Abstract
This study investigates the impact of dynamical representational error (RE) on the analysis of an ocean ensemble Kalman filter‐based data assimilation system, LETKF‐ROMS (Local Ensemble Transform Kalman Filter – Regional Ocean Modelling System) configured for the Indian Ocean and assimilating in situ temperature and salinity observations from Argo. Three different approaches to account for the RE are studied and inter‐compared: (a) static RE (varies in horizontal and vertical direction), (b) dynamic RE (varies in space and time) estimated from concurrent observations, and (c) dynamic RE estimated using concurrent high‐resolution model outputs. RE estimated from the model outputs exhibits rich spatial and temporal variability with an estimated temporal mean RE for temperature below 0.5 and 0.2 °C in the surface and deep layers, respectively, and reaching up to 1 °C in the thermocline layers. The region encompassing the Great Whirl displays a large seasonal variability reaching up to 0.8 °C, and the South Equatorial Current (SEC) a large interannual variability reaching up to 0.4 °C. Neglecting such spatio‐temporal variations of RE and assimilating with a static RE limited the benefits of assimilation by entertaining over‐fitting issues that caused degradations in the Bay of Bengal, the western parts of the Arabian Sea, and the equatorial Indian Ocean. Assimilating with the observations‐based dynamic RE improved the results in these regions, but the best performances were obtained with the configuration using the model‐based dynamic RE, which yielded further improvements, e.g. reduction of sea surface height root‐mean‐square errors reaches 30% with respect to the observations‐based dynamic RE. The latter also better handled the rich spatial variability regions and areas not well sampled by the observations. Improved estimates of the spatial and temporal variations of RE helped to better exploit the assimilated observations and provided enhanced analyses less prone to assimilation shocks.
Three different approaches for accounting for the representational error (RE): static RE, observations‐based dynamic RE, and model‐based dynamic RE, in an ocean ensemble data assimilation system are investigated. Our results reveal that static RE limits the benefits of the assimilated observations. Observations‐based dynamic RE improves the assimilation system results compared to static RE, but is not very robust in under‐sampled regions. Model‐based dynamic RE mitigates the under‐sampling issues and significantly enhances the quality of the assimilation system outputs. Improvements, in terms of root‐mean‐square‐differences, with respect to the other two approaches (presented in Figure G1), reach up to 50%, particularly in the regions characterized by strong variability and sparse observation coverage, such as the south Indian Ocean.Figure G1. Spatial structures of sea level anomaly (SLA) improvements (negative values) and degradations (positive values), in terms of percentage root‐mean‐square differences, in (a) model‐based dynamic RE with respect to static RE approach, and (b) model‐based dynamic RE with respect to observations‐based dynamic RE approach. Root‐mean‐square differences are estimated with respect to AVISO SLA data. Statistics are based on 1 July 2008 to 31 December 2010