high resolution climate scenarios on mediterranean test ... detected for statistical downscaling...

Research PapersIssue RP0233October 2014

Impacts on Soil andCoasts Division

This report represents theDeliverable P92 developed

within the framework of WorkPackage 2.17 Action A of the

GEMINA Project, funded bythe Italian Ministry of

Education, University andResearch and the ItalianMinistry of Environment,

Land and Sea.

High resolution climate scenarios onMediterranean test case areas forthe hydro-climate integrated system

By Paola MercoglianoMeteo System &

Instrumentation Laboratory,CIRA - Capua Impacts on Soiland Coasts Division, CMCC -

[email protected]

Myriam MontesarchioMeteo System &

Instrumentation Laboratory,CIRA - Capua Impacts on Soiland Coasts Division, CMCC -

[email protected]

Guido RiannaImpacts on Soil and Coasts

Division, CMCC - [email protected]

Pasquale SchianoOn-Board Systems and ATM

Division, Italian AerospaceResearch Center (CIRA),

Capua (CE), Italy & Impacts onSoil and Coasts Division,

Centro Euro-Mediterraneo suiCambiamenti Climatici

(CMCC-ISC), Capua (CE), [email protected]

Renata VezzoliImpacts on Soil and Coasts

Division, CMCC - [email protected]

and Alessandra LuciaZollo

Meteo System &Instrumentation Laboratory,

CIRA - Capua Impacts on Soiland Coasts Division, CMCC -

[email protected]

SUMMARY This document is aimed to show the main results of climateprojections, under two RCPs, at 2100 obtained in WP A.2.6 “High-resolutionclimate scenarios” on the geo-hydrological hotspots identified within WPA.2.17 "Analysis of geo-hydrological risk related to climate change" ofGEMINA project. The main goal of WP A.2.17 is the analysis of the effect ofclimate changes on occurrence and magnitude of landslides, floods and lowflows hazards on some specific contexts of the Mediterranean area. Toreach this objective, climate data at the same horizontal resolution (<10 km)of impacts model are required. Within GEMINA project, the generation ofhigh-resolution climate scenarios is one of the aims of WP A.2.6. Thus, inthis research paper, the projected climate changes obtained in WP A.2.6coupling the GCM CMCC-CM with the RCM COSMO-CLM under RCP4.5and RCP8.5 scenarios are discussed on the following test case areasidentified in WP A.2.17: Po river basin, Cervinara and Orvieto sites. Somedetails on validation of the climate model on the same areas are also given.

Keywords: Climate scenarios, RCM, GCM, high resolution, landslides, flood,droughts, Po river basin

02

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

CMCC Research Papers

1. INTRODUCTION AND MOTIVATION

In the last years, the evaluation of the potentialeffect of climate changes on geo-hydrologicalhazards became exceptionally debated.

The availability of long time series of weathervariables and derived parameters regulatingthe hazards has sometimes allowed estimatingon going variations and their significance [33];however, for the future period such assessmenttakes even greater relevance conveying landplanning options or, in general, adaptation poli-cies.

To this aim, the general framework reckons onweather variables provided by climate modelsadopted as input for the specific impact tools; insuch contexts, as is well known, GCMs (GlobalClimate Models) allowing to evaluate at globalscale the response to emission scenarios of theclimate system, are characterized by horizontaland vertical resolutions compromising a properassessment at scale needed for impact studiesrequiring the adoption of downscaling (statisti-cal or dynamical) approaches.

Statistical approaches, preferred for their lim-ited required computational and time effort,are however based on crucial (strong) as-sumption according which feedback mecha-nisms between local and global weather pat-tern, today recognizable, could be suitablealso for the future under the effect of climatechange; conversely, dynamical downscalingapproaches demand huge computational ef-forts but, as physically based numerical mod-els directly nested for specific region on GCM,do not need to state, under steady-state con-ditions, relationships between global and localweather patterns.

For example, for Greece characterized by cli-mate features and orographic complexity com-parable to Italian one, [44] show how an im-proved representation of morphology induce a

substantial enhancement in reproducing pre-cipitation pattern and, at the same time, entailsa strong variation in climate signal detected,on the area, by driving GCM. Gao et al. [16]for South-Eastern Asia, assume that the im-provements could be not only due to a betterrepresentation of orography but also to moreconsistent physical parametrizations employedin RCMs.

However, as showed in [38, 39, 45, 51], a fur-ther mismatch but achieved horizontal reso-lutions in RCMs and the usual scale investi-gated in impact studies subsists; moreover, thevalidation phase on control period reveal howRCM resolutions and resulting necessary phys-ical parametrizations usually induce errors inproper assessment, for example, of cumulativevalues of precipitation or wet days preventingthe direct use of weather variables provided byRCMs as input for impact tools. For overcomingsuch issue, usually RCM outputs are subjectedto statistical approaches, known as bias correc-tion methods able to correct, at least, the errorsassociated to mean value (i.e. delta changeapproach) if not, potentially, those associatedto all main statistical moments (i.e. quantilemapping approaches).

It worth noting that the key prerequisite, re-quired to bias correction approaches, regardtheir attempt to correct only that are recog-nized as ”systematic errors” not affecting theclimate signals, physically estimated, by cli-mate models; for these reasons, beyond thepossible ”quantitative” misrepresentation of cli-mate models, trends and statistical significancerevealed by RCMs should be fully preservedalso after bias-correction; however, a large de-bate about the adoption of bias correction ap-proaches, their constraints and advantages,arise in last years [12, 26] but although their“statistical” features produce issues similar tothose detected for statistical downscaling ap-

03

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

proaches, their adoption in impact analysis, onthe short term, is almost unanimously acknowl-edged.

Such findings has led ISC research division,within GEMINA project, to develop the simula-tion framework widely explained in [49] consist-ing of GCM (CMCC-CM [35]), RCM (COSMO-CLM [37]), different bias correction approachesdepending on investigated impact and its extentand finally, specific impact tools.

So, the aim of this research paper is to displaythe evolutions and trends of the main weathervariables provided by numerical climate mod-els (GCM/RCM) under two RCPs until for thethree selected hot-spots deputed, within WPA.2.17 of GEMINA project, as pilot case studiesfor investigating the potential effect of climatechange on geo-hydrological hazards: namely, i)Po river basin for which the assessment of vari-ations, potentially induced by climate changein discharges and then in occurrence of “ex-treme tails” of it, droughts and floods events,has been carrying out; ii) Orvieto and iii) Cerv-inara where, instead, at the slope scale, theassessment of variations respectively in mag-nitude of slope reaccelerations and occurrenceof flowslide triggering have been performing. Insuch context, the target variables assumed reg-ulating geo-hydrological hazards (namely, dis-charge for Po river basin and soil water pres-sures for landslide case studies), could be as-sumed main function of the soil surface balancefor which precipitation and temperature play asignificant role, the first controlling the ingoingfluxes, the latter as proxy variable of evapotran-spiration (outgoing fluxes).

In the following, first a brief review on climatemodelling chain (Section 2) and case studies(Section 3) is carried out; after, for the threecases (Sections 4 and 5), the most mean-ingful results concerning the trends of precip-itation and temperature estimated by climate

models for XXI century (under RCP4.5 andRCP8.5 concentration scenarios) are reported;because of the great difference in the extent,the physical dynamics and control variables ofthe investigated geo-hydrological hazards, dif-ferent features of weather patterns are takeninto account and different results displaying arepreferred for Po river basin (flood/low flow haz-ard) and Cervinara and Orvieto (landslide haz-ard); however, for each case study, the mainfindings about validation phases are first brieflyrecalled and after, according to the aim of thisresearch paper, the displaying of outputs forXXI century is addressed. Those presentedhere, are only some of the analysis performedon the climate simulations results, e.g. ETCCDIindices [29] and Taylor diagrams [53] are pro-duced for validation purposes on Po river basin;the probability distribution of annual maximumprecipitation at 1 and 120 days are studied tosupport the landslides hazards assessment un-der climate change [32, 51, 31] for Cervinaratest cases.

The high-resolution climate simulations anal-ysed here are generated within the frameworkof WP A.2.6 ”High-resolution climate scenar-ios” of GEMINA project aimed to study the evo-lution of climate and the variability of extremephenomena on the entire Italian territory, Fig.1.Among the climate phenomena of interest wecite, for precipitation: changes in annual max-imum precipitation at different temporal scalesor duration of dry period; for temperature: fre-quency and duration of heat waves.

