Changes in Apparent Temperature Over the Ocean during 1950-2023: Long-term Trends and Contributions of Meteorological Factors

Changes in apparent temperature (APT), due to global warming can have a significant impact on human health and society. A recent study based on ERA5 reanalysis data and HadISD observations post 1979 has shown that the probability of extreme heat discomfort events over the Northern Hemisphere oceans has significantly increased, with the largest changes occurring in the tropics and parts of the Arctic. However, current APT studies have mostly focused on changes on land, while changes in marine APT (MAPT), which pose a potential threat to ocean-based economies and living conditions of people dependent on the sea, still remain unclear to a large extent. We examine historical changes in MAPT and the dominant role of different meteorological factors by separating their contribution to changes during 1950-2023.

The heat stress indicator - APT, is a function of surface air temperature, relative humidity (RH) and near-surface wind speed (NSWS) and is computed from 20CRv3-ERA5 reanalysis monthly average data. MAPT and MAT warm at 0.109±0.011 and 0.122±0.015 ℃/10a respectively. MAPT warms faster than MAT but with similar spatial patterns. We estimate contributions from changes in MAT, RH and NSWS to changes in MAPT using a linear sensitivity analysis. Global MAPT changes are mainly dominated by the change of MAT, RH has little influence and NSWS has some regional influence especially in parts of the equatorial Pacific and the Southern Ocean.

To explore whether reanalysis changes could be reproduced by CMIP6 simulations, contribution by changes in meteorological factors to changes in MAPT from REA and CMIP6 were computed. CMIP6 is consistent with reanalysis with both showing the dominant contribution is from changes MAT, but there are some differences in the spatial pattern of the RH contribution, which only has a small influence and NSWS which may impact regional change. Reanalysis-CMIP6 differences of MAT as well as differences of RH and NSWS in most mid-high latitudes are consistent with internal variability, while differences of RH and NSWS in low latitudes are beyond the range of internal variability. The consistency between reanalysis MAPT/MAT & CMIP-6 multi-model ensemble means we could attribute the bulk of the ‘obs’ changes to anthropogenic climate change.