School of Earth & Environmental Sciences
Mogensen, Laura, Modelling the effect of sea level on SE Australian coastal wetlands: a multistage model validation and comparison study, BSci Hons, School of Earth & Environmental Sciences, University of Wollongong, 2016.
As the rate of sea-level rise is set to accelerate, there is increasing concern regarding the long- term sustainability of coastal wetlands. The validity of a model to reliably represent a particular wetland system considered to be vulnerable is crucial to support efficient management. The primary aim of this study was to examine the adequacy of numerical models in predicting the response of a SE Australian wetland to rising sea levels. A multistage validation process was employed to assess the operational and conceptual validity of three models for the Australian context, with specific focus on the Sea Level Affecting Marshes (SLAM) model originally developed for North American wetlands. A second model, the Spatially Applied Adjusted Temmerman (SAAT) model, originally developed for a Northern European wetland was adjusted and applied in this study. Comparison of the two models with a third developed specifically for an Australian context, the Oliver model, provides further insight into the adequacy of each model to predict the evolution of SE Australian coastal wetlands with rising sea levels.
Basic verification of the SLAM model revealed a significant flaw in the model code, whereby the A1T and A1FI maximum SLR scenarios were interchanged. Predictive validation suggested that the SLAM model had the greatest predictive power over decadal timescales. Inaccuracies noted between modelled and observed data revealed the potential inability of the model to capture important variables influencing the evolution of the Minnamurra site, such as rainfall, groundwater and El Niño–Southern Oscillation (ENSO) related environmental factors. Overall, however, projected model results and conceptual validation of the SLAM model revealed potential conceptual flaws regarding vegetation succession, treatment of wetland surface elevation change (SEC) and simulation of tidal water levels, all of which have the potential to decrease the predictive ability of the model and increase uncertainty of simulated results. The SLAM model was most sensitive to sea-level rise (SLR) and parameters pertaining to the inundation of wetlands, such as tidal range. Stochastic uncertainty analysis allowed for a richer understanding of possible future wetland distributions under rising sea levels but also indicated that the data and conceptual errors within the SLAM model propagated a wide range of uncertainty into deterministic model outcomes. Specific focus on the digital elevation model revealed high accuracy, obtained from expertly refining as-received Light Detection and Ranging (LIDAR) data, was crucial for modelling purposes.
Each of the models applied in this study generated plausible wetland distributions for future scenarios. Comparison of the models indicated that differences were primarily a result of model structure and mathematical expression, indicating that the most applicable model to the Australian context could not be definitively identified.
Despite the potentially large error and uncertainty, modelling remains important in a manager’s tool kit, providing an understanding of the potential response of wetlands to anticipated rising sea levels. It is recommended, however, that stochastic uncertainty analysis be conducted so as to encompass a wider range of possible future scenarios in the planning and decision-making processes regarding the protection of wetlands for the future.