2. CLIMATE MODEL

2.1 REPRESENTATIVECONCENTRATION PATHWAYS: RCP4.5AND RCP8.5

Climate projections are the results of numericalsimulations performed by climate models driven

04

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


Figure 1:COSMO-CLM domain and orography for the

configuration with horizontal resolution of 8 km

by scenarios hypothesizing, on the basis ofdifferent socio-economic assumptions, variousconcentration pathways for greenhouses gasesand pollutants. The most commonly used sce-narios are RCP4.5 and RCP8.5. The RCP4.5is a stabilization scenario where the radiativeforcing will assess at about 4.5 W/m2 before2100, while RCP8.5 is a more extreme sce-nario where radiative forcing will increase up to8.5 W/m2 in 2100 compared to pre-industrialera [27]. Such scenarios are used to driveglobal climate models, that are characterisedby a coarse horizontal resolution that makestheir outputs not suitable for application in geo-hydrological studies. Thus GCMs outputs aredynamically downscaled by regional climatemodels obtaining climate variables at higherhorizontal resolution comparable with the onerequested by geo-hydrological impact models.

2.2 GLOBAL CLIMATE MODEL:CMCC-CM

The global climate model CMCC-CM is acoupled atmosphere-ocean general circulationmodel, which has been implemented and devel-oped in the framework of the European projectCIRCE [18]: the atmospheric model compo-nent is ECHAM5 [34] with a T159 horizon-

tal resolution (0.75◦), while the global oceancomponent is OPA 8.2, in its ORCA2 globalconfiguration, at a horizontal resolution of 2◦.ORCA2 also includes the Louvain-La-Neuve(LIM) model for the dynamics and thermody-namics of sea-ice. A performance assessmentof CMCC-CM in simulating the observed SeaSurface Temperature and precipitation is re-ported in [35], while a comparison with otherstate-of-art GCMs available in the fifth phaseof the Coupled Model Intercomparison Project(CMIP5) is presented in different other workse.g. [13].

2.3 REGIONAL CLIMATE MODEL:COSMO-CLM

A detailed description of the regional climatemodel COSMO-CLM and is reported in [49,28]. COSMO-CLM is the climate version ofCOSMO-LM weather model [37], developed bythe CLM Community. It is a high resolution(less than 50 km) non-hydrostatic RCM ableto explicitly capture small scale severe weatherevents and an improved representation of sub-grid scale physical processes, such as clouds,aerosols, orography, land and vegetation prop-erties. It has been already used in the frame-work of several European projects, such asPRUDENCE [8] and CORDEX [17], highlight-ing its good capability in reproducing the meanclimate features of the areas under study, witha similar range of accuracy with respect to theother RCMs adopted. More details concern-ing the description of the main characteristicsof COSMO-CLM and the configuration usedto perform the simulations presented in thiswork are reported in [50, 29]. Climate simula-tions have been carried out at 0.0715◦ (about 8km) of horizontal resolution using as initial andboundary conditions: (a) ERA Interim Reanal-ysis [11] over the 1981-2010 period, character-ized by a horizontal resolution of about 0.70◦,

05

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

about 79 km, to assess the performances ofthe regional climate model when driven by “per-fect” boundary conditions; (b) the global climatemodel CMCC-CM (0.75◦, about 85 km) to as-sess the uncertainties of the GCM/RCM couplebefore use it to climate change studies,in thiscase the simulation period covers 1971-2005,an extension to 2010 of the simulation is ob-tained forcing the GCM with the RCP4.5 sce-nario.

Numerical simulations over the whole Italiandomain are performed on a cluster of 30 IBMP575 nodes, installed at CMCC (Italy); eachnode has 32 Power6 (4.7 GHz) cores, amount-ing to a total of 960 cores and an aggregatepeak power of 18 TFlops. The elapsed timeto simulate one climatological year using 512cores is about 24 hours.

3. TEST CASES

3.1 ORVIETO AND CERVINARA

The test cases sites of Cervinara (SouthernItaly) and Orvieto (Central Italy), Fig.2, are ofinterest to investigate the impacts of climatechange on different types of landslides phe-nomena. Orvieto is an historical town located100 km North to Rome. It rises on top of a50 m thick tuff slab delimited by subvertical lat-eral cliffs overlying overconsolidated clays. Inthe deeper part, these are stiff and intact, butthe shallowest part of the deposit is jointed andfissured. The clayey slopes are blanketed byan irregular cover of talus and slide debris [32].Since prehistoric times, failures and slow move-ments have been affected the Orvieto slopes :the two historically failures events (Porta Cas-sia, on northern slope, 1900 and Cannicella, onsouthern slope, 1979) were induced by man-made changes to slope geometry or hydraulicconditions; ongoing slow movements (transla-tional) are directly related to soil-atmosphere

interaction. Deep movements occur along pre-existing slip surfaces located within the soft-ened part of clay formation (displacement ratesfrom 2 to 6 mm/years ) while, shallow move-ments, superimposed to the deep ones, in-volve the debris cover and show higher dis-placement rate (displacements between 7 and12 mm/month) [42].

Cervinara slope is located 50 km North-East toNaples; on December, 16, 1999, it has beenaffected by a rapid flowslide triggered by a totalprecipitation of 320 mm in about 50 h, causinghuge damage and five casualties in surround-ing areas. In the area, highly fractured calcare-ous mountains are maintained by loose unsat-urated pyroclastic covers hardly thicker than 2-3 m. These morphological configurations arewidespread in the Campania Region, as re-sult of the activity of some volcanoes includingSomma-Vesuvius.

3.2 PO RIVER BASIN

Po river is the longest river in Italy with a lengthof 652 km from its source in Cottian Alps (atPian del Re) to its mouth in the Adriatic Sea,in the north of Ravenna and it is the largestItalian river with an average discharge of 1540m3/s. The area covered by Po river basin isabout 71000 km2 in Italy and about about 3000km2 in Switzerland and France. The orogra-phy of the basin is quite complex since it isbounded by Alps, Apennines with the Po val-ley between them, Fig.3. In the context of theItalian Law 183/1989, the Po river basin is clas-sified as being of national relevance. Duringthe last centuries, flooding events, due to ex-treme meteorological conditions, of the Po riveror of its tributaries have caused numerous natu-ral catastrophes, two of them, characterized byextraordinary large scale, occurred in the last10 years, [48] and signal of changes in precip-itation and temperature are present in climate

06

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


Figure 2:a) geographical location of the two case-histories; b) Orvieto slope; c) Cervinara landslide

Figure 3:Po basin: hydrological network and main closure sections

07

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

observations [43, 9, 40]. Climate data anal-ysis show, on Po river basin, an increase inannual maximum precipitation with a trend ofabout 0.5◦C/decade since 1960. The signalis detectable in all the seasons [43], in par-ticular, in Summer where maximum tempera-ture are higher than the reference climate [40].Change in precipitation, since 1980, are lessevident than those in temperature, in average,there the precipitation event are more intensebut less frequent, as results the annual totalprecipitation is reduced of 20%. At seasonalscale, the highest reduction rate are found inSpring and Summer (up to 50%) while autum-nal precipitation are almost unvaried; in Win-ter, snowfalls reduces as well [40]. [9] analysetime series of daily cumulated precipitation andof daily minimum and maximum temperaturesin the period 1952-2002 from Piemonte and inthe Valle d’Aosta regions (north-western Italy)finding a significant increase of about 1◦C onaverage temperatures, in particular, for maxi-mum daily temperatures in winter and summermonths; while for precipitation any significanttrend is identified.

4. CLIMATE PROJECTIONS FORORVIETO AND CERVINARA TESTCASES

For the landslide case studies, Orvieto andCervinara, the trends concerning the main fea-tures of temperature and precipitation, for XXIcentury under RCP4.5 and RCP8.5 scenarios,are presented; in particular, firstly, a compre-hensive characterization of seasonal weatherpatterns is carried out displaying: the evolutionof seasonal cumulative rainfall values for thecontrol period 1981-2100 and, through Mann-Kendall test, evaluation of statistical signifi-cance of trends estimated through Thiel-Senapproach (see Appendix); in the same way,trends and statistical significance are computed

for seasonal maximum and minimum tempera-ture. In addition, for 1981-2010 and three futureprojection periods: 2021-2050, 2041-2070 and2071-2100 are shown: (i) mean monthly cumu-lative values; (ii) wet days (daily rainfall>1 mm);(iii) mean monthly maximum daily precipitation;(iv) mean monthly rainfall value for event; (v)mean monthly maximum and minimum temper-ature, moreover, the climate signal, computedas ratio for rainfall and difference for tempera-ture, is reported.

