Adv. Geosci., 12, 153–158, 2007 www.adv-geosci.net/12/153/2007/ © Author(s) 2007. This work is licensed under a Creative Commons License. Advances in Geosciences Winter precipitation and cyclones in the Mediterranean region: future climate scenarios in a regional simulation P. Lionello1 and F. Giorgi2 1 University 2 ICTP, of Salento, Lecce, Italy Trieste, Italy Received: 5 March 2007 – Revised: 6 July 2007 – Accepted: 22 October 2007 – Published: 16 November 2007 Abstract. Future climate projections show higher/lower winter (Dec-Jan-Feb) precipitation in the northern/southern Mediterranean region than in present climate conditions. This paper analyzes the results of regional model simulations of the A2 and B2 scenarios, which confirm this opposite precipitation change and link it to the change of cyclone activity. The increase of the winter cyclone activity in future climate scenarios over western Europe is responsible for the larger precipitation at the northern coast of the basin, though the bulk of the change is located outside the Mediterranean region. The reduction of cyclone activity inside the Mediterranean region in future scenarios is responsible for the lower precipitation at the southern and eastern Mediterranean coast. 1 Introduction Previous studies, based both on global and regional simulations show opposite changes of winter (Dec-Jan-Feb) precipitation in different part of the Mediterranean region (Giorgi and Lionello, 2007). In most simulations, precipitation is projected to increase in a relatively narrow band in the northern part and to decrease in the rest, with very large reductions in some areas at the western (Morocco) and eastern (Middle Est) border of the Mediterranean region. Figure 1 shows the projected percentage precipitation change according to the MGME (Multi Global Model Ensemble) for the A2 (top) and B1 (bottom) scenarios. These public data, which are stored at the Program for Climate Model Diagnosis and Intercomparison (PCMDI, http://www-pcmdi.llnl.gov), include the results of a large set of global climate simulations carried out with different models by about 20 research groups for the 20th and 21st century under different greenhouse gas forcing scenarios as a contribution to the fourth Correspondence to: P. Lionello (piero.lionello@unile.it) Assessment Report (AR4) of the Intergovernmental Panel on Climate Change (IPCC). The transition line between positive and negative change crosses the northern part of the Mediterranean region, and its position depends little on the emission scenario, while the intensity of the negative (positive) change in the southern Mediterranean (central and northern Europe) dramatically increases with the emission level. This study analyzes three 30-year regional climate simulations, one for present day conditions (1961–1990) (CTR experiment) and two for future conditions (2071–2100) carried out with the regional climate model RegCM under the IPCC A2 and B2 emission scenarios. The A2 is a high emission scenario, lying towards the high end of the IPCC range, while the B2 is a low emission scenario lying towards the low end of the range. The aim of this study is to discuss the link between changes of winter precipitation and of cyclonic activity in this regional simulation for producing a consistent overall picture of climate change projection for both quantities in the Mediterranean area. 2 The RegCM model: present climate simulation In this study, the RegCM model grid spacing is 50 km and the model domain covers the European region and adjacent oceans. The model is driven at the lateral boundaries by meteorological fields from the Hadley Centre global atmospheric model HadAM3H (1.25×1.875 lat-lon horizontal resolution). Sea surface temperatures (SSTs) are from corresponding simulations with the Hadley Centre global coupled model HadCM3. For more information on the model and for a general discussion of the simulations, the reader is referred to Giorgi et al. (2004a, b), for the discussion of the cyclone climate change signal to Lionello et al. (2007). Figure 2 compares the winter CRU (Climate Research Unit, East Anglia University, (New et al., 2000)) and the CTR Published by Copernicus Publications on behalf of the European Geosciences Union. 