5:15 PM - 7:15 PM
[ACG40-P05] Effect of data assimilation on AMOC and its variability
Keywords:data assimulation, AMOC, Ocean-sea ice model
The Global Thermohaline Circulation driven by density differences is a global-scale ocean circulation that connects the surface and the deep layer. The Atlantic meridional component of the circulation is called the Atlantic Meridional Overturning Circulation (AMOC), driven by the deep-water formation site in the North Atlantic. The AMOC and its variability significantly influence climate through heat transport. Consequently, many studies use ocean reanalysis or model simulations to investigate the relevant topics and overcome the poor temporal and spatial resolution from observations. However, there are few studies to notice the impact of data assimilation (DA) on the AMOC simulated by models.
In this study, we assess the representation of AMOC and the variability characteristics in two hindcast simulations – one with DA and the other without. Based on the Ensemble Kalman Filter method, a state-of-the-art data assimilation system for the North Atlantic and the Arctic (TOPAZ4) assimilated all traditional ocean and sea ice observations from 1991 to 2019 (RA). TOPAZ4 uses version 2.2 of the Hybrid Coordinate Ocean Model (HYCOM) coupled to a simple one-category sea ice model. Being a control run (FR), the model was freely driven by the atmospheric forcing from ERA5 in 1991-2019, after 20 years of spin-up beginning at the rest state. The mean state of AMOC of both experiments and the time series of the annual mean AMOC of them are in the attached figures. The AMOC of both runs is underestimated compared with that observed by RAPID. After 2006, when RAPID observations are available, the mean AMOC in FR is 11.1 Sv and 10.3 Sv for RA.
However, the correlation of AMOC between RA and RAPID has increased slightly relative to that between FR and RAPID. In addition to the AMOC strength, other AMOC characteristics, such as the deep convection strength and the northward heat transport, are compared in the two runs. Although compared with the long timescale of AMOC, the period in this study is limited, it still reveals the change characteristics affected by DA, which helps to improve the understanding of the influence of DA on AMOC
In this study, we assess the representation of AMOC and the variability characteristics in two hindcast simulations – one with DA and the other without. Based on the Ensemble Kalman Filter method, a state-of-the-art data assimilation system for the North Atlantic and the Arctic (TOPAZ4) assimilated all traditional ocean and sea ice observations from 1991 to 2019 (RA). TOPAZ4 uses version 2.2 of the Hybrid Coordinate Ocean Model (HYCOM) coupled to a simple one-category sea ice model. Being a control run (FR), the model was freely driven by the atmospheric forcing from ERA5 in 1991-2019, after 20 years of spin-up beginning at the rest state. The mean state of AMOC of both experiments and the time series of the annual mean AMOC of them are in the attached figures. The AMOC of both runs is underestimated compared with that observed by RAPID. After 2006, when RAPID observations are available, the mean AMOC in FR is 11.1 Sv and 10.3 Sv for RA.
However, the correlation of AMOC between RA and RAPID has increased slightly relative to that between FR and RAPID. In addition to the AMOC strength, other AMOC characteristics, such as the deep convection strength and the northward heat transport, are compared in the two runs. Although compared with the long timescale of AMOC, the period in this study is limited, it still reveals the change characteristics affected by DA, which helps to improve the understanding of the influence of DA on AMOC