A comparative study of the sensitivity of an ocean model outputs to atmospheric forcing: ERA-Interim vs. ERA5 for Adriatic Sea Ocean modelling

Abstract

Advancements in atmospheric data have the potential to improve the accuracy of ocean modeling, as these models rely heavily on atmospheric parameters as key forcing inputs. One such dataset is the ECMWF reanalysis, with ERA5 being the latest version, succeeding ERA-Interim (ERA-I or ERAI). However, limited research has explored whether ERA5 improves ocean model accuracy compared to ERA-I. We use the ROMS model on the Adriatic Sea under two atmospheric forcing scenarios: ERA-I and ERA5. Results show that ERA5 calculates higher temperature and salinity values than ERA-I. ERA5 shows better alignment with satellite and Mediterranean reanalysis data than ERA-I. For temperature, ERA5 has a higher bias range (–2.29℃ to 0.83℃) compared to ERA-I (–2.34℃ to 0.80℃) and achieves a lower minimum bias, particularly in summer (0.02℃). Against Mediterranean reanalysis data, ERA5’s temperature bias range (–2.06℃ to 1.54℃) is lower range than ERA-I’s (–3.14℃ to 1.51℃). For salinity, ERA5 also has a smaller bias range (–0.02 PSU to 0.27 PSU) and achieves zero bias in spring, indicating a more accurate seasonal alignment than ERA-I. The warmer water temperatures in ERA5 are attributed to higher values of atmospheric parameters such as shortwave radiation flux, sensible heat flux, and air temperature, while, increased salinity is linked to more negative latent heat flux up to 10 W/m², longwave radiation up to 5 W/m², and higher wind speeds. These factors collectively lead to improved ocean modeling performance in ERA5.