- 21st September 2017
- Posted by: Sophie Bunker
- Category: Blog
On Wednesday 6 September Jennifer Hornsby attended the EVAN conference in Southampton. She presented Matt Elliot, previous Technical Director for Coastal Sediment Dynamics, work on the ‘Comparison of tropical storm water level modelling an ensemble approach’ within the Storm Surges session.
The conference provided a forum for the exchange of experiences and ideas on the extreme value analysis of Natural Hazards. Studies of extremes of geophysical variables and works related to natural hazards were welcomed. Her presentation abstract is highlighted below.
Comparison of tropical storm water level modelling an ensemble approach
Existing studies of tropical storm water levels (combined tide and surge) provide a spatially dispersed and sometimes divergent body of information for the description of coastal flooding hazard. This information is often largely neglected in favour of the most recent study when attempting to resolve hazard at a new location. Evaluation of previous tropical storm water level modelling may provide a valuable context for analysis, that is otherwise disregarded. An ensemble approach has been developed to synthesise more than 50 studies along the Western Australian coast, with objectives to both draw together the knowledge and contrast of the effect of different modelling approaches.
Unification of dispersed information, built from a succession of local studies, required representation of geographically variable factors contributing to water level. Statistical analysis of tropical cyclone meteorology was combined with detailed analysis of tide gauge observations to determine spatial variations. This incorporated a geographic shift of active processes, from surge developed through onshore winds on the north coast, to an increased response to storm systems tracking parallel to the west coast. The statistical model enables comparison of studies undertaken at various locations, and therefore provides an ensemble range reflecting influence of different assessment methods.
The study demonstrated a need for careful use of rare extreme events due to the sensitivity of model outcomes to how these observations are included in model validation. Processes of spatial aggregation and downscaling were developed for both tropical cyclone meteorology and observational data, including geomorphic evidence of extreme flooding. Further challenges to providing a unified framework were caused by extra-tropical storm transition, the wide range of surge-tide ratios, and the complexity of interacting seasonal phases for tropical storms, tides and mean sea level variability.
Comparison of previous studies demonstrated the selection of study methodology has at least an equivalent effect on results as geographic variability. Statistics derived from tide gauge records alone gave substantially lower estimates of recurrence within the ensemble range than from hydrodynamic modelling. Regional studies, with simplified representation of processes, also typically produced lower estimates of extremes than locally focused studies, although the latter were often inconsistent between sites. The potential influence of tropical storms for coastal flooding on the western coast appears to be understated, likely due to the focus of existing models on surge developed through onshore winds and neglect of extra-tropical transition.