To improve the sea ice initial condition in the Arctic, we assimilate the satellite-derived sea ice concentrations within the Data Assimilation Research Testbed (DART) system, based on the ensemble Kalman filter (EnKF), coupled with the sea ice model in the Community Earth System Model (CESM). The EnKF-based assimilation results show that the Arctic sea ice initial condition is significantly improved by assimilating the satellite-derived sea ice concentration data. However, during the Arctic sea ice freezing season, the assimilation impact tends to be degraded due to the reduction of the ensemble spread within the EnKF-based assimilation system. To counteract the ensemble spread reduction, we apply the randomized-dormant ensemble Kalman filter (RD-EnKF) method in which the model backgrounds are more perturbed by leaving the dormant ensemble members out of the total ensemble members from the analysis update, inflating the ensemble spread. Compared with the assimilation results using the EnKF method, the additional analysis benefits are obtained due to the increment of the ensemble spread derived by applying the RD-EnKF method, in particular, during the Arctic sea ice freezing season.

Ensemble Kalman Filters (EnKFs), which assimilate observations based on statistics derived from samples of ocean states called ensemble, have become the norm for ocean data assimilation (DA) and forecasting. These schemes are commonly implemented with inflation and localization techniques to increase their ensemble spread and to filter out spurious long-range correlations resulting from the limited-size ensembles imposed by computational burden constraints. Such ad hoc methods were found not necessary in ensemble DA experiments with simplified ocean/atmospheric models and large ensembles. Here, we conduct a series of 1-year-long ensemble experiments with a fully realistic EnKF-DA system in the Red Sea using tens-to-thousands of ensemble members. The system assimilates satellite and in-situ observations and accounts for model uncertainties by integrating a 4km-resolution ocean model with ECMWF atmospheric ensemble fields, perturbed internal physics and initial conditions for forecasting. Our results indicate that accounting for model uncertainties is more beneficial than simply increasing the ensemble size, with the improvements due to large ensemble leveling off at about 250 members. Besides, and in contrast to what is commonly observed with simplified models, the investigated ensemble DA system still required localization even when implemented with thousands of members. These findings are explained by (i) amplified spurious long-range correlations produced by the low-rank nature of the ECMWF atmospheric forcing ensemble, and (ii) non-Gaussianity generated by the perturbed internal physical parameterization schemes. Large ensemble forcing fields and non-Gaussian DA methods might be needed to take full benefits from large ensembles in ocean DA.

Society’s ability to make wise decisions depends onan accurate understanding of the current state of Earthand on an ability to predict future states.The Data Assimilation Research Testbed (DART) is an example of a suite of toolsdesigned to improve our understanding through the combination of observationswith our theoretical understanding embodied in forecast models.DART’s ensemble based data assimilation provides uncertainty quantification as a function of time, location, and variable.Current research using DART includes: Improving streamflow prediction during intense rainfall events, which lead to flooding, using DART and the Weather Research and Forecasting model and the Noah-MP land model (WRF-Hydro). Building an integrated atmosphere and ocean forecasting system using DART and WRF for the Red Sea Initiative. Understanding air pollution using a global meteorology-aerosol-chemistry prediction system that assimilates aerosol optical depth, carbon monoxide, and weather observations into the Community Atmosphere Model with Chemistry (CAM-Chem). Assimilating observations of the Earth system from satellites into the Model for Prediction Across Scales (MPAS; regional and global) using observation operators from the Joint Effort for Data assimilation Integration (JEDI), bias correction for satellite retrievals from the Gridpoint Statistical Interpolation (GSI), and the assimilation environment of DART. Deciphering the flow dependency of forecast errors in the tropics and the relative importance of wind and mass information for tropical analyses. Connecting the U.S. Department of Energy’s E3SM atmospheric model with a broad spectrum of observations to perform short ensemble hindcast simulations for model development and evaluation. Generating atmospheric reanalysis data sets from CAM, which enables efficient data assimilation in other components of the Earth system; ocean, land, cryosphere, … Improving DART by giving users more control over how observations are assimilated, and supporting the assimilation of additional observations, such as radiances through the use of the RTTOV software.