154 Lionello P. and Giorgi F.: Precipitation and cyclones P. Lionello and F. Giorgi: Precipitation and cyclones Lionello P. and Giorgi F.: Precipitation and cyclones cyclone ch Unit, TR simmodel to ding the drier ind, espe- ajectory ea Level ed to the and-pass s of the led syne littera- ECMWF he CTR ntitative a, except lation to Mediterngitude, yclones, tracking ng of the of steepAll the mum are ny small pression. he same ted with ults in a pressure per than comparmulation stretchtern Eunumber orrelated ch overrecipitaproduce herwise, ver large n are remap of rack inain. The Fig. 2. Winter precipitation according to the CRU data (top) and to Fig. 2.theWinter precipitation to the CRU data CTR simulation of theaccording RegCM model (bottom). Units:(top) mm.and to Fig. 3. Synop the CTR simulation of the RegCM model (bottom). Units: mm. the CTR simu Fig. 1. Winter (Dec-Jan-Feb) multi model average (MGME) precipFig. 1. Winter (Dec-Jan-Feb) multi model average (MGME) preitation percentage change in the Mediterranean region: A2 scenario cipitation percentage change in the Mediterranean region: A2 sceminus present climate (top) and B1 scenario minus present climate nario minus present climate (top) and B1 scenario minus present (bottom). climate (bottom). effect of the ridges at the northern Bordermodel of the simulation and mountain it shows the capability of the RegCM region the is responsible for the number of cytoMediterranean reproduce correctly observed climatology, including clone centers along North East coast ofthe thedrier basin the maxima along the the coasts and and on the Alps, and in fig.4. However, thethe tracking algorithm has been applied interior areas, although maxima are underestimated, esto the along band-pass filteredGreek fields,and so Turkish that thecoasts. cyclone trajecpecially the western tory differ from those that would be identified in the Thefeatureis cyclone activity is analyzed using a cyclone trajectory original SLP fields. The 15hPa threshold has been adopted identification algorithm and computing the SLP (Sea Level for fig.3standard in orderdeviation. to includeBoth onlymethods intense are cyclone, Pressure) appliedlikely to theto contribute significantly to the precipitation monthly total. SLP fields which have been pre-processed with a band-pass (1–7 day) Lanczos digital filter. The strong features of the band-pass filtered SLP standard deviation (here called synoptic signal) maps are often called storm tracks in the litterature. 3 Climate change signal in the RegCM simulations Figure 3 shows the synoptic signal of the ERA-40 (ECMWF Re-Analysis, Simmons and Gibson, 2000) and of The winter precipitation change in the RegCM simuthe CTR simulation. The climate comparison shows a good quantilations substantially confirms the results of the MGME analAdv. Geosci., 12, 153–158, 2007 ysis1 and it shows contrasting changes in the Northern part, to become wetter, and of the the southern part, projected tative projected agreement over large parts Mediterranean area, to become drier. Changes increase with the emission level except(they overareTurkey. There is a tendency of the CTR simularger in the A2 than in the B2 scenario, where they lation are to overestimate synoptic outside significant onlythe in small areas)signal and aremostly particularly sig-the Mediterranean region over eastern Europeover (East 20 E nificant in the A2 scenario being positive theof Alps andlonnorthern Adriatic Coastand on one negative At the gitude,theexcept over Greece) the hand, Blackand Sea. coast of Turkeyproduces and of Middle on the TheMediterranean identification algorithm a listeast, of cyclones, other hand. Figs.5 shows the statistically significant changes with the SLP minima positions and values. The tracking at a 90% confidence level according to a Mann-Whithey test. method Also (Lionello et al.,activity 2002)the is Mediterranean based on partitioning of the for cyclone is a transition slp fields in depressions by the identification of sets of steepregion with an intensification on its north-western area and a est descent paths to the slpsynoptic minimum. reduction overleading the eastern area,same both as signalAll and the of by cyclone centers are concerned (figs.6minimum and 7). In are pointsnumber crossed a path leading to the same Fig.7toonly deeper than 15hPa are included. assigned the minima same depression. Subsequently, anyNote small that, however, the area where the largest intensification