A real-time ocean reanalyses intercomparison project in the context of tropical pacific observing system and ENSO monitoring

作者：Xue Yan^*; Wen C; Kumar A; Balmaseda M; Fujii Y; Alves O; Martin M; Yang X; Vernieres G; Desportes C; Lee T; Ascione I; Gudgel R; Ishikawa I

来源：Climate Dynamics, 2017, 49(11-12): 3647-3672.

DOI：10.1007/s00382-017-3535-y

• Ocean reanalysis intercomparison
• Ocean data assimilation system
• Ocean initialization
• Tropical Pacific observing system
• TAO/TRITON
• ENSO monitoring and prediction
• Argo floats
• Ensemble approach
• Signal to noise ratio

An ensemble of nine operational ocean reanalyses (ORAs) is now routinely collected, and is used to monitor the consistency across the tropical Pacific temperature analyses in real-time in support of ENSO monitoring, diagnostics, and prediction. The ensemble approach allows a more reliable estimate of the signal as well as an estimation of the noise among analyses. The real-time estimation of signal-to-noise ratio assists the prediction of ENSO. The ensemble approach also enables us to estimate the impact of the Tropical Pacific Observing System (TPOS) on the estimation of ENSO-related oceanic indicators. The ensemble mean is shown to have a better accuracy than individual ORAs, suggesting the ensemble approach is an effective tool to reduce uncertainties in temperature analysis for ENSO. The ensemble spread, as a measure of uncertainties in ORAs, is shown to be partially linked to the data counts of in situ observations. Despite the constraints by TPOS data, uncertainties in ORAs are still large in the northwestern tropical Pacific, in the SPCZ region, as well as in the central and northeastern tropical Pacific. The uncertainties in total temperature reduced significantly in 2015 due to the recovery of the TAO/TRITON array to approach the value before the TAO crisis in 2012. However, the uncertainties in anomalous temperature remained much higher than the pre-2012 value, probably due to uncertainties in the reference climatology. This highlights the importance of the long-term stability of the observing system for anomaly monitoring. The current data assimilation systems tend to constrain the solution very locally near the buoy sites, potentially damaging the larger-scale dynamical consistency. So there is an urgent need to improve data assimilation systems so that they can optimize the observation information from TPOS and contribute to improved ENSO prediction.

• 出版日期2017-12
• 单位中国气象科学研究院