Application of the nudging technique to produce initial states for the INM RAS climate model seasonal hindcasts

This study examines the implementation of nudging in the INM RAS global coupled climate model to improve the initial states for seasonal hindcasts. We compare results of three nudging experiments with both reanalyses data and ensemble of CMIP6 historical experiments. The results show that nudging significantly reduces model biases in atmosphere and ocean fields. The experiment with nudging excluding lower atmospheric levels yields the best performance by minimizing the impact of the relaxation procedures on the model physics in the atmosphere boundary layer. We compare the quality of the hindcasts for November–March obtained with different initialization techniques. The initialization approach using nudging outperforms other initialization methods for hindcasts of the Northern Hemisphere winter season with one-month lead time. However, for the first forecast month the full field initialization demonstrates the best results. The study highlights that nudging in the ocean model is essential to maintain hindcast skill over longer lead times and proposes nudging as a promising approach for initialization of the annual-to-decadal climate predictions.