Ensemble data assimilation in the Red Sea: sensitivity to ensemble selection and atmospheric forcing

Habib Toye, Peng Zhan, Ganesh Gopalakrishnan, Aditya R. Kartadikaria, Huang Huang, Omar Knio, Ibrahim Hoteit

Research output: Contribution to journalArticlepeer-review

14 Scopus citations

Abstract

We present our efforts to build an ensemble data assimilation and forecasting system for the Red Sea. The system consists of the high-resolution Massachusetts Institute of Technology general circulation model (MITgcm) to simulate ocean circulation and of the Data Research Testbed (DART) for ensemble data assimilation. DART has been configured to integrate all members of an ensemble adjustment Kalman filter (EAKF) in parallel, based on which we adapted the ensemble operations in DART to use an invariant ensemble, i.e., an ensemble Optimal Interpolation (EnOI) algorithm. This approach requires only single forward model integration in the forecast step and therefore saves substantial computational cost. To deal with the strong seasonal variability of the Red Sea, the EnOI ensemble is then seasonally selected from a climatology of long-term model outputs. Observations of remote sensing sea surface height (SSH) and sea surface temperature (SST) are assimilated every 3 days. Real-time atmospheric fields from the National Center for Environmental Prediction (NCEP) and the European Center for Medium-Range Weather Forecasts (ECMWF) are used as forcing in different assimilation experiments. We investigate the behaviors of the EAKF and (seasonal-) EnOI and compare their performances for assimilating and forecasting the circulation of the Red Sea. We further assess the sensitivity of the assimilation system to various filtering parameters (ensemble size, inflation) and atmospheric forcing.
Original languageEnglish (US)
Pages (from-to)915-933
Number of pages19
JournalOcean Dynamics
Volume67
Issue number7
DOIs
StatePublished - May 26 2017

Fingerprint Dive into the research topics of 'Ensemble data assimilation in the Red Sea: sensitivity to ensemble selection and atmospheric forcing'. Together they form a unique fingerprint.

Cite this