of cyclone activ it is associate track over th These cha herent. It is sociated with associated to surrounding all intense ra of a cyclone, air to the loc 2001). The strong depression is merged into the closest large size depression. tation is confi 1 of total seas The B2 scenario by usedjoining in this study is not available in same the A trajectory, obtained the location of the MGME note that maps, B1 and B2 similar, with low pressuredataset. centerHowever, in successive is are associated with tic signal an 550 and 600ppm CO concentration in 2100, respectively. the CTR sim the evolution of each depression. The procedure results in a trajectory, an initial and final point, a sequence of pressure minima, and of areas covered by the cyclone. Figure 4 shows the density of cyclone centers deeper than 15 hPa in the band-pass filtered ERA-40 and the CTR simulation fields. The comparison between the two figures shows that the CTR simulation overestimates the number of cyclones over an area stretching from central Europe to Turkey www.adv-geosci.net/12/153/2007/ yclones P. Lionello and F. Giorgi: Precipitation and cyclones 3 U data (top) and to m). Units: mm. Fig. 3. Fig. Synoptic signal according data(top) (top) 3. Synoptic signal accordingtotothe the ERA-40 ERA-40 data andand to to thesimulation CTR simulation of the RegCMmodel model(bottom). (bottom). Units: the CTR of the RegCM Units:hPahPa e Northern part, n part, projected cyclone activity takes place is outside the Mediterranean and e emission level it is associate with an intensification of the mid-latitude storm ario, where they acrosstrack south-eastern Europe. It will be shown in the next over the north-east Atlantic. particularly sigSect. 3 that the number of cyclones at the southern border of ver the Alps and These changes of precipitation and cyclone activity are coin thisherent. area isIt is correlated with precipitation over isGreece negative At the well known that the the passage of a cyclone asdle east, on the and Turkey. Suchrain. overestimation, combined withis often the unsociated with In the Mediterranean region this nificant changes associatedoftoprecipitation, the advection of humid air against the Mediterslopes derestimation suggests that eastern nn-Whithey test. basin. In fact,rain it hasinbeen almost raneansurrounding cyclones the produce less the shown modelthatsimulation an is a transition intense rain events arethe in coincidence withcyclone the presence than inallreality. Otherwise, frequency of centers estern area and a of a cyclone, positioned so that its circulation advects humid is roughly correct over large parts of the domain. optic signal and air to the location where precipitation occurs (Jansá et al. , figs.6 and 7). In 2001). characterizing the Mediterranean region are reFeatures included. Note link between position of cyclones producedThebystrong the model results in Figs. 3 and and 4. precipiThe map ntensification of tation is confirmed computing the Spearman rank-correlation of the ofwinter synoptic signal (Fig. 3) shows the storm track total seasonal precipitation in selected areas with synopt available in the intensification on the lee of the Alps and Atlas Mountain. tic signal and frequency of cyclones deeper than 15hPa for are similar, with The effect of the mountain at the northern pectively. the CTR simulation. Theridges geographical distributionBorder of the of the Mediterranean region is responsible for the number of cyclone centers along the North and East coast of the basin in Fig. 4. However, the tracking algorithm has been applied to the band-pass filtered fields, so that the cyclone trajectory features differ from those that would be identified in the original SLP fields. The 15hPa threshold has been adopted for Fig. 3 in order to include only intense cyclone, likely to contribute significantly to the precipitation monthly total. www.adv-geosci.net/12/153/2007/ 4 155 Lion Fig. 5. Winte nus A2 scena 4. Cyclone frequency 15hPa) according thethe Fig. 4.Fig.Cyclone (deeper than (deeper 15 hPa)than frequency accordingto to denote drier c 8 number of ERA-40 data (top) to theCTR CTRsimulation. simulation. Units: 10 10 ERA-40 data (top) andand to the Units: number changes are s month . Data have been smoothed on squared areas minima km of minima km−2 month−1 . Data have been smoothed on squared cording to the