Changes in air-sea fluxes over the North Atlantic during 1950-2019 as derived from ERA5 data

Air-sea heat fluxes play a key role for many processes in the North Atlantic Ocean, such as the lateral transport of energy or the formation of storm tracks. Thus, an accurate estimation of air-sea heat flux trends is pivotal and helps to understand implications of climate change. To do so, reanalysis products are attractive candidates due to their excellent spatial coverage over multiple decades. However, trend estimations based on reanalysis data are challenging as changes in the observing system can introduce temporal discontinuities.

In this study, we explore the reliability and temporal stability of net air-sea heat flux trends from ERA5 forecasts in the North Atlantic basin over the period 1950-2019.  The assessment is complemented with an indirect estimate of the net surface flux derived from the atmospheric energy budget. Causes of trends in latent and sensible heat fluxes are identified based on monthly analyzed state quantities from ERA5, such as wind speed, moisture, and temperature. Additionally, the impact of the North Atlantic Oscillation (NAO) and Atlantic Multi-decadal Oscillation (AMO) as well as analysis increments, as introduced by the ERA5 data assimilation, is investigated.  

Our results show a robust increase of latent heat fluxes in the tropical North Atlantic over the past seven decades, which is likely caused by the intensification of the Hadley cell favouring subsidence and advection of drier air masses. In the Norwegian Sea, positive net air-sea heat flux trends (increased ocean heat uptake) are largely dominated by changes in sensible heat fluxes, which are driven by a trend towards more southerly winds and the advection of warmer air. In the Gulf Stream region, the AMO likely drives the multi-decadal variability of net air-sea heat fluxes, while long-term trends over the 1950-2019 period remain insignificantly small. Furthermore, we find significant changes over the North Atlantic Warming Hole and western North Atlantic associated with more frequent positive NAO phases during the past 30 years. From our analysis, we conclude that analysis increments most likely influence the magnitude of these trends, especially at low latitudes where the impact can be as large as ~2 W m^-2 dec^-1, while the basin-wide trend pattern remains unaffected. The net effect of the found regional changes in fluxes is assessed by the spatial average trend over the whole North Atlantic north of 26°N, which yields a positive but statistically insignificant trend of 0.5 W m^-2 dec^-1 over the past 70 years. Potential implications for trends in the AMOC are discussed.