engleski

Temporal changes in temperature extremes since the beginning of 20th century over Croatia

The time series variability and trends of mean temperature and temperature extremes since the beginning of the 20th century were analyzed for different climatic regions of Croatia. The temperature extremes based on daily series were developed using a unified approach based on internationally agreed climate indices developed by the WMO/WCRP/JCOMM Expert Team on Climate Change Detection and Indices (ETCCDI). Five meteorological stations were chosen as the representatives of different climates in Croatia. The linear trends have been tested for significance by means of nonparametric Mann-Kendall rank statistics. In order to remove short-term fluctuations in data series, they have been smoothed by means of the 11-year binomial moving average filter. For the series, which showed the significant trend identified by Mann-Kendall coefficient t, a progressive analysis of the time series by means of the statistic u(t) was performed in order to determine the beginning of trend. The positive mean annual temperature trends are evident from the beginning of the period, and became more expressed in last 50 and especially in last 25 years. The greatest contribution to positive trends in annual values at the coast gave the trends in warm season, while in the continental part the warming is more pronounced in the cold season. Most from the 10 warmest years belongs to the last 10-year period. The analysis of temperature indices, as the indicator of temperature extremes, and defined as the days with minimum and maximum air temperature exceeding the thresholds defined by percentiles, showed the positive trends in warm and negative trends in cold indices. The trends of temperature indices are greater at the coast than in the continental part. The warming in the capital of Croatia Zagreb, evident in mean temperatures as well as in indices, might be the result of urbanization.

indices of temperature extreme indices; trend analysis; Man Kendall rank statistics; binomial moving average filter; and progressive trend analysis

nije evidentirano