| Title: |
A more rigorous approach to calibrating and assessing the uncertainty of coastal numerical models |
| Authors: |
Simmons, JA; Marshall, LA; Turner, IL; Splinter, KD; Cox, RJ; Harley, MD; Hanslow, DJ; Kinsela, MA |
| Source: |
Australasian Coasts & Ports Conference 2015, Auckland, New Zealand, 2015-09-15 - 2015-09-18 |
| Publication Year: |
2015 |
| Collection: |
UNSW Sydney (The University of New South Wales): UNSWorks |
| Description: |
There has been widespread effort to develop sophisticated numerical models to predict coastal change at a variety of time and spatial scales. All these models contain free parameters that require calibration to the available field data and little guidance (beyond the adoption of the default values provided) is presently available to inform the selection of best-fit parameter values. In practice, the means of optimising these parameters often lacks a sufficiently rigorous assessment of the impacts of parameter interdependence and parameter-induced model uncertainty is rarely quantified. The Generalised Likelihood Uncertainty Estimation (GLUE) method has been employed extensively in the field of hydrology and has proven to be a conceptually simple and efficient method to evaluate model sensitivities to parameter values and identify any inherent model structural errors. The GLUE method approaches the problem of ‘equifinality’ (i.e. the likely existence of multiple ‘optimum’ parameter sets) using Monte Carlo simulation applied to create many different combinations of possible model parameters. The outcomes of the Monte Carlo analysis are parameter posterior distributions, which can then be used by the modeller to determine parameter values most likely to produce model predictions with the highest skill. This paper describes the new application of the GLUE method to the XBeach storm erosion model, using data from a site in Italy where the XBeach model has been previously applied without such a rigorous evaluation of parameter sensitivity. The results presented demonstrate the more generic effectiveness and applicability of GLUE in the field of coastal engineering. The sensitivity of XBeach to each trialled free parameter is determined in a rigorous and transparent manner, and uncertainty bounds are obtained. This enables the modeller to better understand and quantify model skill in predicting observed and potential future erosion. |
| Document Type: |
conference object; report |
| Language: |
unknown |
| Relation: |
http://purl.org/au-research/grants/arc/DP150101339; https://hdl.handle.net/1959.4/unsworks_38478 |
| Availability: |
https://hdl.handle.net/1959.4/unsworks_38478 |
| Rights: |
metadata only access ; http://purl.org/coar/access_right/c_14cb ; CC-BY-NC-ND ; https://creativecommons.org/licenses/by-nc-nd/4.0/ |
| Accession Number: |
edsbas.20F8B0A |
| Database: |
BASE |