with size 3 and 11 grid points for ERA-40 and CTR, respectively. areas with size 3 and 11 grid points for ERA-40 and CTR, respectively. cyclone activity responsible for winter precipitation on the continental part of Italy (including the southern side of the 3 Climate change signal in the RegCM simulations Alps) and the Croatian coast of the Adriatic Sea is shown in fig.8. The seasonal value of precipitation in these two areas (whereprecipitation it is projected climate to increase) is correlated with the synThe winter change in the RegCM simusignal and the numberthe of central lationsoptic substantially confirms resultspressure of the minima MGMEover analthe western Mediterranean and western Europe (fig.8). The ysis1 and it shows contrasting changes in the Northern part, Mediterranean coast of Turkey and the eastern Coast of the projected to become wetter, and the southern ispart, projected Mediterranean (where the winter precipitation projected to to become drier. Changes increase the emission level diminish) show a different pattern: with precipitation over those (they two are regions larger isincorrelated the A2 with thansynoptic in the signal B2 scenario, in a regionwhere including the whole Mediterranean and north and with they are significant only in small areas) andAfrica are particularly the frequency of cyclone centers the northern coast significant in the A2 scenario, beingalong positive over the Alps of the Mediterranean (fig.9). Therefore, the comparison beand the northern Adriatic coast on one hand, and negative tween fig.8 and fig.9 shows the different positions of cyclone at the Mediterranean coast of Turkey and of Middle east, on the other hand. Figure 5 shows the statistically significant changes at the 90% confidence level according to the MannWhithey test. 1 The B2 scenario used in this study is not available in the MGME dataset. However, note that B1 and B2 are similar, with 550 and 600 ppm CO2 concentration in 2100, respectively. Adv. Geosci., 12, 153–158, 2007 centers and o ern and sout Using the preting the to the concl number of c mainly in it change of p change in th instead caus storm track o 4 Conclus The analysis B2 scenario g to the mber of ed areas tively. on the of the own in o areas he syna over ). The of the cted to r those ion ind with coast on beyclone 156 Lionello P. and Giorgi F.: Precipitation and cyclones Fig. 5. Winter precipitation climate change signal. Top: CTR minus A2 scenario. Bottom: CTR minuschange B2 scenario. values Fig. 5. Winter precipitation Climate signal. Positive Top: CTR midenote drier conditions in climate scenarios. In the colored areas nus A2 scenario. Bottom: CTR minus B2 scenario. Positive values changesdrier are statistically significant at the 90% In confidence levelareas acdenote conditions in climate scenarios. the colored cording to Mann-Whitney test. Units: mm. confidence level acchanges arethe statistically significant at the 90% cording to the Mann-Whitney test. Units: mm. Also for cyclone activity the Mediterranean is a transition region with an intensification on its north-western area and a centers and of cyclone activity that produce rain in the northreduction over the eastern area, both as synoptic signal and ern and south-eastern coasts of the Mediterranean. number of cyclone centers are concerned (Figs. 6 and 7). In Using the spatial distribution of the correlation for interFig. 7 only minima deeper than 15 hPa are included. Note preting the climate change signal shown in figs.5-7 brings that, however, the area where the largest intensification of to the conclusion that the reduction of cyclone activity and cyclone activity takes place is outside the Mediterranean and number of cyclones inside the Mediterranean region (and it is associate with an intensification of the mid-latitude storm mainly in its eastern part) is responsible for the negative track over the north-east Atlantic. change of precipitation in its eastern part. The positive Theseinchanges of precipitation andMediterranean cyclone activity are cochange the northern part of the region is herent. It is well known that the passage of a cyclone is asinstead caused by the increased strength of the mid-latitude sociated with rain. In mostly the Mediterranean regionitself. this is often storm track occurring outside the region associated to the advection of humid air