paid on improving habitat conditions for fish and for waders like the famous black stork (Ciconia nigra) . The set of criteria formulated for the ecological objectives have been supplemented with criteria representing cost minimisation and timber production. This latter is an important economical use of the area, which has to be taken into consideration in the future as well, After identifying the decision and criteria spaces a decision making 'game' has been played with the help of the ESEMOPS based DSS, The purpose of the game was to locate a best compromise solution according to the foreseen preferences of a potential DMG dealing with the revitalisation of the Gemenc floodplain.It can thus be concluded that developing such a DSS is quite an interdisciplinary task involving the disciplines of ecology, fluid mechanics, numerical modelling, statistics and operations research. Accordingly, the key contribution of the present dissertation is the new, interdisciplinary concept that forms the basis of the proposed DSS. The major challenges that had to be faced during the development of the DSS were selection, adaptation and integration of techniques from the different disciplines.