Premium
Trends in summer extreme temperatures over the Iberian Peninsula using nonurban station data
Author(s) -
Acero F. J.,
García José Agustín,
Gallego María Cruz,
Parey Sylvie,
DacunhaCastelle Didier
Publication year - 2014
Publication title -
journal of geophysical research: atmospheres
Language(s) - English
Resource type - Journals
eISSN - 2169-8996
pISSN - 2169-897X
DOI - 10.1002/2013jd020590
Subject(s) - percentile , poisson distribution , heat wave , peninsula , extreme value theory , standard deviation , poisson regression , return period , environmental science , intensity (physics) , statistics , extreme heat , mathematics , climatology , demography , geography , climate change , geology , physics , population , flood myth , oceanography , archaeology , quantum mechanics , sociology
Heat wave event trends over the Iberian Peninsula (IP) are studied using extreme value theory, specifically the peaks‐over‐threshold (POT) approach. Summer (June–August) daily temperature records from 20 observatories regularly distributed over Iberia in places far from urban effects were available for the common period 1961–2010. Heat waves are defined as days occurring above the 95th percentile of the temperature distribution, considering both maximum ( T max ) and minimum ( T min ) temperatures. These events were identified using a “run declustering” scheme to select independent extreme events exceeding the threshold. Also, the dates of occurrence of the independent events were fitted to a Poisson process. Trends in the following parameters were studied: the scale parameter of the POT approach, the Poisson intensity, mean, return level period, and low (25th percentile) and high (75th percentile) values. The optimal trends in the Poisson intensity considering both T max and T min show a major increase in the occurrence of heat waves. Also, the rise in the return level trend was less than that in the mean of T min and T max , and the analysis of the values of T min and T max showed a greater increasing trend in the low values (25th percentile) than in the high values (75th percentile), especially for T max , leading to a decrease in the variance. Over the IP, temperature extremes are increasing but not as much as the mean because the variance is tending to decrease. This highlights the important role of variance in the evolution of extremes.