Machine Learning Eliminates Reanalysis Warm Bias and Reveals Weaker Winter Surface Cooling over Arctic Sea Ice

Authors:
  • POP Ola
    P Ola

Abstract
The surface energy budget of the Arctic Ocean governs sea ice growth and melt. Correctly representing it in models is a prerequisite for understanding and projecting Arctic climate change. Direct observations of surface fluxes are scarce, and reanalysis datasets suffer from systematic biases. Here, we train a neural network with observational data to bias-correct hourly ERA5 reanalysis surface fluxes. The neural network reduces the RMSE for net shortwave radiation by ~40%, downward longwave radiation by ~16% and the total surface energy budget by ~55%. It eliminates the wintertime warm bias of ~4K in ERA5 and reduces wintertime surface cooling by ~50%, in agreement with independent satellite-observed sea ice growth rates. Our bias-corrected data capture the observed clear and cloudy states of the Arctic winter boundary layer. The bias-corrected data can be used to study the Arctic climatology, evaluate climate models and drive sea-ice-ocean models.