Consequences of different air-sea feedbacks on ocean using MITgcm and MERRA-2 forcing: Implications for coupled data assimilation systems. (December 2018)

• Record Type: Journal Article
• Authors: Strobach, Ehud; Molod, Andrea; Forget, Gael; Campin, Jean-Michel; Hill, Chris; Menemenlis, Dimitris; Heimbach, Patrick
• Is Part Of: Ocean modelling. Volume 132(2018)
• Page Start: 91
• Page End: 111
• Publication Date: 2018-12
• Subjects: Ocean modeling -- Bulk formulae -- Atmospheric reanalysis -- Ocean reanalysis; Oceanography -- Periodicals; Océanographie -- Périodiques; Oceanography; Periodicals; 551.46
• DOI: 10.1016/j.ocemod.2018.10.006
• Languages: English
• ISSNs: 1463-5003

Abstract

Highlights: Comparison between MERRA-2 and ECCO-v4 net surface heat fluxes. Feedbacks between the ocean and the atmosphere. MITgcm forced with MERRA-2 fluxes reduces SST by 2.5°C compared to ECCO-v4. MITgcm forced with MERRA-2 state variables reduces SSS by 0.172psu compared to ECCO-v4. Implications for coupled data assimilation are discussed Abstract: Ocean surface flux estimates from atmospheric and oceanic reanalyses contain errors that compensate for inaccuracies in the respective atmosphere and ocean models used to generate these reanalyses. A conundrum for climate studies is the discrepancy between surface fluxes that minimize model-data differences for an atmosphere-only model vs surface fluxes that minimize model-data differences for an ocean model. As a first step towards a consistent coupled ocean-atmosphere data-assimilation (DA) system, we compare surface net heat flux from a state-of-the-art atmospheric reanalysis, the Modern-Era Retrospective analysis for Research and Applications, Version 2 (MERRA-2), to net heat flux from a state-of-the-art ocean state estimate, the Estimating the Circulation and Climate of the Ocean Version 4 (ECCO-v4). The possible impacts of the MERRA-2 and ECCO-v4 air-sea net heat flux difference in a coupled DA system were assessed using a set of experiments designed to imitate different "flavors" of a coupled DA system in an ocean-only setup. This was done by forcing the ECCO-v4 underlying ocean model - the Massachusetts Institute of Technology general circulation model (MITgcm) - with different sets of MERRA-2 fields and utilizing different forcing methods. By doing so we were able to turn off different air-sea feedbacks which, in a coupled DA setup, are partially muted by the constraining observations. The set of experiments, therefore, represents a range of active feedbacks in different "flavors" of coupled data-assimilation systems. For the period 1992–2011, MERRA-2 net heat flux has a global mean difference of − 4.9 Wm − 2 relative to ECCO-v4. When MERRA-2 surface fields are used to force MITgcm, imbalances in the energy and the hydrological cycles of MERRA-2, which are directly related to the fact that MERRA-2 was created without an interactive ocean, propagate to the ocean. The experiment in which MITgcm is forced with MERRA-2 fluxes (MERRA-2-flux experiment) results in a 2.5°C global mean Sea Surface Temperature (SST) cooling, a 1m reduction in global mean sea level, and other drastic changes in the large scale ocean circulation relative to those resulting when the MITgcm is forced with the optimized ECCO-v4 net heat flux (the ECCO-v4 experiment itself). When MITgcm is forced with MERRA-2 state variables (MERRA-2-state experiment), the SST is somewhat restored to the observed SST, but the errors are shifted to the water cycle, resulting in a global mean sea level increase of 2.7 m. To further explore the pros and cons of these two approaches, we introduce a new intermediate forcing method in which the ocean is forced with turbulent fluxes but has a long wave feedback. This method, unlike MERRA-2 state, preserves the MERRA-2 water and salinity cycles, and it reduces the SST error compared to the MERRA-2-flux experiment, but the SST is not as good as that in the MERRA-2-state experiment. Our results have implications for ocean-model forcing recipes and clearly reveal the undesirable consequences of limiting the feedbacks in either these types of experiments or in coupled DA.