Sub-seasonal prediction skill for the stratospheric meridional mass circulation variability in CFSv2

This study evaluates the prediction skill for the stratospheric mass circulation variability in winter (November–March) in the Climate Forecast System, version 2 (CFSv2), from 2011 to 2018. Three stratospheric mass circulation indices measuring meridional mass transport into the polar stratosphere by the total flow (ST60N), wavenumber-1 (ST60N_W1), and wavenumber-2 waves (ST60N_W2) are considered. The variability of ST60N is mainly contributed by ST60N_W1 and ST60N_W2, and these indices are good indicators of the timing and locations of the continental-scale cold-air outbreaks (CAOs) at mid-latitudes. Thus, their potentially useful prediction skill can be utilized to make sub-seasonal forecasts of CAOs in a dynamical and statistical hybrid paradigm. Systematic forecast bias is found in both the 7-year averaged winter mean and seasonal cycle, which is tied to the overestimation of damping in amplitude and westward tilting variations of total waves and difficulties in forecasting the exact contributions from different spatial scales of waves. The intraseasonal variations of stratospheric mass circulation indices, with the systematic forecast bias corrected, can be modestly predicted at a forecast lead time of about 20 days, in terms of both the anomaly value and timing of negative and positive peak events. The sub-seasonal predictability of the ST60N value mainly comes from the ST60N_W1, while the predictability of the timing of positive and negative peaks comes from the ST60N_W2. The 20-day prediction limit of the stratospheric mass circulation indices is mainly due to the 2-week limit of the CFSv2 model in predicting the variability of anomalous wave tilt angle, whereas the prediction limit of the wave amplitude anomaly can exceed 50 days.