Blocking predictability in operational medium-range ensemble forecasts

Abstract

This study assesses the forecasting performance of operational medium-range ensemble forecasts: BOM, CMA, CMC, CPTEC, ECMWF, JMA, KMA, NCEP, and UKMO, in terms of atmospheric blocking during DJF (December-January-February) of 2006/07, 2007/08, and 2008/09. The state-of-the-art medium-range ensemble forecasts performed well in simulating the frequencies of Euro- Atlantic (EA) and Pacific (PA) blockings, even after 216-hr lead time, whereas they did not simulate well the frequencies of Greenland and Ural blockings, even in the middle of the forecast range. The ensemble forecasts are not always able to capture the blockings with high probability in the latter half of the forecast range. During this latter half, blocked flows were frequently predicted with low probability during the active blocking period, whereas they were seldom predicted with similar probability during the nonactive blocking period. This result might suggest that the active blocking period is more chaotic than the non-active blocking period. In addition, it was more difficult to predict an onset of EA blocking than to predict an onset of PA blocking, and probabilistic blocking forecasting over the PA sector was more skillful than that over the EA sector. These results suggest that PA blocking has a higher predictability than does AT blocking.