Premium
An objective and automatic method for identification of pattern changes in wind direction time series
Author(s) -
Klausner Ziv,
Fattal Eyal
Publication year - 2011
Publication title -
international journal of climatology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.58
H-Index - 166
eISSN - 1097-0088
pISSN - 0899-8418
DOI - 10.1002/joc.2100
Subject(s) - smoothing , series (stratigraphy) , maxima and minima , computer science , identification (biology) , wind speed , time series , parametric statistics , wind direction , algorithm , mathematics , meteorology , statistics , geology , geography , machine learning , paleontology , botany , biology , mathematical analysis
Finding the point in time at which the wind pattern changes its direction is an important problem with a broad scope of application in atmospheric boundary layer modelling. In many cases such a change of pattern can be extracted using subjective methods, which involves a subjective post analysis of the measured time series. A considerably more difficult task is identifying such a change using an objective and automated method. This task is usually treated by smoothing the original time series, extracting threshold parameters using a statistical analysis of the past wind data, and setting threshold values via trial and error. We present here an objective and automatic method for solving this problem. The method identifies extrema and bending points in the cumulative sum terms of sine and cosine of the wind direction time series, which are based on the directional cumulative sum principle. Our method is non‐parametric, and needs neither prior calibration of threshold values nor smoothing of the data. Furthermore, our method is unique in its ability to detect the pattern change even if the direction change occurs gradually. An example of the application of the method is presented using 6 months of data collected at a single meteorological station located on the central coastal plain of Israel. The method was applied to identify the timing of the occurrence of the nocturnal land wind period (beginning and end). Out of 176 days examined, the method identified correctly more than 90% of the periods. Copyright © 2010 Royal Meteorological Society