Impact of non-Gaussian statistics of atmospheric variables on extreme intramonth anomalies

Abstract

© 2017, Pleiades Publishing, Ltd. The analysis of asymmetry of probability distribution functions (PDF) is carried out for key atmospheric variables using the JRA-55 reanalysis data in the troposphere of the Northern Hemisphere for 1976–2014. The nonzero asymmetry of the PDF indicates the deviation of the PDF from the normal distribution. The analysis was carried out for two time-scale intervals: synoptic variability (SV) of 2–7 days and low-frequency variability (LV) of 9–30 days. Statistically significant deviations from the normal probability distribution occur in the regions of the most frequent formation of atmospheric baroclinic perturbations, i.e., over the western parts of the oceans in midlatitudes and downstream in the atmosphere. In the SV time-scale interval, a negative asymmetry of the vertical velocity is revealed in isobaric coordinates for the entire thickness of the free troposphere, which agrees with the overall dominance of cyclonic anomalies in this interval of time scales. In the LV interval, the asymmetry of this variable in the entire free troposphere is positive, which indicates the dominance of anticyclonic anomalies at these time scales. For the zonal velocity, temperature, and geopotential, the asymmetry sign of the PDF for variability with time scales of 2–7 days is different for the upper and lower free troposphere. The asymmetry of the PDF for atmospheric variables indicates the important role of the intermode interaction in the formation of baroclinic perturbations. The corresponding deviations of synoptic variability from the normal distribution, which is found in the upper troposphere of the subpolar and polar latitudes, can be related to the interaction of these perturbations with the winter polar vortex. These deviations of PDF from the normal distribution substantially increase the probability of the appearance of large (in absolute value) anomalies as compared to the case of the Gaussian PDF.