Finally, the trend of proxy variables that morespecifically influence the slope stability for thetwo case studies are considered. The refer-ence variables for slow (very slow movements)affecting clayey slopes of Orvieto are: maxi-mum yearly precipitation values cumulated over30 and 60 days (of particular interest, for shal-low movements in debris covers) and over 120and 180 days (associated to deep movementsin clayey formations); for flowslide triggeringin pyroclastic covers, Cervinara case, back-analysis of case histories [30, 14] in similar geo-morphological contexts have recognized as sig-nificant the effect of a main precipitation eventon 1-2d scale coupled to antecedent precipita-tions on time span directly linked to hydraulicbehaviour of pyroclastic cover and slope fea-tures, i.e. pyroclastic mantle thickness or slopeangle: thus the proxy variables considered fortrend analysis are: maximum yearly 2 days pre-cipitation, cumulative values over 30 and 60days, and the trends in cumulative value of rain-fall over 58 days before the 2d for which themaximum 2d rainfall value is estimated for everyyear. For each variable, trend slopes detectedby Thiel-Sen approach and statistical signifi-cance through Mann-Kendall approach are alsocomputed.

At the beginning, for completeness, on the con-trol period 1981-2010, for both test cases, theperformances of climate models in terms of (i),

08

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


(ii), (iii) and (iv) indicators are compared withobserved data to briefly recall which are theskills of adopted climate modelling in reproduc-ing observed weather patterns on investigatedareas; however, a wider discussion about suchissue is reported in [32, 33, 31].

4.1 SUMMARY OF PERFORMANCESOF CLIMATE MODELS ON CONTROLPERIOD

Figure 4 reports observed (black lines) and sim-ulated (red lines) values of cumulative monthlyrainfall (first row), monthly wet days (sec-ond row), monthly maximum daily precipitation(third row) and monthly average maximum andminimum temperature (forth row) on the con-trol period 1981-2010 for both Orvieto (left col-umn) and Cervinara (right column) case stud-ies. First of all, two remarks are relevant: forallowing the comparison between climate con-ditions, on current period, between the two casestudies, the trends are reported adopting thesame intervals; and, since both Orvieto andCervinara cases are essentially investigated atpoint scale (represented by the investigatedslopes) while the effective resolution of climatemodels can be assumed about equal to 3-7∆xthe nominal resolution, in this case about 8 km,according the weather variable of interest andthe specific features of used RCM [36, 3, 21].For Orvieto, an assessment on the effective res-olution (respectively equal to 3-, 5- and 7-timesthe nominal resolution) is shown in Fig.4(a) find-ing very slight variations in performances bothin terms of precipitation that of temperature asfunction of the resolution, thus, for all the furtherelaborations shown in the following only the in-termediate value 5∆x of effective resolution isdisplayed.

Coming back to the analysis of performances ofclimate models, concerning Orvieto, seasonalcumulative values show an annual cycle char-

acterized by two peaks the first one located inlate winter season and the second one, moreprominent, during the Autumn; the climate sim-ulation is able to properly reproduce the firstone while larger underestimations are returnedduring the Autumn and the entire dry season(Fig.4(a)); as displayed, in wet days, Fig.4(c),although the cumulative values are well repro-duced during the first part of the year, they arethe result of an excessive number of estimatedwet days while, in the second part, a better cor-respondence is returned but in the face of theabove shown substantial underestimation. Re-garding monthly average maximum daily pre-cipitation, Fig.4(e), observed values return sea-sonal patterns similar to those shown for cumu-lative values but also, in this case, climate sim-ulations partially fail to reproduce the autumnalpeak estimating values about equal to 20 mm/dvs 35 mm/d observed during the wet seasonand lower than 10 mm vs about 20 mm/d fromobservations in the dry season. Finally, for tem-perature Fig.4(g), observed values provide theusual annual cycle detected in Mediterraneanarea with winter values ranging between 1◦Cand 10◦C and summer ones in the interval 15-30◦C; however, climate simulation show verydifferent performances for maximum (where bi-ases are about equal to 3◦C during the entireyear) and minimum temperature (with biasesoften lower than 1◦C); such difference couldbe partly due to various current weaknesses ofclimate simulations to properly reproduce tem-perature dynamics during the daytime [6, 7].For what concern Cervinara case study, despiteits position (300 km south of Orvieto), its loca-tion on Apennines of Campania Region entailsrainfall amounts substantially higher than thoserecorded for Orvieto town; again, a seasonalpattern with two peaks is detected but, for thisone, the rainfall heights during late Winter andAutumn reach respectively, on average, valueshigher than 150 mm/month and 250 mm/month,

09

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

Fig.4(b). In this respect, climate simulation failsto properly reproduce cumulative values withhighly lower rainfall heights (maximum valuesabout equal to 100 mm/month); despite this sig-nificant underestimation (which affects in com-parable way also the daily maximum precipita-tion values), an higher number of wet days isestimated during the first part of the year whilethe subsequent underestimation is however di-rectly associated to strong underestimation ofcumulative values, Fig.4(d) and (f). Finally, con-cerning temperature trends, the foregoing con-siderations about Orvieto case can be verba-tim reiterated, Fig.4(h). Therefore, in generalterms, the performances of climate simulationfor landslide hot spots substantially follow theaverage results on the entire Italian domain[6, 7]; in the same way, the main recognizableweaknesses are essentially associated to gen-eralized underestimation of cumulative valuesand maximum daily precipitation but with a con-current overestimation of wet days; in term oftemperature values, climate simulations showvery high skills in reproducing minimum tem-perature values while higher biases affect max-imum temperature evolutions.

4.2 CLIMATE SCENARIOS FORORVIETO CASE STUDY

In order to frame, in broad terms, the expectedvariations in key variables, precipitation andtemperature, while taking into account the dif-ferences in the seasonal dynamics, in Fig.5the trends of seasonal cumulative precipitationscovering 1981-2100 period under both RCP4.5and RCP8.5 concentration scenarios are dis-played (trend lines, evaluated through Theil-Sen method, and indication about statisticalsignificance, assessed through Mann-Kendallapproach, are also marked); while, for seasonaltemperature, the trend slope and the indicationof statistical significance are reported in Tab.1.

In term of precipitation, a sharp difference in be-haviour is revealed between fundamentally dry(MAM and JJA) and wet (SON and DJF) sea-sons; the first ones display a substantial reduc-tion particularly marked for the summer season(statistical significant at 0.1% for both RCPs)and for the spring season under RCP8.5 sce-nario (statistical significant at 1%) while, for thesecond ones, values remain almost unchanged(DJF under RCP4.5) or characterized by notsignificant increase (i.e. DJF under RCP8.5and SON under RCP4.5) or slight decrease(SON under RCP8.5). In term of temperature,the trends are instead completely unambigu-ous: for all seasons and under both RCPs anincrease, statistical significant at 0.1%, is as-sessed for minimum and maximum tempera-ture; under RCP8.5, the increases attain 7-8◦C/century in summer season while they re-main, on average, at 4-5◦C in other seasons;under the intermediate RCP4.5, the increasestands instead, averagely, on 3◦C.

Moreover, in Fig.6, on monthly scale, for con-trol period 1981-2010 (for which 2006-2010RCP4.5 is adopted) and 2021-2050, 2041-2070 and 2071-2100 projection periods, thevariations in different features of precipitationpatterns (and associated climate signals) areinvestigated. In this regard, the strong assump-tion according which the deficiencies of the cli-mate model simulations affect, in the same way,the projections regardless of the considered pe-riod or the influence of radiative forcing, allowsto consider reliable the absolute signal climatedetected between two different simulated peri-ods. The cumulative values show roughly whatseen in Fig.5; decreases in dry seasons re-sult main function of time horizon and “sever-ity” of RCP scenarios (attaining reduction un-til 80% for 2071-2100) while, during the wetseason, the increases generally do not exceed20%; against these cumulative values, reduc-

10

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


Figure 4:Orvieto case study: a) monthly cumulative values observed (black line) and simulated by climate models adopting an effective

resolution equal to 3∆x (red dashed line), 5∆x (red continuous line), 7∆x (red line with filled red dots); c) wet days for observedand simulated values; e) monthly maximum daily precipitation for observed and simulated values; g) maximum and minimum

temperature for observed and simulated values. Cervinara case study: b) monthly cumulative values observed (black line) andsimulated by climate models adopting an effective resolution equal 5∆x (red continuous line); d) wet days for observed and

simulated values; f) monthly maximum daily precipitation for observed and simulated values; h) maximum and minimumtemperature for observed and simulated values