against the slopes surrounding the basin. In fact, it has been shown that almost 4all intense Conclusions rain events are in coincidence with the presence of a cyclone, positioned so that its circulation advects humid air The regional climate simulations ofetthe and to theanalysis locationofwhere precipitation occurs (Jansá al.,A2 2001). B2 scenario confirms the global ensemble results and it Adv. Geosci., 12, 153–158, 2007 andF.: F. Precipitation Giorgi: Precipitation and cyclones LionelloP.P.Lionello and Giorgi and cyclones Fig. 6. Synoptic signal. Top: A2 minus CTR scenario. Bottom: B2 minus CTR scenario. Positive values denote larger CTR synoptic Fig. 6. Synoptic Climate change signal. Top: A2 minus scesignal in climate scenarios. In the colored areas changes stanario. Bottom: B2 minus CTR scenario. Positive valuesare denote tistically significant at in theclimate 90% confidence according the larger synoptic signal scenarios. level In the coloredtoareas Mann-Whitney test. Units: HPa. at the 90% confidence level acchanges are statistically significant cording to the Mann-Whitney test. Units: HPa. The strong link between position of cyclones and precipishows wetter winter conditions at the northern boundary of tation is confirmed computing the Spearman rank-correlation the Mediterranean region and drier along the southern and of total seasonal precipitation in selected areas with synopeastern coast of the basin. A purpose of this paper is to extic signal and frequency of cyclones deeper than 15 hPa for plain such changes in the precipitation field (fig.5) on the bathe CTR simulation. The geographical distribution of the sis of changes in synoptic signal (fig.s6) and cyclone number cyclone activity responsible for winter precipitation on the (fig.7) because of the respective correlation fields shown in continental part of Italy (including the southern side of the figs.8 and 9. Alps) and the Croatian coast of the Adriatic Sea is shown in The high resolution of the RegCM simulation allows the Fig. 8. The seasonal value of precipitation in these two areas identification of cyclones and the evaluation of the climate (where it is projected to increase) is correlated with the synchange signal. Cyclone activity is projected to become optic signal and the number of central pressure minima over weaker over the eastern Mediterranean and stronger in the the western Mediterranean and western Europe (Fig. 8). The north-western areas. In fact, the north western Mediterranean Mediterranean coast of Turkey and the eastern coast of the is partially affected by the intensification of the mid-latitude Mediterranean (where the winter precipitation is projected to storm-track which occurs over central and western Europe in diminish) show a different pattern: precipitation over those the scenario simulations. www.adv-geosci.net/12/153/2007/ Fig. 7. minima tom: B optic si statistic Mann-W The tion wi clone m the sou activity tation i over w tivity a are con This region and it i the mi ios pro TR scedenote d areas evel ac- ary of rn and to exhe baumber own in ws the limate ecome in the ranean atitude ope in P. Lionello and F. Giorgi: Precipitation and cyclones 5 Fig. 7. Cyclone frequency change signal (considering only cyFig. Cyclonedeeper frequency change only cyclone clone7.minima than 15 hPa).signal Top: (considering A2 minus CTR scenario. minima than CTR 15hPa). Top: A2 minusvalues CTR scenario. BotBottom: deeper B2 minus scenario. Positive denote higher tom: B2 minus CTR scenario. Positive values denote larger frequency in climate scenarios. In the colored areas changes synare optic signal significant in climate at scenarios. In the colored changes are statistically the 90% confidence levelareas according to the 8 number −2 month statistically significant at the1090% confidence levelkm according to−1 the Mann-Whitney test. Units: of minima . Mann-Whitney test. Units: 10 number of minima km month . two regions correlated with synoptic signalofinthe a region inThe spatialis distribution of the correlation precipitacluding the whole Mediterranean and north Africa and with tion with the synoptic signal and with the frequency of cythe frequency cyclone coast of clone minima of shows that,centers on onealong hand,the thenorthern precipitation at the Mediterranean (Fig. 9). Therefore, the comparison bethe south-eastern coast of