Evaluation of the INM RAS climate model skill in climate indices and stratospheric anomalies on seasonal timescale

Abstract

The study of winter seasonal predictability with the climate model INM-CM5-0 is presented. Initial conditions were produced using ERA-Interim reanalysis data for atmosphere, SODA3.4.2 reanalysis data for ocean and the bias-correction algorithm. The seasonal 5-month re-forecasts consisting of 10 ensemble members with small initial condition perturbations for each year over the 35-yr period are conducted. A comparison of the multiyear mean winter averaged anomaly correlation for basic variables in several regions with similar results of SLAV model was conducted. An increase in the anomaly correlation for the years with El Niño and La Niña events was shown. The predictability of NAO and PNA indices was studied. INM-CM5-0 provides very high skill in predicting the winter NAO (correlation coefficient of 0.71 with ERA-Interim reanalysis and 0.68 with instrumental CRU data for 1991–2010). It was shown, that the stratospheric variability provides a significant contribution, although potentially is not the only cause of model high skill in NAO index predictability. Correlation coefficients for PNA index in December-February is 0.60. In the years of the most pronounced El Niño the values of PNA index have significantly positive values, and for La Niña years they are noticeably less than zero.