Table 1Orvieto case study slopes of trend lines detected, through Theil-Sen approach, in evolutions of seasonal maximum and minimum

temperature for RCP4.5 (upper part) and RCP8.5 (bottom part); statistical significance, assessed through Mann-Kendall approach,is also reported (+ at 10%, * at 5%, ** at 1%, *** at 0.1%)

Max T2m Min T2mStatistical significance ◦C/year Statistical significance ◦C/year

RCP4.5 DJF *** 0.02 *** 0.03MAM *** 0.03 *** 0.03JJA *** 0.04 *** 0.04SON *** 0.03 *** 0.03


11

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

iFigure 5:

Orvieto case study: trends of seasonal precipitation on time periods 1981-2100 for RCP4.5 (left column) and RCP8.5 (rightcolumn)scenarios; are also reported trend line estimated through Theil-Sen and indications about statistical significance (+ at 10%,

* at 5%, **at 1%, *** at 0.1%) estimated through Mann-Kendall approach

12

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


tions (with rare exceptions) of rainy days gener-alized are estimated during the entire year. Thecoupled effect of the two behaviours is there-fore revealed in Fig.6(d) where, on monthlyscale, average precipitation heights for eventare shown (trivially given as the ratio betweenthe two previous quantities): of course, an in-crease in climate signal is returned for monthscharacterized by an increase in cumulative val-ues; however, also in other periods during theyear where a reduction of total values is es-timated, the amount for event is expected in-creasing (average growing about equal to 20-30%) or, at least, nearly steady. In this regard,the summer season under the RCP8.5 scenarioconstitutes an exception due to the reductionin cumulative values so substantial to repre-sent, probably, the main reason of reductionin wet days. However, concerning maximumdaily precipitations, the main displayed trendsare confirmed; the increase in maximum val-ues during the wet season, exceed 30% duringthe Autumn while lower in late Winter; how-ever, during dry season, again a reduction isestimated; in this regard, however, it is well re-calling also the weaknesses of the current re-gional climate models in reproducing convec-tive dynamics essentially governing precipita-tion events during the dry season partially af-fecting the shown trends.

Displayed results are quite consistent withthose found by other studies on the Mediter-ranean area (i.e. PRUDENCE and ENSEM-BLE project); in the specific, the atmosphericwarming especially tends to affect the precip-itation patterns (dry days and precipitation forevent) because of higher atmospheric moistureretention capacity; in term of precipitation val-ues, at the same time, soil moisture depletioninduced by warming during the dry season, en-tail a sharp reduction of precipitation amountsmainly under the RCP8.5 scenario.

The last issue deals, in greater detail, the trendof “proxy” variables that, through the back anal-ysis of the slope movements and field monitor-ing [42, 41, 5], have been assumed as directlyrelated to the movements of the landslide bod-ies: in Fig.7, for both RCPs, are then displayedthe trends on 1981-2100 period of maximumyearly value of cumulative precipitations on 30,60, 120 and 180 days.

Indeed, the low permeability characterizing thesoils in situ (10−7 m/s for debris cover until 10−9

m/s for intact deep clays) induce a “time lag”between rainfall events and the arrival of thecorresponding infiltrated rate at the soil depthin which the sliding surfaces arise; the increaseof water content, producing the growth of waterpressures, induce a reduction of frictional resis-tances and then the potential reacceleration oflandslide bodies; moreover, the temporal shiftis main function of soil depth and inversely pro-portional to permeability of involved soils; then,the cumulative values on 30 and 60 days areessentially correlated to shallower movementswhile 120 and 180 days to the deepest ones.However, considering for each year as proxyvariable only the maximum value could providea generic information on possible occurrenceof the movements but not about their duration.In Fig.7, for both RCPs and for the four refer-ence time span, the trends show an increasebut only for the shorter durations it is assumedstatistical significant (at 5% for RCP4.5 and 1%for RCP8.5). This findings might suggest an al-most invariance of the existing rate movementfor deep landslide bodies and a presumableworsening of slope stability conditions for shal-lower movements; on the other hand, on suchreference time periods, also the effect of evap-oration removing water from the soil before itarrives in depth could play a remarkable role inthe special way, investigating future scenarios,characterized as shown in Tab.1, by a strong ex-

13

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

Figure 6:Orvieto case study: bottom part: values on 1981-2010 (red line), 2021-2050 (blue line), 2041-2070 (green line), 2071-2100 (orangeline); upper part: associated climatic signal: 2021-2050 vs 1981-2010 (blue line), 2041-2070 (green line), 2071-2100 (orange line)

for mean cumulative precipitation values (a), wet days (b), mean monthly maximum precipitation value (c), mean monthlyprecipitation for event (d), average monthly maximum and minimum temperature (e)

14

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


Figure 7:For Orvieto case study: trends on time periods 1981-2100 for RCP4.5 (left column) and RCP8.5 (right column)scenarios of: (a) and

(b) maximum yearly precipitation value over 180 days; (c) and (d) maximum yearly precipitation value over 120 days; (e) and (f)maximum yearly precipitation value over 60 days; (g) and (h) maximum yearly precipitation value over 30 days

15

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

pected increase in the temperature values. Forthese reasons, more refined analysis (through,for example, physically-based models adoptedin [31]) are required to better understand if cli-mate changes could affect in different waysshallow and deep movements (as conceivablefrom Fig.7) or the increase of temperature couldbe such to induce a substantial soil moisturedepletion and subsequent improvements of av-erage slope stability conditions.

4.3 CLIMATE SCENARIOS FORCERVINARA CASE STUDY

The framework carried out for the other casestudy, constituted by flowslides affecting Cerv-inara (Avellino, Southern Italy), is fully similarto that, above described, for slow movementsof Orvieto. Also in this case, Fig.8, underboth RCPs, a dramatic decrease in cumula-tive values is estimated during the dry season(statistically significant for Summer and Springat 0.01% under RCP8.5 and at 1% for Sum-mer under RCP4.5); for winter season, for thetwo scenarios, remarkable variations in cumu-lative values are not assessed while, finally, forSpring season, the behaviour substantially dif-fer with the “stabilization” scenario (RCP4.5) re-turning a slight (not significant) increase andthe more “extreme” RCP8.5, confirming, also,in this case a significant (at 5%) decrease incumulative values.

Against these differences, the projections aboutthe temperature, Tab.2, are totally in agree-ment in estimating a substantial increase inminimum and maximum temperatures (statis-tically significant at 0.1% for all seasons andboth RCPs); the magnitude of such increasesis roughly equivalent to that estimated for Orvi-eto (Central Italy).

Refining the representation showing the data,for the 30 years periods considered, on monthly

scale, Fig.9, a further confirmation of theseresults through the ratio defining the climaticsignal between future and current time inter-vals constantly below the unit value (with re-ductions attaining 60% in dry seasons underRCP8.5); the coupled effect of decreases ofcumulative values and increase in atmosphericmoisture retention capacity (induced by warm-ing) strongly affect the number of wet days, es-timated decreasing, with rare exceptions, forall reference future time spans, Fig.9(b). Alsoin this case, the combined effect of evolu-tions of monthly cumulative values and wetdays is assessed through average “Precipita-tion for Event” index, Fig.9(d); for this one,under RCP4.5, the climatic signal, for all thetime intervals, is kept almost unchanged in thefirst part of the year (approximately Winter andSpring) while it is evaluated experiencing in-creases up to 30% in late Summer and Au-tumn partially proving that in such periods, theeffect of increased atmospheric retention ca-pacity could play a greater role compared toreduction in cumulative values in reduction ofwet days; as regards the RCP8.5 scenario, thetrend seems more fluctuating but mainly for theperiod 2071-2100, the decrease in cumulativevalues could regulate the estimated future pre-cipitation patterns. Finally, concerning averagevalues of maximum daily precipitation, underRCP4.5 scenario, for a large part of the year,increases attaining, on average, 30% are esti-mated while for RCP8.5, the high reduction intotal amounts involve a reduction also in aver-age maximum values, mainly in dry seasons,while in wet seasons increases up, on aver-age, 20% are estimated mainly for farther 2071-2100. In this regard, it is worth remembering,however, again the possible weaknesses of theclimate modelling in reproducing convective dy-namics, especially significant, in the warmerseasons. With slight differences between maxi-mum and minimum values, also climate signals

16

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


Figure 8:for Cervinara case study; trends of seasonal precipitation on time periods 1981-2100 for RCP4.5 (left column) and RCP8.5 (right

column)scenarios; are also reported trend line estimated through Theil-Sen and indications about statistical significance (+ at 10%,* at 5%, ** at 1%, *** at 0.1%) estimated through Mann-Kendall approach

Table 2For Cervinara case study slopes of trend lines detected, through Theil-Sen approach, in evolutions of seasonal maximum and

minimum temperature for RCP4.5 (upper part) and RCP8.5 (bottom part); statistical significance, assessed through Mann-Kendallapproach, is also reported (+ at 10%, * at 5%, ** at 1%, *** at 0.1%)

Max T2m Min T2mStatistical significance ◦C/year Statistical significance ◦C/year



17

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

provided for temperature increases, are sub-stantially equivalent to that detected for Orvieto.

