Ensemble initialization of the oceanic component of a coupled model through bred vectors at seasonal-to-interannual timescales

Abstract

Abstract. We evaluate the ensemble spread at seasonal-tointerannual timescales for two perturbation techniques implemented in the ocean component of a coupled model: (1) lagged initial conditions as commonly used for decadal predictions; (2) bred vectors as commonly used for weather and seasonal forecasting. We show that relative to an uninitialized reference simulation the implementation for bred vectors can improve the ensemble spread compared to lagged initialization at timescales from one month up to three years.As bred vectors have so far mostly been used at short timescales, we initially focus on the implementation of the bred vectors in the ocean component. We introduce a depthdependent vertical rescaling norm, accounting for the vertical dependence of the variability, and extending the commonly used upper-ocean rescaling norm to the full water column. We further show that it is sufficient for the (sub-surface) ocean to breed temperature and salinity (i.e., scalar quantities), and rely on the governing physics to carry the temperature and salinity perturbations to the flow field.Using these bred vectors with a rescaling interval of 12 months, we initialize hindcast simulations and compare them to hindcast simulations initialized with lagged initial conditions. We quantify the ensemble spread by analyzing Talagrand diagrams and spread-error ratios. For both temperature and salinity, the lagged initialized ensemble is particularly under-dispersive for the first few months of predictable lead time. The ensemble initialized with bred vectors improves the spread for temperature and salinity for the 0-700 m and 1000-3500 m means, compared to the lagged ensemble at lead times of several months to one year. As the lead time increases to years, the differences between the two ensemble initialization techniques become more difficult to discern. While the results need to be confirmed in an initialized framework, the present analysis represents a first step towards improved ensemble generation at the transition from seasonal to interannual timescales, in particular at lead times up to one year.

The impacts of oceanic deep temperature perturbations in the North Atlantic on decadal climate variability and predictability

“…Ocean initial condition uncertainties and their impacts on climate prediction have been also addressed through bred vectors (Baehr and Piontek 2014) and anomaly transform methods (Romanova and Hense 2015) yielding a weak improvement of prediction reliability at seasonal timescales. Recently, Marini et al (2016) have achieved a greater ensemble spread for sea surface temperature (SST) during the first 3 years of simulations when oceanic singular vectors are used rather than atmospheric-only perturbations.…”

Seasonal Predictability over Europe Arising from El Niño and Stratospheric Variability in the MPI-ESM Seasonal Prediction System

“…Overall, the model yields a good prediction of tropical Pacific surface temperature for lead times of up to 6 months when initialized in November (Baehr et al 2014). As an illustration of the model skill for predicting ENSO, Fig.…”

“…The atmosphere is coupled to the MPI Ocean Model (MPI-OM; Jungclaus et al 2013) at a resolution of GR15L40 (1.58 horizontal resolution in the tropics, 40 vertical levels), and to a land model including a hydrological discharge model and an interactive sea ice model. For a detailed description of the model, the performed model runs, and the evaluation of the model skill, see Baehr et al (2014).…”

“…In the ocean, the ECMWF Ocean Reanalysis System 4 (ORA-S4; Balmaseda et al 2013) is used with a relaxation time scale of 10 days, and for sea ice the observational National Snow and Ice Data Center (NSIDC) sea ice concentration data (Fetterer et al 2002) are employed with an effective relaxation time scale of 20 days (Tietsche et al 2013). From the nudged model experiment, nine ensemble members are started on 1 November of each year between 1980 and 2011 with a duration of one year each, using the breeding method for ensemble generation (Baehr and Piontek 2014). The first month after the initialization (i.e., November) is discarded because of the possibility of initialization shock.…”

Dynamical attribution of oceanic prediction uncertainty in the North Atlantic: application to the design of optimal monitoring systems

“…There are various strategies to accurately sample the initial condition uncertainty and to accurately define the ensemble (e.g., Persechino et al 2013). However, except for a few dedicated studies (e.g., Du et al 2012;Baehr and Piontek 2014;Germe et al 2017b), the role of the oceanic uncertainties has often been overlooked in these ensemble design strategies, despite being a possible source of predictability (Hawkins et al 2016).…”