Argo data assimilation into HYCOM with an EnOI method in the Atlantic Ocean

Abstract

Abstract. An ocean data assimilation system to assimilate Argo temperature (T ) and salinity (S) profiles into the HYbrid Coordinate Ocean Model (HYCOM) was constructed, implemented and evaluated for the first time in the Atlantic Ocean (78 • S to 50 • N and 98 • W to 20 • E). The system is based on the ensemble optimal interpolation (EnOI) algorithm proposed by Xie and Zhu (2010), especially made to deal with the hybrid nature of the HYCOM vertical coordinate system with multiple steps. The Argo T -S profiles were projected to the model vertical space to create pseudoobserved layer thicknesses ( p obs ), which correspond to the model target densities. The first step was to assimilate p obs considering the sub-state vector composed by the model layer thickness ( p) and the baroclinic velocity components. After that, T and S were assimilated separately. Finally, T was diagnosed below the mixed layer to preserve the density of the model isopycnal layers. Five experiments were performed from 1 January 2010 to 31 December 2012: a control run without assimilation, and four assimilation runs considering the different vertical localizations of T , S and p. The assimilation experiments were able to significantly improve the thermohaline structure produced by the control run. They reduced the root mean square deviation (RMSD) of T and S calculated with respect to Argo independent data in 34 and 44 %, respectively, in comparison to the control run. In some regions, such as the western North Atlantic, substantial corrections in the 20 • C isotherm depth and the upper ocean heat content towards climatological states were achieved.The runs with a vertical localization of p showed positive impacts in the correction of the thermohaline structure and reduced the RMSD of T (S) from 0.993 • C (0.149 psu) to 0.905 • C (0.138 psu) for the whole domain with respect to the other assimilation runs.

Assimilation of Argo Temperature and Salinity Profiles Using a Bias-Aware EnOI Scheme for the Labrador Sea

“…From these, 121 anomaly fields were generated by subtracting each daily mean state from its mean. This yields an ensemble of size 121, which is consistent with the ensemble size chosen in other studies (Mignac et al 2015;Xie and Zhu 2010;Counillon and Bertino 2009;Sakov and Sandery 2015).…”

“…The former is more likely to occur in regions of high mesoscale variability as well as in the thermocline region, while the latter is more likely to be the case for the deeper, more quiescent, levels of the model. Note that while the instrument error used for temperature is higher than that used in previous studies (Mignac et al 2015;Turpin et al 2016), it is the ratio between the observation and background error variances (as well as the spatial structure of the covariance fields) that determines the impact of the observation on the state estimate. For the formulation of the observation error used here, a different value of instrument error could be used with a different value of k to achieve similar results.…”

“…However, given the sparse distribution of the observations in space and time, it is more likely that the lack of initialization shock is due to the small a value used [see Eqs. (3) and (9), and the discussion in Mignac et al (2015)].…”

Analog ensemble data assimilation and a method for constructing analogs with variational autoencoders

“…Overall, the results are quite promising. EnOI is a widely used method (Xie et al ., 2011; Backeberg et al ., 2014; Mignac et al ., 2015; Deng et al ., 2018; Wu et al ., 2018) because of its acceptable performance and significantly reduced cost compared to EnKF, and ensemble background covariances are widely used in EnVar and hybrid data assimilation methods (Gharamti et al ., 2014; Bannister, 2017). The results here suggest that improvements could be obtained using either found analogs or constructed analogs; the increased cost of using analogs will be situation‐dependent, but if the costs can be made lower than the cost of forecasting an ensemble, then the analog EnOI or EnVar methods may be an attractive alternative.…”

Assessing the impact of Argo floats temperature measurements on the numerical weather prediction forecast skill

“…Many studies related to Argo data assimilation in ocean circulation models (e.g. Xie et al, 2010;Nilsson et al, 2011;Mignac et al, 2015) have revealed improvement as regards their forecast skill. In particular, MedArgo data are routinely assimilated (using localized Singular Evolutive Extended Kalman filtering techniques) on a weekly basis in one of the operational forecasting systems that are currently operating at HCMR, involving the Mediterranean basin, 1/10° resolution (POSEIDON system), and the Aegean Sea, 1/30° resolution (Korres et al, 2009;Korres et al, 2010;Korres & Kassis, 2012).…”