Finally, the last part of discussion is aimed todiscuss the trends estimated for those variablesthat can be assumed as “proxy” in triggeringof flowslide events in Cervinara area. Accord-ing what is above recalled, trends of maximumyearly values of cumulative precipitation on 30and 60 days (1981-2100) are assumed to berelevant to assess the tendencies in poten-tial “antecedent precipitation” while maximumyearly precipitation on 2 days is considered asreference for “main” triggering event, Fig.10.

Moreover, as, for flowslides in pyroclastic soils,the trigger is regulated by concomitant pres-ence of a particularly wet period followed byan intense precipitation event, in order to as-sess future trends only of the main eventsin the periods of potential interest, also thetrends in the maximum yearly precipitation val-ues on 2 days occurring only during wet sea-son, October-March, and the correspondingantecedent precipitation on 58 days are consid-ered. For what concern the trends on 30 and 60days, slight increases (almost unchanging andnot statistically significant) are evaluated underboth RCPs while, for the main event on 2 days,a more remarkable increase is assessed bothconsidering the entire year that the only wetseasons; moreover, for RCP4.5 scenario, it isassociated to a significant increase (although at10%) of the corresponding antecedent precipi-tations on 58 days. Under the strong assump-tion of assuming all the others factors, currently,governing the phenomena as steady, such find-ings might suggest a not-negligible worseningof average slope stability conditions resultingin a possible increase in the number of oc-currences of flowslides events. On the otherhand, this preliminary estimate, mostly qual-itative, carried out in “isothermal conditions”,neglects the potential beneficial effect of the in-

creased evapotranspiration on the slope stabil-ity; for these reasons, also in this case, clearerindications about the possible trends requirethe adoption of more physically based tools al-lowing, for example, to estimate soil surface wa-ter balances and the linked variations in soilsuction regulating the triggering of landslideevents.

18

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


Figure 9:Cervinara case study: bottom part: values on 1981-2010 (red line), 2021-2050 (blue line), 2041-2070 (green line), 2071-2100(orange line); upper part: associated climatic signal: 2021-2050 vs 1981-2010 (blue line), 2041-2070 (green line), 2071-2100(orange line) for mean cumulative precipitation values (a), wet days (b), mean monthly maximum precipitation value (c), mean

monthly precipitation for event (d), average monthly maximum and minimum temperature (e)

19

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

Figure 10:for Cervinara case study: trends on time periods 1981-2100 for RCP4.5 (left column) and RCP8.5 (right column) scenarios of: (a)

and (b) maximum yearly precipitation value over 58 days before the corresponding maximum 2 days event simulated in wet season;(c) and (d) maximum yearly precipitation value over 2 days simulated in wet season; (e) and (f) maximum yearly precipitation value

over 2 days considering the entire year; (g) and (h) maximum yearly precipitation value over 60 days; (i) and (l) maximum yearlyprecipitation value over 30 days

20

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


5. CLIMATE PROJECTIONS FOR PORIVER BASIN TEST CASE

For the flood and drought case study area,i.e. Po river basin, the main features of tem-perature and precipitation, for XXI century un-der RCP4.5 and RCP8.5, are investigated interms of anomalies with respect to the controlperiod 1982-2011 of seasonal weather patternand of annual cycle; moreover trend and dis-tribution probability are analysed at seasonalscale. The statistical significance of trends istested using Mann-Kendall test. The capabil-ity of the GCM/RCM couple to reproduce theobserved climate in terms of seasonal patternsand annual cycle is presented before discussthe climate projections. Differently to landslidephenomena, to deal with floods/droughts phe-nomena on Po river daily precipitation and tem-perature data are sufficient due to the basin ex-tension, in a smaller basin the study of floodswill require sub-daily time series.

5.1 SUMMARY OF PERFORMANCESOF CLIMATE MODELS ON CONTROLPERIOD

This section presents the validation of theclimate outputs, precipitation and 2 metermean temperature, simulated by COSMO-CLMdriven by ERA-Interim reanalysis [11] and bythe GCM CMCC-CM [35] with respect to agridded climate dataset provided by the Hydro-Meteo-Climate Service of the Regional Agencyfor Prevention and Environment (ARPA SIMC)of the Emilia-Romagna Region. The dataset isbased on the meteorological data published inthe Hydrological Yearbooks and integrated withobservations, validated and quality checked,collected by the stations network in Po rivercatchment. The interpolation method used togenerate the dataset is the inverse of the dis-tance; for temperature the orographic effect is

considered using a fixed gradient. The pre-cipitation data span from 1971 to 2010 andthose of temperature cover the period 1990-2010 due to the low density of the tempera-ture measurements network before 1990. Thevalidation period is 1991-2010, Fig.11, alterna-tively, the validation over a 30 year long pe-riod of the RCM outputs both driven by cli-mate reanalysis and by the GCM, is avail-able in [6, 7]. Figure 11 shows the sea-sonal bias maps of COSMO-CLM driven byERA-Interim Reanalysis and CMCC-CM globalmodel against observations. Concerning tem-perature (Fig.11(a)), ERA-Interim driven simu-lation shows a very good agreement with theobservations in Spring and Autumn, with a biasclose to 0◦C in most of the domain under study.In Winter, a general cold bias is registered (withpeaks of -3◦C), probably ascribed to an un-derestimation of the real orography, whereasin Summer a general temperature overestima-tion occurs (with the exception of the Alpinearc). The summer warm bias can be caused bya tendency of COSMO-CLM to underestimatelatent heat fluxes (with a consequent overesti-mation of sensible heat fluxes) [2], along withan underestimation of cloud cover. It is worthnoting that a general tendency of RCMs to un-derestimate winter temperature, in particularover mountainous chains and Alps, e.g. [10],and overestimate the summer ones, e.g. [4],especially in semi-arid regions and south Eu-rope is reported in literature. However, with re-spect to other studies at coarser resolution, thebias found in this work is lower, highlighting therole of the horizontal resolution [29]: for exam-ple, some RCMs driven by ERA-Interim reanal-ysis involved in EURO-CORDEX project [23]at horizontal resolution of 0.11◦ exhibit even -5◦C of difference with respect to observationsin Winter, whereas in Summer, overestimationover Italian domain generally ranges between+2 and +3◦C. The CMCC-CM driven simulation

21

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

Figure 11:Bias of (a) 2-meter mean temperature (◦C) and (b) precipitation (%) of ERA-Interim/COSMO-CLM and CMCC-CM/COSMO-CLM

simulations for all the seasons

Table 3Average value of 2 meter mean temperature (◦C) and precipitation (mm/day) of observation (OBS) and synthetic errors (BIAS,

MAE, RMSE) of the same fields from COSMO-CLM driven by ERA-Interim and CMCC-CM with respect to observation for the fourseasons.

