Ensemble downscaling of a regional ocean model. (January 2020)

• Record Type: Journal Article
• Authors: Ghantous, Malek; Ayoub, Nadia; De Mey-Frémaux, Pierre; Vervatis, Vassilios; Marsaleix, Patrick
• Is Part Of: Ocean modelling. Volume 145(2020)
• Publication Date: 2020-01
• Subjects: Ensemble modelling -- Downscaling -- Model uncertainties -- Coastal modelling -- Bay of Biscay -- Data assimilation; Oceanography -- Periodicals; Océanographie -- Périodiques; Oceanography; Periodicals; 551.46
• DOI: 10.1016/j.ocemod.2019.101511
• Languages: English
• ISSNs: 1463-5003

Abstract

Abstract: We downscaled a free ensemble of a regional, parent model to a high-resolution coastal, child ensemble in the Bay of Biscay. The child ensemble was forced at the open boundaries by the parent ensemble, and locally by perturbing the winds. By comparing ensembles generated by each of these forcing perturbations separately and combined we were able to consider the ensemble from either of two paradigms: (1) characterising high-resolution, coastal model errors using local and non-local forcing perturbations, or (2) downscaling regional model errors into the coastal domain. We found that most of the spread in the child ensembles was generated from the ensemble of open boundary conditions, with the local wind perturbations on their own generating substantially less ensemble spread. Together, the two sources of error increased the ensemble spread by only a small amount over the non-local perturbations alone. In general, the spread in sea surface height was greater in the child ensembles than in the parent ensemble, probably due to the more refined dynamics, while the spread in sea surface temperature was lower, likely due to the way the open boundary conditions were averaged. Deep below the surface, though, the child ensemble featured a large spread even where the parent model's spread was very weak. This enhanced error response is a promising result for an ensemble data assimilation system, as it could be exploited to correct the model deep below the surface.