the basin is linked to the cyclonic 8 and showsand, the on different positions of cyclone tween Figs. activity inside the 9basin, another hand, the precipicenters and of cyclone activity areas that produce rainactivity in the tation in the northern part is linked to the cyclonic northern and south-eastern coasts of the Mediterranean. over western Europe. Therefore, the changes of cyclone acUsing spatial distribution for intertivity andthe of precipitation shownof bythe thecorrelation RegCM simulations preting the climate change signal shown in Figs. 5–7 brings are consistent. to This the conclusion that the reduction of cyclone activity and analysis shows that the response of the Mediterranean number of cyclones inside the Mediterranean region (and region to climate change is not necessarily homogeneous mainly its easternbypart) is responsible negative and it is in conditioned changes of intensityfor andtheposition of change of precipitation in its eastern part. The the mid-latitude storm track. Since future climatepositive scenarios produce an intensification and a northward shift (away www.adv-geosci.net/12/153/2007/ 6 157 Fig. 8. Spearman Rank correlation of the seasonal precipitation Fig. 8. Spearman Rank correlation of the seasonal precipitation at the Croatian coast of the Adriatic Sea and over the continental at the eastern Mediterranean coast and over the continental part of part of Italy with the synoptic signal (top) and with the frequency of Italy coast with the synoptic signal (top) and with the frequency of cyclone centers deeper than 15 hPa (bottom). In the colored areas cyclone centers deeper than 15hPa (bottom). In the colored areas correlation is statistically significant at the 90% confidence level. correlation is statistically significant at the 90% confidence level. change in the northern part of the Mediterranean region is instead caused by the increased strength of the mid-latitude from the Mediterraean) of the mid-latitude storm track this storm track occurring mostly outside the region itself. has contrasting effect in different areas within the Mediterranean. The increased intensity of the storm track prevails in the Mediterranean areas producing an increased winter 4 north Conclusions precipitation. The effect of its shift prevails in the southeastern areas, of forregional which the Mediterranean is A2 a main The analysis climate simulationsitself of the and source of moisture, where the reduced activity reB2 scenario confirms the global MGMEcyclonic results and it shows sults in winter a diminished precipitation at the coast in future wetter conditions at the northern boundary of clithe mate scenarios. region and drier along the southern and eastMediterranean ern coast of the basin. A purpose of this paper is to explain such changes in the precipitation field (Fig. 5) on the basis of changes in synoptic signal (Fig. 6) and cyclone freAcknowledgements. Thethe authors thank Dr X.Bi for fields his help with quency (Fig. 7) using respective correlation shown the and 9. U.Boldrin for his contribution to the programs used in graphics Figs. 8 and in this study. Adv. Geosci., 12, 153–158, 2007 Fig. 9. at the T signal (t 15hPa ( nificant Refere Giorgi, trend I: Pre 2004 Giorgi, trend II: Fu 858, Giorgi, ranea Jansà M Carre Part 2 Lionello gion: pitation part of ency of d areas evel. k this editerails in winter southmain ty rere cli- p with ms used 158 Lionello P. and Giorgi F.: Precipitation and cyclones P. Lionello and F. Giorgi: Precipitation and cyclones This analysis shows that the response of the Mediterranean region to climate change is not necessarily homogeneous and it is conditioned by changes of intensity and position of the mid-latitude storm track. Since future climate scenarios produce an intensification and a northward shift (away from the Mediterraean) of the mid-latitude storm track this has contrasting effect in different areas within the Mediterranean. The increased intensity of the storm track prevails in the north Mediterranean areas producing an increased winter precipitation. The effect of its shift prevails in the southeastern areas, for which the Mediterranean itself is a main source of moisture, and where the reduced cyclonic activity results in a diminished precipitation at the coast. Acknowledgements. The authors thank X. Bi for his help with the graphics