Satellite data assimilation and estimation of a 3D coastal sediment transport model using error-subspace emulators

Journal Publication ResearchOnline@JCU
Margvelashvili, N.;Andrewartha, J.;Herzfeld, M.;Robson, B.J.;Brando, V.E.
Abstract

This paper describes sequential assimilation of data into a three-dimensional coastal ocean model using fast and cheap statistical surrogates of the model (emulators). The model simulates resuspension and deposition of fine sediments in a macro-tidal environment of the Fitzroy Estuary and Keppel Bay, North-East Australian coast. The assimilation algorithm was applied first to synthetic observations produced by a twin model run, and then with real data obtained from satellite observation. The latter are derived from remote sensing algorithms customised to the study region. The main objective of simulations was to test the data assimilation scheme using synthetic observations and identify potential issues and challenges when assimilating real data sets. The assimilation algorithm proved capable of substantially reducing a prior uncertainty of the model for both the scenario with the synthetic observations and the scenario with the satellite data. Significant remaining error in western Keppel Bay after assimilating satellite data is diagnostic of an underlying error in the system conceptualisation in other words, it indicates that the primary source of error is not in the parameter values specified, but in the model structure, in the interpretation of satellite data or in the other input data. The results of our study show the utility of the developed technique for the data assimilation into the three-dimensional sediment transport model of the Fitzroy estuary and Keppel Bay. More research is required to understand the capacity of this technique to generalise to other models and regions. Crown Copyright (C) 2012 Published by Elsevier Ltd. All rights reserved.

Journal

Environmental Modelling and Software

Publication Name

Environmental Modelling and Software

Volume

40

ISBN/ISSN

1364-8152

Edition

N/A

Issue

N/A

Pages Count

11

Location

N/A

Publisher

Pergamon-Elsevier Science Ltd

Publisher Url

N/A

Publisher Location

N/A

Publish Date

N/A

Url

N/A

Date

N/A

EISSN

N/A

DOI

10.1016/j.envsoft.2012.09.009