ANALYSIS OF DAILY TEMPERATURES DATA IN TÂRGU MUREŞ FROM THE PERIOD 1951-2014
OTTILIA RUSZ
ABSTRACT. – Analysis of daily temperatures data in Târgu Mureş from the period 1951-2014. Statistical analysis of 365 x 4 time series (daily mean temperatures DMT, daily maximum temperatures TN, daily minimum temperatures TX and diurnal temperature ranges DTR), each content 64 values (1951-2014) were made from data registered at Târgu Mureş meteorological station. In case of DMT standard deviations of summer days are less than of winter days, while in case of DTR spring and autumn days have greater values of this parameter. Kolmogorov-Smirnov normality tests reveal that in almost cases time series follow a normal distribution, but some other test (e.g. Shapiro-Wilk) show that in winter there are time series of days in which case one cannot assume normality. By far most outliers were found in winter days (e.g. 15 outliers were registered in 1985, which is one of the coldest years). Linear trend was analysed using Mann-Kendall test. In case of DMT statistical significant increasing trends are present in greater numbers in days of warm months (e.g. in August there are 10 such days), which is similar to monthly linear trends. In order to decide whether a time series have constant variance or not, range-mean plots were created. Mainly downward sloping lines are present, e.g. in January (DMT) there are 15 days when the variance is decreasing in the mean. Two parts regressions have different points (year) when the regression line is changing, but usually the second line has an upward slope. In case of extreme temperatures (TN and TX) 1st and 3rd quartiles were determined, and analysed how the number of values greater respectively less than the quartiles is distributed in time. So for example, in case of summer days (TN) two thirds of values greater than 3rd quartile are placed in the second parts of studied periods, while in winter half parts of values greater than the 1st quartile are placed in the last 32 years. Analyses from this paper were made using the following statistical software: XLSTAT, Gretl, AnClim and Microsoft Excel.
Keywords: daily temperatures, Targu Mures, outliers, trends, quartiles