and U. Boldrin for his contribution to the programs used in this study. Edited by: P. Alpert, H. Saaroni, and E. Heifetz Reviewed by: two anonymous referees References Fig. 9. Spearman Rank correlation of the seasonal precipitation Fig. 9. Turkish Spearman correlation of thecoast seasonal at the and Rank Eastern Mediterranean with precipitation the synoptic atsignal the Turkish and Eastern Mediterranean coastcenters with the synoptic (top) and with the frequency of cyclone deeper than signal (top) and with the frequency of cyclone centers deeper 15 hPa (bottom). In the colored areas correlation is statistically than sig15hPa (bottom). In the colored level. areas correlation is statistically significant at the 90% confidence nificant at the 90% confidence level. The high resolution of the RegCM simulation allows the identification of cyclones and the evaluation of the climate change signal. Cyclone activity is projected to become References weaker over the eastern Mediterranean and stronger in the north-western In fact, northinterannual western Mediterranean Giorgi, F., Bi, X.areas. and Pal, J. S.:the Mean, variability and is trends partially by the intensification of the in a affected regional climate change experiment overmid-latitude Europe. Part I: Present day climate (1961-1990). Dyn., 22,Europe 733–756 storm-track which occurs over centralClim. and western in 2004a. the scenario simulations. Giorgi, Bi, X.distribution and Pal, J.S.: Mean, interannual and The F., spatial of the correlation of variability the precipitatrends a regional climate change over Europe. tion within the synoptic signal and experiment with the frequency of Part cyII: Future climate scenarios (2071-2100). Clim. Dyn., 23, 839clone minima shows that, on one hand, the precipitation at 858, 2004b. the south-eastern coast of the basin is linked to the cyclonic Giorgi, F., Lionello P.: Climate Change projections for the Mediteractivity inside the basin, and, on another hand, the precipiranean region, Global Planet. Change, submitted, 2007 tation in northern is linked the cyclonic activity Jansà M.A.the , Genovès A., part Picornell M.A. ,to Campins J., Riosalido R., over western Europe. Therefore, the changes of cyclone acCarretero O.: Western Mediterranean cyclones and heavy rain. tivity of precipitation by the RegCM simulations Part and 2: Statistical approach,shown Meteorol. Appl., 8, 43–56, 2001 are consistent. Lionello P., Dalan F., Elvini E. Cyclones in the Mediterranean Re- Giorgi, F., Bi, X., and Pal, J. S.: Mean, interannual variability and trends in a regional climate change experiment over Europe. Part I: Present day climate (1961–1990), Clim. Dyn., 22, 733–756 2004a. Giorgi, F., Bi, X., and Pal, J. S.: Mean, interannual variability and trends in a regional climate change experiment over Europe. Part II: Future climate scenarios (2071–2100), Clim. Dyn., 23, 839– 858, 2004b. Giorgi, F. and Lionello P.: Climate Change projections for the Mediterranean region, Global Planet. Change, accepted, 2007. Jansà, M. A., Genovès, A., Picornell, M. A., Campins, J., Riosalido, R., and Carretero, O.: Western Mediterranean cyclones and heavy rain. Part 2: Statistical approach, Meteorol. Appl., 8, 43–56, 2001. Lionello, P., Dalan, F., and Elvini, E.: Cyclones in the Mediterranean Region: the present and the doubled CO2 climate scenarios, Clim. Res., 22, 147–159, 2002. Lionello, P., Bhend, J., Buzzi, A., Della-Marta, P. M., Krichak, S., Jansá, A., Maheras, P., Sanna, A., Trigo, I. F., and Trigo, R.: Cyclones in the Mediterranean region: climatology and effects on the environment, in Mediterranean Climate Variability, edited by: Lionello, P., Malanotte-Rizzoli, P., and Boscolo, R. Elsevier, Amsterdam, The Netherlands, 324–372, 2006. Lionello, P., Boldrin, U., and Giorgi, F.: Future changes in cyclone climatology over Europe as inferred from a regional climate simulation, Clim. Dyn., doi:10.1007/s00382-007-0315-0, 2007. New, M. G., Hulme, M., and Jones, P. D.: Representing twentieth century space time climate fields. Part II: Development of a 1901–1996 mean monthly terrestrial climatology, J. Climate, 13, 2217–2238, 2000. Simmons, A. J. and Gibson, J. K.: The ERA-40 Project Plan, ERA40 Project Report Series n. 1, 2000. gion: the present and the doubled CO2 climate scenarios. Clim. Adv. Geosci., 12, 153–158, 2007 www.adv-geosci.net/12/153/2007/