Statistical Analysis of Categorical Time Series of Atmospheric Elementary Circulation Mechanisms - Dzerdzeevski Classification for the Northern Hemisphere.

Mihael Brenčič,Inés Álvarez

doi:10.1371/journal.pone.0154368

Abstract

Northern hemisphere elementary circulation mechanisms, defined with the Dzerdzeevski classification and published on a daily basis from 1899–2012, are analysed with statistical methods as continuous categorical time series. Classification consists of 41 elementary circulation mechanisms (ECM), which are assigned to calendar days. Empirical marginal probabilities of each ECM were determined. Seasonality and the periodicity effect were investigated with moving dispersion filters and randomisation procedure on the ECM categories as well as with the time analyses of the ECM mode. The time series were determined as being non-stationary with strong time-dependent trends. During the investigated period, periodicity interchanges with periods when no seasonality is present. In the time series structure, the strongest division is visible at the milestone of 1986, showing that the atmospheric circulation pattern reflected in the ECM has significantly changed. This change is result of the change in the frequency of ECM categories; before 1986, the appearance of ECM was more diverse, and afterwards fewer ECMs appear. The statistical approach applied to the categorical climatic time series opens up new potential insight into climate variability and change studies that have to be performed in the future.

Highlights

Weather and climate can be characterised only by a great number of physical parameters forming a complex picture that is difficult to comprehend
The first is represented by elementary circulation mechanisms (ECM) categories 13s, 11a, 13w, 12a, 12bw and 10a and the second is represented by the rest of the ECM categories
Category missing represents a small share at the end of the diagram, showing that it is not part of the patterns and trends in the whole ECM time series

Summary

Introduction

Weather and climate can be characterised only by a great number of physical parameters (e.g. air temperature, air pressure, humidity etc.) forming a complex picture that is difficult to comprehend. These parameters are characteristic for particular points in atmospheric space, forming extensive scalar and vector fields that result in air mass movements and air distribution patterns. Complete characterisation based on the entire set of relevant physical parameters is never possible, and other approaches must be applied to characterise the status of the atmosphere in a particular region.

Results

Discussion

Conclusion