LETTERS PUBLISHED ONLINE: 12 DECEMBER 2010 | DOI: 10.1038/NGEO1032 Observational evidence for soil-moisture impact on hot extremes in southeastern Europe Martin Hirschi 1,2 * , Sonia I. Seneviratne 1 * , Vesselin Alexandrov 3 , Fredrik Boberg 4 , Constanta Boroneant 5 , Ole B. Christensen 4 , Herbert Formayer 6 , Boris Orlowsky 1 and Petr Stepanek 7 Climate change is expected to affect not only the means of climatic variables, but also their variabilities 1,2 and extremes such as heat waves 2–6 . In particular, modelling studies have postulated a possible impact of soil-moisture deficit and drought on hot extremes 7–11 . Such effects could be responsible for impending changes in the occurrence of heat waves in Europe 7 . Here we analyse observational indices based on measurements at 275 meteorological stations in central and southeastern Europe, and on publicly available gridded observations 12 . We find a relationship between soil-moisture deficit, as expressed by the standardized precipitation index 13 , and summer hot extremes in southeastern Europe. This relationship is stronger for the high end of the distribution of temperature extremes. We compare our results with simulations of current climate models and find that the models correctly represent the soil-moisture impacts on temperature extremes in southeastern Europe, but overestimate them in central Europe. Given the memory associated with soil moisture storage, our findings may help with climate-change- adaptation measures, such as early-warning and prediction tools for extreme heat waves. A preferential warming of the hot tail of temperature distri- butions as a consequence of climate change has been reported in observational studies for the European continent, where changes in the upper and lower tails of daily minimum, maximum and mean temperature distributions have been analysed 14–16 . Strong trends in heat-wave intensity, length and number have in particular been identified in the eastern Mediterranean region 17 . In the context of modelling studies, soil moisture has been shown to possibly play an important role for the occurrence of hot extremes in Europe 7,10,11 and impending changes thereof with climate change 7,8,18 . Using quantile regression 19,20 (see Methods), we investigate whether such a relation between dry conditions and hot extremes can indeed be established from observations in Europe based on a newly available observational database. Quantile regression was developed as an extension to the ordinary least squares regression to estimate the response not only in the mean of a variable, but in all parts of its data distribution. It has been widely used in econometrics 19,20 , and in some ecological studies 21 . Recently, the method has been applied for the identification of trends in the Baltic sea level 22 . For the analysis, we use the standardized precipitation index (SPI; ref. 13, see Methods), a commonly used drought index 23,24 , and two temperature indices from the CECILIA climate and 1 Institute for Atmospheric and Climate Science, ETH Zurich, Universitaetsstrasse 16, 8092 Zurich, Switzerland, 2 Federal Office for Meteorology and Climatology, MeteoSwiss, Kraehbuehlstrasse 58, 8044 Zurich, Switzerland, 3 National Institute of Meteorology and Hydrology, 66 Tsarigradsko Chaussee, 1784 Sofia, Bulgaria, 4 Danish Meteorological Institute, Lyngbyvej 100, 2100 Copenhagen, Denmark, 5 National Meteorological Administration, Sos. Bucuresti-Ploiesti 97, 013686 Bucharest, Romania, 6 University of Natural Resources and Life Sciences, Gregor Mendel Strasse 33, 1180 Vienna, Austria, 7 Czech Hydrometeorological Institute, Kroftova 43, 616 67 Brno, Czech Republic. *e-mail:m.hirschi@alumni.ethz.ch; sonia.seneviratne@env.ethz.ch. BOKU CHMI NIMH NMA Longitude (° E) Latitude (° N) 35 40 45 50 55 60 10 15 20 25 30 Figure 1 | Station observations and analysis domains. Location of high-density station observations from the CECILIA climate and extreme database (responsible institutions: University of Natural Resources and Life Sciences (BOKU), Austria; Czech Hydrometeorological Institute (CHMI), Czech Republic; National Institute of Meteorology and Hydrology (NIMH), Bulgaria; National Meteorological Administration (NMA), Romania), as well as analysed central and southeast European domains (boxes). extreme database (see Methods): the percentage of hot days (%HD) and the maximum heat-wave duration (HWD max ). This database was developed in the framework of the EU-FP6 project CECILIA and provides indices computed from two observational sources: (1) the gridded E-Obs dataset 12 and (2) high-density station data from the participating institutions (Fig. 1). The latter set of indices represents an invaluable new observational reference for the investigation of extreme events in central and NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience 1 © 2010 Macmillan Publishers Limited. All rights reserved.