Extratropical response to Fast and Slow episodes of Madden-Julian Oscillation in observation and using intervention experiments with CFSv2

The Madden-Julian Oscillation (MJO) is a potential source of predictability in the extratropics in extended range weather forecasting. The nature of MJO is sporadic and therefore, the mid-latitude response may depend on the nature of the MJO event, in particular the phase speed. We discuss the results of our observational and modeling study of mid-latitude circulation response to Fast and Slow MJO episodes using wintertime ERA-Interim reanalysis data and the CFSv2 coupled model of NOAA. The observational study shows that the mid-latitude response to different propagating speeds is not the same. The propagation speed is defined by the time the OLR takes to propagate from phase 3 to phase 6. The mid-latitude response is assessed in terms of composite maps and frequency of occurrence of robust circulation regimes. Fast episode composite anomalies of 500hPa height show a developing Rossby wave in the mid-Pacific with downstream propagation through MJO phases 2- 4. Development of NAO+ teleconnection pattern is stronger in Slow that in Fast MJO episodes, and occurs with a greater time lag after MJO heating is in the Indian Ocean (phase 3). Previous results find an increase in occurrence of NAO- regime following phase 6. We have found that much of this behavior is due to the slow episodes. Based on these observational results, intervention experiments using CFSv2 are designed to better understand the impact of heating/cooling and to estimate mid-latitude response to Fast and Slow MJO episodes. The added heating experiments consist of 31 year reforecasts for December 1 initial conditions from CFS reanalysis (1980-2011) in which the identical MJO evolution of three-dimensional diabatic heating has been added, thus producing fast and slow MJO episodes with well-defined phase speeds. We will discuss the results of these experiments with a focus on understanding the role of phase speed and interference in setting up the response, and to understand the mechanisms that distinguish fast and slow types of response We will also discuss the diagnostics using Predictable Component Analysis to distinguish the signal forced by common diabatic heating signal from noise, and weather regime response to fast and slow MJO using cluster analysis.