Attributions to Errors in Air-Sea Fluxes in Numerical Model Products

Air-sea fluxes of energy, momentum, and gases from numerical model output (data assimilation products and simulations/predictions) have rarely been validated because of a lack of in situ observations. Observations from recent deployments of Saildrone uncrewed surface vehicles (USVs) allow estimates of sea surface fluxes in regions where no observed fluxes were available before. In this study, observations from saildrones are used to validate surface latent and sensible heat fluxes in ensemble forecasts of ECMWF and three global reanalysis products (MERRA2, ERA5, CFSR2) in the Pacific Arctic and subtropical Atlantic oceans. Attributions to errors in surface fluxes in these numerical products are quantified in terms of errors in the surface state variables (temperature, humidity, wind). These attributions share similarities and differences between the prediction and reanalysis products and between the Arctic and Atlantic oceans. The results point to surface air humidity and temperature in the numerical products as the main sources of errors in sea surface latent and sensible heat fluxes.