T2m (◦C) P (mm/day)ERA-Interim/ CMCC-CM/ OBS ERA-Interim/ CMCC-CM/ OBS

Season Index COSMO-CLM COSMO-CLM COSMO-CLM COSMO-CLM COSMO-CLM COSMO-CLMAverage 0.14 -0.63 1.02 2.08 2.89 1.97

DJF BIAS -0.89 -166 - 0.10 0.91 -MAE 0.91 2.30 - 0.39 1.79 -RMSE 1.01 2.85 - 0.50 2.13 -Average 9.12 6.86 9.51 3.42 4.53 2.86

MAM BIAS -0.38 -2.64 - 0.56 1.67 -MAE 0.49 2.82 - 0.68 2.28 -RMSE 0.63 3.30 - 0.89 2.79 -Average 19.82 17.51 19.13 2.72 2.86 2.58

JJA BIAS 0.69 -1.62 - 0.14 0.27 -MAE 0.81 2.10 - 0.66 1.39 -RMSE 0.98 2.50 - 0.89 1.87 -Average 9.96 8.92 10.12 3.31 3.07 4.13

SON BIAS -0.16 -1.20 - -0.82 -1.06 -MAE 0.50 1.87 - 1.02 2.29 -RMSE 0.59 2.30 - 1.40 3.02 -

22

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


is affected by a general cold bias in all sea-sons (higher than the simulation driven by "per-fect" boundary conditions), more pronouncedin Spring, where peaks of -5◦C are reached; itis partially due to the general tendency of theAtmosphere-Ocean General Circulation Mod-els to underestimate the temperature [24]; inparticular, [18] has verified that CMCC-CM isgenerally affected by a cold bias up to -2◦Cover the Mediterranean area. For precipita-tion, Fig.11(b), both simulations are affectedby an overestimation over Alps. It is moreevident in Spring, when the simulation forcedby CMCC-CM shows the highest bias. Overthe plain area, instead, an underestimation oc-curs, stronger in Autumn. The error occurringover mountainous areas has already been re-ported in other works, e.g. [19], and it can becaused by the joint underestimation of the oro-graphic effects joint and of precipitations, thelatter caused by the hydrometeor deflection dueto wind, that lead to an undercatchment of theobserved rainfall [1, 15]. Table 3 reports thevalues of three objective quantities for perfor-mance evaluation, namely BIAS (Model minusObservation), MAE (Mean Absolute Error) andRMSE (Centered Root Mean Square Error);they have been computed for both ERA-Interimand CMCC-CM driven climate simulations spa-tially averaged over the whole Po river basin.

BIAS =1

N

N∑i=1

(Si −Oi) (1)

MAE =1

N

N∑i=1

|(Si −Oi)| (2)

RMSE =

√√√√ 1

N

N∑i=1

(Si −Oi)2 (3)

where N is the size of the sample, Si and Oi

are, respectively, simulated and observed valueat the i-th time step. The index BIAS providesestimate of the average error, but it is affected

by a possible error compensations, whereasMAE and RMSE yield the average magnitudeerror without indicating the direction of the devi-ation (in addition, the latter tends to emphasizelarge errors than smaller ones). Results showa very good agreement between ERA-Interimforced simulation and observations, both for2 meter mean temperature and precipitation(mean BIAS never exceeds 1◦C and 1 mm/dayin absolute value). Spring and Autumn arecharacterized by the lowest T2m BIAS (-0.38◦Cand -0.16◦C, respectively), but they are charac-terized by the highest precipitation error, witha general Spring overestimation and Autumnunderestimation (however smaller than 40%).CMCC-CM driven simulation is characterizedby a more pronounced error, in all seasonsand for both the variables under study. BIASis higher than 100% in winter season, both fortemperature and precipitation; that is stronglyoverestimated also in Spring (more than 1.67mm/day on average).

5.2 CLIMATE PROJECTIONS

In the next section, we present the out-puts of the climate projections at 2021-2050,2041-2070 and 2071-2100 under RCP4.5 andRCP8.5 scenarios on Po river basin. For eachperiod, the estimated precipitation and temper-ature are illustrated in terms of spatial and tem-poral anomalies with respect to 1982-2011, andin terms of trends and probability distributionfunctions (pdfs). The statistical significance of(linear) trend has been tested through the nonparametric Mann-Kendall test.

Period 2021-2050 The first period under anal-ysis is 2021-2050 and within this period, theRCP4.5 and RCP8.5 scenarios agree in pro-jecting a decrease in summer precipitations, (-20% and -15%, respectively) and an increase inthe autumnal ones (+2.1% and +8.2%, respec-tively), in particular, along the main channel

23

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

of Po river (Fig.12(a)). Under the stabilizationscenario RCP4.5, in Winter, the precipitationreduces of about 8.4% mostly on the north-eastern side of basin, while, under RCP8.5,it increases of about 7.4%. In Spring, underRCP4.5 scenario the simulated precipitation islower than in the control period (-11%) andit is almost unvaried (-1.1%) considering theRCP8.5 projection. Temperature are increas-ing under both scenarios over the entire basin,considering RCP4.5 emissions, the tempera-ture anomaly ranges between 0.9◦C (in Winter)and 1.7◦C (in Summer), and it varies between1.3◦C (Spring and Autumn) and 1.7◦C (in Win-ter) under RCP8.5, Fig.12(b).

Figure 13 shows the anomaly in monthly pre-cipitation (a) and 2 meter mean precipitation(b). For precipitation, Fig.13(a), the two scenar-ios are in agreement, monthly anomalies rangebetween -32% in July and +18% in November(RCP4.5) or between -47% in July and +26%in December, if RCP8.5 emissions are con-sidered. The shape of the monthly tempera-ture anomaly, Fig.13(b), is similar under bothemission scenarios but the magnitude is dif-ferent, 0.7-3.1◦C for RCP4.5 and 2.3-4.9◦C forRCP8.5.

In 2021-2050, RCP4.5 and RCP8.5 climateprojections show a substantial agreement inseasonal trends and variability of average pre-cipitation and 2 meter mean temperature asshown in Fig.14. According to Mann-Kendalltest, any of the seasonal precipitation trend isstatistically significant at 5%, while, for tem-perature, the linear trend hypothesis is notrejected in Summer (Autumn) under RCP4.5(RCP8.5) scenario. With more details, in Win-ter, RCP4.5 and RCP8.5 precipitations arecharacterised by slopes close to zero (0.01and -0.003 mm/day/year, respectively) bothhigher than the one of the control period (-0.02mm/day/year); in Spring, the slopes are posi-

tive: 0.03 and 0.02 mm/day/year, respectivelyversus 0.01 mm/day/year of the control period;in Summer, a negative tendency is found (-0.03and -0.02 mm/day/year, respectively) compa-rable with the one (-0.03 mm/day/year) of thecontrol period; and, in Autumn, the trend is pos-itive (0.01 mm/day/year) for RCP4.5 projectionsand negative (-0.02 mm/day/year) according toRCP8.5 scenarios, in the control period the au-tumnal trend is positive (0.02 mm/day/year).For temperatures, both simulations in all sea-sons show positive trends; in particular, in Win-ter, RCP4.5 trend is higher than RCP8.5 one(0.04 vs 0.02◦C/year) but lower than in thecontrol period (0.05◦C/year); the same con-siderations hold for Spring with trends equalto 0.05◦C/year (control period); 0.02 (RCP4.5)and 0.002◦C/year (RCP8.5). In Summer, thetemperature growth rate of RCP4.5 simula-tion (0.07◦C/year) is comparable with the oneof the control period (0.06◦C/year) and higherthan RCP8.5 one (0.02◦C/year); in Autumn,RCP8.5 simulation is characterised by a slope(0.05◦C/year) comparable with the control pe-riod (0.05◦C/year) and while RCP4.5 simulationslope (0.02◦C/year) is lower.

The seasonal pdfs of precipitation and 2 me-ter mean temperature are reported in Fig.15.In Winter, the probability of null precipitation(P ≤ 1 mm/day) is almost unvaried with re-spect to 1982-2011 period, while low precipi-tation rate show a slight increase in their fre-quency; for temperature a shift of about 1◦C,toward right, in the modal value is evidencedunder both scenarios, that show a different vari-ability: between -15◦C and 9◦C (RCP4.5) andbetween -12◦C and 9◦C (RCP8.5); the vari-ability in the control period is -15◦C to 8◦C. InSpring, the probability of P ≤ 1 mm/day is in-creasing, while the pdf of P > 1 mm/day is sub-stantially unvaried, also the pdf of temperatureis almost constant. In Summer, the null precip-

24

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


2021-2050 Precipitation 2 meter mean temperature RCP4.5 RCP8.5 RCP4.5 RCP8.5

DJF

MA

M

JJA

SO

N

%

°C

(a) (b)

Figure 12:Anomaly of (a) precipitation (%) and (b) 2-meter mean temperature (◦C) of CMCC-CM/COSMO-CLM simulations under RCP4.5

and RCP8.5 scenarios in the 2021-2050 period with respect to 1982-2011.

Figure 13:Monthly anomaly of (a) precipitation (%) and (b) 2-meter mean temperature (◦C) of CMCC-CM/COSMO-CLM simulations under

RCP4.5 and RCP8.5 scenarios in the 2021-2050 period with respect to 1982-2011.

25

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

Figure 14:Trend in average seasonal precipitation and 2 meter mean temperature in the 2021-2050 under RCP4.5 (blue) and RCP8.5 (red).

Trend in the control period (black) is reported for comparison.

