Tide-surge adjoint modeling: A new technique to understand forecast uncertainty

Chris Wilson, Kevin J. Horsburgh, Jane Williams, Jonathan Flowerdew, Laure Zanna

Research output: Contribution to journalArticlepeer-review


For a simple dynamical system, such as a pendulum, it is easy to deduce where and when applied forcing might produce a particular response. However, for a complex nonlinear dynamical system such as the ocean or atmosphere, this is not as obvious. Knowing when or where the system is most sensitive, to observational uncertainty or otherwise, is key to understanding the physical processes, improving and providing reliable forecasts. We describe the application of adjoint modeling to determine the sensitivity of sea level at a UK coastal location, Sheerness, to perturbations in wind stress preceding an extreme North Sea storm surge event on 9 November 2007. Sea level at Sheerness is one of the most important factors used to decide whether to close the Thames Flood Barrier, which protects London. Adjoint modeling has been used by meteorologists since the 1990s, but is a relatively new technique for ocean modeling. It may be used to determine system sensitivity beyond the scope of ensemble modeling and in a computationally efficient way. Using estimates of wind stress error from Met Office forecasts, we find that for this event total sea level at Sheerness is most sensitive in the 3 h preceding the time of its unperturbed maximum level and over a radius of approximately 300 km. We also find that the pattern of sensitivity follows a simple sequence when considered in the reverse-time direction.

Original languageEnglish (US)
Pages (from-to)5092-5108
Number of pages17
JournalJournal of Geophysical Research: Oceans
Issue number10
StatePublished - Oct 2013


  • adjoint
  • ensemble
  • sea level
  • sensitivity
  • surge
  • tide

ASJC Scopus subject areas

  • Geochemistry and Petrology
  • Geophysics
  • Oceanography
  • Space and Planetary Science
  • Earth and Planetary Sciences (miscellaneous)


Dive into the research topics of 'Tide-surge adjoint modeling: A new technique to understand forecast uncertainty'. Together they form a unique fingerprint.

Cite this