Changes in extreme air temperatures in the mid-sized European city situated on southern base of a mountain (Zagreb, Croatia)

Characteristics and changes in minimum and maximum air temperatures and associated climate indices are analysed for Zagreb city (Croatia). Daily data from the period 1960–2019 at four meteorological stations (urban, suburb, airport and mountain) are used. Generally, changes in extreme temperatures showed to be the least expressed at the mountain site compared to other city stations. An increase (decrease) in the frequency of warm (cold) extremes is obtained using both stationary (applied to non-overlapping 30-year periods) and non-stationary (applied to the whole period of the analysis) generalized extreme value distribution. Likelihood ratio test revealed that inclusion of linearly time-dependent location parameter significantly improves the performance of the model, especially for maximum air temperature. The effect of a period used as a reference for the percentile-based indices estimation is analysed using three different periods: standard climatological period 1961–1990, recent climatological period 1981–2010 and the whole period of the analysis 1960–2019. Regardless of the base period used, generally significant increase (decrease) in warm (cold) indices is detected for both summer and winter season. Still, increase (decrease) in warm (cold) indices is stronger when period 1961–1990 (1981–2010) is used as a reference. The effect of different period used as a reference is more expressed for summer season. Summer trend rates showed to be stronger compared to the winter ones, especially for the right side of the distributions (warm indices). This is due to a more pronounced increasing trend in the latest 30-year period compared to the earlier period. The obtained results imply potential increase (decrease) in risk associated with warm (cold) events which showed to be more expressed for warm summer extremes.