26

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


itation is constant for RCP4.5 and increasingfor RCP8.5, and precipitations between 1 to 10mm/day show a higher frequency than in thecontrol period; the shape of the temperaturepdfs is similar to the control period but shifted ofabout 1◦C towards right. In Autumn, the prob-ability of P ≤ 1 mm/day is similar to the controlperiod for both scenarios, and low precipitationare less frequent than in the 1982-2011 period;the shape of the temperature pdf is similar tothe control period, with a plateaux at 6◦C thatpersists also in the 2021-2040 period.

Period 2041-2070 The second period consid-ered is 2041-2070. Within this period, RCP4.5and RCP8.5 climate projections show morevariability than in the 2021-2050 period previ-ously analysed. Both RCP4.5 and RCP8.5 sce-narios project the decrease of spring and sum-mer precipitations (Fig.16(a)). Under the sta-bilization scenario RCP4.5, in Spring, the pre-cipitation varies of -7.9% mostly on the north-eastern side of basin, while under RCP8.5 thereduction is slightly higher, -9.2%, and the mostaffected areas are localised on the Apennines(southern boundary of the basin). Under bothscenarios, about 1/3 of the summer precipita-tion is lost on the entire basin, in particular inthe Po valley, while in Autumn, the precipitationis expected to increase, but the spatial distribu-tion of the anomaly is different. According toRCP4.5, the precipitation increases (18%) allover the basin, instead following RCP8.5, posi-tive changes are mostly localised along the Poriver main channel and negative on the Apen-nines, and in average the total precipitationis constant being the average anomaly equalto 0.9%. Under RCP4.5, winter precipitationslightly reduces, -3.8%; mostly in the west-ern part of Po Valley and on the Apennines,while on Alps is almost unvaried; consideringRCP8.5 precipitation increases, 7.3%, mostlyon the north-eastern side of the basin. For

temperature, Fig.16(b), both scenarios projecta positive anomaly in all the seasons, rangingbetween 1.7◦C in Spring and 3.1◦C in Sum-mer and between 2.4◦C in Spring and 3.7◦C inSummer, for RPC4.5 and RCP8.5 projections,respectively. Figure 17 shows the anomaly inmonthly precipitation (a) and 2 meter meanprecipitation (b). For precipitation, Fig.17(a),the highest uncertainties, i.e differences be-tween the projections, are detected in January:RCP4.5 presents a negative (-11%) anomalywhile RCP8.5 a positive (+17%) one and inSeptember, where RCP4.5 anomaly is positive(+24%) and the one associated to RCP8.5 isnegative (-8.3%). The shape of the monthlytemperature anomaly, Fig.17(b), is quite similarfor both scenarios but magnitude is quite dif-ferent, 0.8-3.3◦C for RCP4.5 and 1.8-4.1◦C, forRCP8.5. Precipitation projected trends, Fig.18are both positive in Winter, 0.01 mm/day/yearfor RCP4.5 and 0.02 mm/day/year for RCP8.5;both negative in Spring, -0.01 mm/day/year(RCP4.5) and -0.07 mm/day/year (RCP8.5) andSummer, -0.01 mm/day/year (RCP4.5) and -0.03 mm/day/year (RCP8.5) and positive forRCP4.5, 0.02 mm/day/year and negative -0.02mm/day/year for RCP8.5. The RCP8.5 springand summer trends are statistically significantaccording to Mann-Kendall test with a confi-dence level of 5%. The precipitation variabil-ity of control period and projections is com-parable in Winter and Spring, while in Sum-mer both projections show lower precipitationand, in Autumn, the RCP4.5 is characterisedby a higher interannual variability than RCP8.5and the control period. Temperature trendsare positive for both scenarios in all the sea-sons, and, with the exception of Winter, allstatistically significant. The RCP4.5 temper-ature trends are equal to 0.001◦C/year (Win-ter), 0.04◦C/year (Spring), 0.06◦C/year (Sum-mer) and 0.08◦C/year (Autumn); for RCP8.5estimated trends are 0.06◦C/year (Winter),

27

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

Figure 15:Probability distribution function (pdf) of average seasonal precipitation and 2 meter mean temperature in the 2021-2050 under

RCP4.5 (blue) and RCP8.5 (red). The control period pdf (black) is reported for comparison.

28

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i



DJF

MA

M

JJA

SO

N

%

°C

(a) (b)





29

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

0.10◦C/year (Spring), 0.17◦C/year (Summer)and 0.10◦C/year (Autumn) and, in average,temperatures are higher than in the control pe-riod.

The seasonal pdfs of precipitation and 2 me-ter mean temperature are reported in Fig.19.In Spring and Summer, the probability of P ≤1mm/day is characterised by the highest vari-ations, from 38% in the control period to a45% either for RCP4.5 and RCP8.5, and from60% to 63% (RCP4.5) and 67% (RPC8.5), re-spectively. Summer is also characterised bythe highest increase in low (but higher than1 mm/day) precipitations frequency, while lim-ited changes are present in the other seasons.With the exception of Autumn where a fre-quency plateaux is found, the temperature pdfsare all characterised by a shift in the modalvalue. In average, RCP4.5 and RCP8.5 tem-peratures are characterised by higher extremevalues than the control period.

Period 2071-2100

The third period considered is 2071-2100. Con-cerning precipitation (Fig.20(a)), RCP4.5 pro-jection shows a positive anomaly in winter pre-cipitations, in particular over Po Valley whereasnegligible differences are found on Alps: onaverage, the positive variation is about 11%.RCP8.5 projection expects an increase of about38% in the precipitation over the whole Po riverbasin. The RCP4.5 signal in winter precipita-tion is now positive, i.e. there is more precipi-tation than in the control period. A possible ex-planation is related to the positive temperatureanomaly in Autumn/Winter that forces the evap-oration from the soil increasing the air humidityfostering the winter precipitation. In Spring, un-der both scenarios precipitations reduce on theeastern part of the basin, the variation is moremarked under RCP8.5. On the western Alps,according to RCP4.5 a light increase of pre-cipitation is expected while following RPC8.5

it reduces; in average, the winter precipitationvaries of -1.4% (RCP4.5) and -14% (RCP8.5).In Summer, the RCP4.5 precipitation reductionis almost the same (about -28%) of 2041-2070period, but with no more evident dependencyon altitude; under RCP8.5 scenario precipita-tion variation is -57%. In Autumn, RCP4.5precipitation increases of about +5.3% and thevariations are localised on Piemonte and east-ern side of Po river main channel, similarly tothe anomaly shown by RCP8.5 projections on2041-2070; RCP8.5 precipitation reduces onApennines and, partially, on Alps while it is un-varied on the Po Valley, the overall reductionis less than 5%. Concerning temperatures thepositive anomaly detected in 2041-2070 pro-jections is present and more marked, on aver-age, for RCP4.5, it will range between +2.3◦Cin Winter and Spring and +3.5◦C in Summerand, for RCP8.5, between 4.1◦C in Spring and7◦C in Summer, Fig.20(b). Figure 21 showsthe anomaly in monthly precipitation (a) and2 meter mean precipitation (b). With the ex-ception of March, October and November, theuncertainty in the magnitude of precipitationchanges is high, beside that, the tendency isclear: precipitation increases between Novem-ber and April(RCP4.5)/March(RCP8.5) and de-creases in the rest of the year. ConsideringRCP4.5 scenario the changes range between-36% in August and +23% in November and be-tween -67% in July and 46% in December forRCP8.5 projections. Thus, the 2071-2100 pro-jection of precipitation are affected by a higheruncertainty than the previously analysed pe-riods. This is reasonable considering the in-creasing differences in the projected emissions.Also in 2071-2100, the two simulations agree inthe shape of the monthly temperature anomaly,Fig.17(b), that show a variability between 2.7-4.2◦C for RCP4.5 and 3.6-5.2◦C for RCP8.5.

The RCP4.5 projection shows positive trends

30

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i




31

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i



32

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i



DJF

MA

M

JJA

SO

N

%

°C

(a) (b)





33

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

in Winter (0.03 mm/day/year) and Spring (0.01mm/day/year) precipitations and negative inSummer (-0.04 mm/day/year) and Autumn (-0.03 mm/day/year), only, for Summer the nullhypothesis of Mann-Kendall test is not rejectedwith a confidence level of 5%. The RCP8.5projected precipitation is characterised by apositive trend in Winter (0.06 mm/day/year)and negative in the rest of the year, -0.04 mm/day/year (Spring), -0.02 mm/day/year(Summer) and -0.01 mm/day/year (Autumn),any of these trend is statistically significant. It isworth to note that, Winter and Spring precipita-tions show an inter-annual variability compara-ble to the control period, Fig.22. For tempera-ture, under RCP4.5 scenarios, slightly negativetrends are computed in Winter (-0.02◦C/year)and Spring (-0.003◦C/year), a positive trendis found in Summer (0.05◦C/year) while, inAutumn, the trend is null. RCP8.5 tempera-tures are characterised by positive, and sta-tistically significant, trends in all the seasons:0.008◦C/year in Winter, 0.09◦C/year in Spring,0.12◦C/year in Summer and 0.09◦C/year in Au-tumn.

