Seaice data assimilation and forecasting using an Ensemble
Sea-ice data assimilation and forecasting using an Ensemble Transform Kalman Filter (ETKF) Paul Sandery 7 May 2018 CSIRO OCEANS AND ATMOSPHERE
ACCESS-ESM GFDL FLOR GFDL CM 2. 5 GFDL CM 2. 6 etc GFDL CM 2. 1 CSIRO DFP V 2 with BGC 2 | Coupled sea-ice DA in CAFE-ETKF | Paul Sandery En. KF-C • • • Observations 96 member ensemble 7 day assimilation cycle Coupled state vector Asynchronous DA ETKF bias correction SST and SLA
Assimilated Observations • RADS altimetry • SST – NAVO-AVHRR, AMSR-E, AMSR-2, Wind. Sat, PATHFINDER, VIIRS • SSS – SMOS L 2 debiased SMAP • In-situ T/S from CARS and WMO GTS including ARGO, CTD, XBT, RAMA, PIRATA, TAO-TRITON. • OSISAF sea ice concentration • Under ice freezing point SST derived from OSISAF All used for verification 3 | Coupled sea-ice DA in CAFE-ETKF | Paul Sandery
4 | Coupled sea-ice DA in CAFE-ETKF | Paul Sandery
Context • Sub-optimal model • Summer southern ocean SST warm bias • No where near enough Antarctic summer sea ice 5 | Coupled sea-ice DA in CAFE-ETKF | Paul Sandery
Dramatic improvement with SST & SLA bias correction Global 28 day forecast errors using all forward independent observations 6 | Coupled sea-ice DA in CAFE-ETKF | Paul Sandery
Assimilating sea-ice concentration 96 MEMBER GFDL CM 2. 1 (MOM 5 -SIS-AM 2) ENSEMBLE • Assimilation Methods • SIC observations are 2 D • SIS has 5 ice thickness categories • RED, RFT (Smith et al, 2016) • Augment state vector (Barth et al, 2015) • Assimilate freezing point temperature under ice based on SIC observations (Sandery, 2018) • Combination of Barth and Sandery 7 | Coupled sea-ice DA in CAFE-ETKF | Paul Sandery
Assimilating sea-ice concentration 96 MEMBER GFDL CM 2. 1 (MOM 5 -SIS-AM 2) ENSEMBLE • SIC observation error function to avoid overfitting (Sakov et al, 2012) 8 | Coupled sea-ice DA in CAFE-ETKF | Paul Sandery
Assimilating sea-ice concentration 96 MEMBER GFDL CM 2. 1 (MOM 5 -SIS-AM 2) ENSEMBLE • SIC observation error function to avoid overfitting (Sakov et al, 2012) 9 | Coupled sea-ice DA in CAFE-ETKF | Paul Sandery
Assimilating under-ice freezing point temperature 10 | Coupled sea-ice DA in CAFE-ETKF | Paul Sandery
Assimilating under-ice freezing point temperature CONTROL ASSIMILATION 11 | Coupled sea-ice DA in CAFE-ETKF | Paul Sandery
CONTROL 12 | Coupled sea-ice DA in CAFE-ETKF | Paul Sandery SIC ASSIM SIT ASSIM
• Model agnostic approach to state estimation • Aiming for a series of clean multi-year DA experiments testing sea-ice assimilation methods • Aim to improve forecast models by using ETKF for parameter estimation – train model parameters on observations 13 | Coupled sea-ice DA in CAFE-ETKF | Paul Sandery
Thank you Dr Paul Sandery t +61 3 6232 5035 e paul. sandery@csiro. au w www. csiro. au/dfp CSIRO OCEANS AND ATMOSPHERE
- Slides: 14