Workshop do DIMNT/CGCT/INPE, 6-7 de outubro de 2022

Assimilação de dados nas componentes do sistema Terrestre:

status e perspectivas futuras no contexto do MONAN

​

Assimilação de Dados Oceanográficos

com o Sistema HYCOM+RODAS: Status e Perspectivas

Clemente A. S. Tanajura (UFBA)

Filipe Bitencourt Costa (CHM e UFBA)

Geoff Souza Dorfschäfer (UFBA)

Leonardo Brito Pires (UFBA)

Janini Pereira (UFBA)

Vitor Fernando da Silva Vidal

Davi Mignac Carneiro (UK MetOffice)

Rafael Santana (University of Auckland))

Alex Novaes de Santana (SOCIB)

Leonardo Lima (CMCC)

1

Content

2

• The REMO Data Assimilation System (RODAS)

​

• Experiments

​

• Future Plans

REMO is a research group

Research and technological development in operational oceanography and physical oceanography with focus on the South Atlantic and regions along the Brazilian coast using assimilative models and observational data. It was formed in 2007.

​

​

​

Products:

• Ocean weather forecasts
• Hydrodynamic databases (reanlayses)
• In situ observational data

www.rederemo.org

Ensemble Optimal Interpolation (EnOI)

EnOI: B is estimated with model states from a free run

​

The REMO Ocean Data Assimilation System (RODAS) was constructed based on EnOI; M Ensemble members are not static (Xie and Zhu 2010, Tanajura et al 2014, Mignac et al 2015; Costa e Tanajura 2015; Carvalho et al. 2019; Tanajura et al. 2020, Santana et a;, 2020; Dorfschäfer et al. 2020; Costa e Tanajura 2022)

11 members

11 members

M = 132

Operational Ocean Forecast in REMO-CHM

HYCOM 1/4^o 1/12^o 1/24^o with tides L21

Large-scale circulation in the Atlantic and downscaling to METAREA V and region along the Brazilian S-SE coast in the Brazilian Navy CHM operational system.

• 3-5 day forecasts
• NCEP/NOAA GFS and COSMOS forcing
• Simplified version of RODAS
• Assimilation of OSTIA SST, AVISO SLA and Argo T/S profiles
• Dissemination by the CHM web page

Brazilian Navy Hydrography Center (CHM) Operational System

Operational Ocean Forecast in REMO-CHM

Large-scale 1/12^o domain increased, Metarea V domain now with 1/24^o resolution L32 with tides.

• 3-5 day forecasts
• NCEP/NOAA GFS and COSMOS forcing
• Simplified RODAS version
• Assimilation of OSTIA SST, AVISO SLA and Argo T/S profiles
• Not yet disseminated in the web, but in app

This and the other nested system are running operationally today in CHM

HYCOM+RODAS 11-yr Reanalysis

• 11-year integration (2008-2018) with HYCOM 1/12^º L32+RODAS
• Atmospheric Forcing NCEP/NOAA CFSR each 6 h
• Assimilation each 3 days of

UK MetOffice SST analysis (1/20^º resolution)

SSH from CMEMS (gridded with 1/4^º resolution)

T/S profiles from 13,988 XBTs, 1,718 CTDs, 8 PIRATA buoys and 45,325 Argo profilers.

• Produced 3-D outputs each 3 h, and 2-D outputs in the mixed layer each 1 h

Strategy to use synthetic S to pair single T XBT and PIRATA profiles

Dorfschäfer et al. JGR (2020)

HYCOM+RODAS 11-yr reanalysis

Locations of the in situ data assimilated in the reanalysis from January 1, 2008 to December 31, 2018. On the left, 13,988 XBTs are in red, 1,718 CTDs in green and 8 PIRATA buoys in blue. On the right, 45,325 Argo T/S profiles are in blue. (Tanajura et al. 2022, under preparation)

HYCOM+RODAS 11-yr reanalysis

SST RMSD (^oC) with respect to OSTIA for the Control free run, reanalysis B_H_MV.2 and the COAPS/FSU HYCOM+NCODA system.

SST RMSD (^oC)

HYCOM+RODAS 11-yr reanalysis

SST RMSD (^oC)

SSH Correlation with respect to AVISO for the Control free run, reanalysis (B_H_MV.2) and the COAPS/FSU HYCOM+NCODA system.

HYCOM+RODAS 11-yr reanalysis

Control

Reanalysis

Control

Reanalysis

GLORYS

GLORYS

NCODA

NCODA

T RMSD (^oC)

S RMSD

30 d Predictability Assessment

Carvalho J, Costa, FB, Mignac D, Tanajura CAS, JOO, 2019

48 30-d hindcasts were performed and compared with HYCOM+RODAS analyses

Persistence

Persistence

Persistence

Hindcast

Hindcast

Hindcast

Free

Free

Free

OSE with HYCOM+RODAS

Tanajura et al.,Ocean Dyn, 2020

Assess the relative importance of

OSTIA SST

AVISO SLA

Argo T/S

Meridional current on Feb 27, 2012, along the NOAA XBT line AX97

OSE with XBTs and Synthetic Salinity

Dorfshäfer et al., JGR, 2020

OSSE: Assimilation of SWOT SSH

SLA SWOT data: along-track on 2 bands 50 km wide

2 km x 2 km resolution & 8 mm accuracy

Synthetic Data produced by Gaultier et al. (2021) software

OSSE: Assimilation of SWOT SSH

Assimilation Sea Surface Salinity (SSS) from SMOS

SSS data was assimilated in a 4-d cycle in 2010 and 2012 into HYCOM 1/12

Assimilation could impose interannual variability

SSS Difference 2010 - 2012

T(^oC) Difference With – Without PIRATA Jan-May 2012

OSE with PIRATA

Maps of time correlation of SLA (m) and AVISO for the Control Run, RODAS_EnOI and RODAS_EnKF

HYCOM+RODAS version with EnKF

RODAS_EnKF with only 11 members;

Trial period 01/06/2017 a 31/12/2017

Source: D.Sc. Thesis by Filipe Bitencourt Costa (Jul/2021)

Final Considerations

​

• HYCOM+RODAS skills can be improved, but its results are already good to be employed operationally and for research

​

• RODAS can be implemented into any ocean model, and effort is under way to implement it into ROMS

​

• Preliminary work was done to realize an EnKF version and in the future an Hybrid EnOI+EnKF will be explored

​

• 2 projects involving RODAS are submitted:

- BRICS to work with assimilation of data in the deep ocean and learn how to assimilate sea ice

- GOOS-BR to perform OSEs and OSSEs

​

• Partnership with CPTEC/INPE is now established to implement HYCOM+RODAS in CPTEC EGEON machine and collaborate with MONAN in ocean and/or coupled ocean-ice-atm data assimilation

​