The comparison among the seasonal precipita-tion pdfs show that in Spring, Summer and Au-tumn, the probability of P ≤ 1 mm/day is higherthan in the control period, with the highest prob-abilities associated to RCP8.5 values. In Win-ter, the RCP4.5 (RCP8.5) probability of P ≤ 1

mm/day is equal to (less than) the control periodone, Fig.23. Changes in low precipitation (buthigher than 1 mm/day) frequency are detectedin Summer (increase) and Autumn (decrease).For temperature the shift in the modal values isconfirmed and more marked than in 2041-2070period, Fig.23.

34

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i




35

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i



36

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


6. CONCLUSIONS

In the present work, a further intermediatestep in the developing of the assessment ofthe potential effect of climate changes on geo-hydrological hazards is carried out: the es-timate of the future trends (until 2100 underRCP4.5 and RCP8.5) of the variables, temper-ature and precipitation, mostly regulating soilprocesses which determine the hazards in se-lected hot spots: Po river basin (flood andlow flow hazard), Cervinara and Orvieto (re-spectively, for fast and slow slope instabilityprocess); it follows the assessment of perfor-mances of simulation chain (for single “rings” orfull asset) carried out in [32, 31, 5, 51, 45, 47,46] and essentially consisting in: i)validationof numerical climate models, ii) developmentsof bias correction approaches and iii) calibra-tion and validation of impact tools. The climatevariables are provided by high resolution cli-mate projections, performed within A.2.6 workpackage of GEMINA project [6, 7, 29, 52]; in ac-cordance with was also recalled in the introduc-tion, the high horizontal resolution (about equalto 8 km) characterizing the adopted RCM simu-lations, attempts to “narrow” the current gap ex-isting between the spatial scale of climate simu-lation and impact tools; moreover, it representsa crucial added value in estimating the varia-tion in geo-hydrological hazards strictly linkedto a proper evaluation of extreme values, be-yond representing an indisputable advantage,in general, also for average values on a domain,like Italy, characterized by a very complex geo-morphology. The results achieved confirm thegeneral findings according which, in a not sofar future, Mediterranean areas could experi-ence (1) warmer temperature inducing, in turn,(2) a more pronounced seasonality in precipi-tation, with wetter Autumns (RCP4.5) or Win-ters (RCP8.5) and drier Summers. However,although the “direction” of the indicated varia-

tions seems clear, their magnitude is affectedby well known uncertainties related to climateprojections (first and foremost, the one linked tothe assumptions underlying RCPs). Since pre-cipitation and temperature are the main driversof surface processes, the uncertainty in theirprojections has to be taken into account to cor-rectly estimate the geo-hydrological hazards re-lated to climate change. Regarding the specificnature of hot spots, in the work, parametersof interest and their representation has beentailored according their peculiar features (i.e.extent, reference time spans of hazards) whilekeeping in mind the current weaknesses of cli-mate simulation. Despite the number of casesis limited, even in this case as for the validation,a key point of the carrying out research is rep-resented by the development of a frameworkentirely and blindly replicable in the study ofother geo-hydrological (and not only) hazards.To this aim, it worth noting that the main diffi-culty is related to finding case studies properlydocumented allowing the calibration and vali-dation of the impact tools on Italian domain; onthe other hand, however, they are case stud-ies representative for dramatically widespreadphenomena in the Italian context. Finally, thenext step, already in progress, is represented byevaluation of the effects of changes of weathervariables on hazards through the adoption ofthe specific impact tools.

37

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

A. MANN KENDALL TEST

The non parametric Mann-Kendall test[25, 22]tests the H0 hypothesis of trend absence ver-sus the H1 hypothesis of the existence of apreferential order in the observations. The teststatistics S is computed on the basis of the rela-tive ordering of all possible pairs of data points:

S =

n−1∑i=1

n∑j=i+1

sign(xj − xi). (4)

where sign(·) is 1 if (·) > 0 and -1 if (·) < 0.The variable S is characterised by mean µS = 0

and variance σ2S = n(n−1)(2n+5)/18 The test

variable Zs is relate to S by

Zs =S − sign(S)

σS, (5)

the variable Zs is distributed as N(0, 1), thusa comparison between Zs and the inverse ofN(0, 1) provides the critical value for Mann-Kendall variable. The monotonic trend hasbeen estimated according to Thiel-Sen ap-proach [20] as:

β = median

[xj − xij − i

], ∀i < j (6)

38

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


Bibliography

[1] J.C. Adam and D.P. Lettenmaier. Adjust-ment of global gridded precipitation forsystematic bias. Journal of Geophysical Re-search, 108(D9), 2003.

[2] O. Bellprat, S. Kotlarski, D. Luthi, andC. Schar. Objective calibration of regionalclimate models. Journal of Geophysical Re-search, 117:D23, 2012.

[3] L. Bierdel, P. Friederichs, and S. Bentzien.Spatial kinetic energy spectra in theconvection-permitting limited-area NWPmodel COSMO-DE. MeteorologischeZeitschrift, 21(3):245–258, 2012.

[4] F. Boberg and J.H. Christensen. Overesti-mation of mediterranean summer temper-ature projections due to model deficien-cies. Nature Climate Change, 2:433–436,2012.

[5] D. Boldini, L. Comegna, G. Rianna, andP. Tommasi. Evapotranspiration estimatein a clayey slope affected by landslidephenomena. Rivista Italiana di Geotecnica,1:21–33, 2014.

[6] E. Bucchignani, P. Mercogliano, M.P.Manzi, M. Montesarchio, and V. Rillo. As-sessment of ERA-Interim driven simula-tion over Italy with COSMO-CLM. CMCC -Research Paper, RP0183, 2013.

[7] E. Bucchignani, P. Mercogliano, M. Mon-tesarchio, M.P. Manzi, and A.L. Zollo. Per-formance evaluation of COSMO-CLM overItaly and climate projections for the XXIcentury. In Climate change and its implica-tions on ecosystem and society: Proceedings ofI SISC (Societa Italiana di Scienze del Clima)Conference, pages 78–89, 2013.

[8] J.H. Christensen, B. Hewitson, A. Busuioc,A. Chen, X. Gao, I. Held, R. Jones, R.K.Kolli, W.-T. Kwon, R. Laprise, V. Magana Rueda, L. Mearns, C.G. Menendez,

J. Raisanen, A. Rinke, A. Sarr, andP. Whetton. Climate Change 2007: ThePhysical Science Basis. Contribution of Work-ing Group I to the Fourth Assessment Reportof the Intergovernmental Panel on ClimateChange, chapter Regional Climate Projec-tions. Cambridge University Press, Cam-bridge, United Kingdom and New York, NY,USA, 2007.

[9] N. Ciccarelli, J. von Hardenberg,A. Provenzale, C. Ronchi, A. Vargiu,and R. Pelosini. Climate variability innorth-western Italy during the second halfof the 20th century. Global and PlanetaryChange, 63:185–195, 2008.

[10] E. Coppola and F. Giorgi. An assessmentof temperature and precipitation changeprojections over Italy from recent globaland regional climate model simulations.International Journal of Climatology, 30:11–32, 2010.

[11] D.P. Dee, S.M. Uppala, A.J. Simmons,P. Berrisford, P. Poli, S. Kobayashi, U. An-drae, M.A. Balmaseda, G. Balsamo,P. Bauer, P. Bechtold, A.C.M Beljaars,L. van de Berg, J. Bidlot, N. Bormann,C. Delsol, R. Dragani, M. Fuentes, A.J.Geer, L. Haimberger, S.B. Healy, H. Hers-bach, E.V. Holm, L. Isaksen, P. Kallberg,M. Kohler, M. Matricardi, A.P. McNally,B.M. Monge-Sanz, J.-J. Morcrette, B.-K.Park, C. Peubey, P. de Rosnay, C. Tavolato,J.-N. Thepaut, and F. Vitart. The ERA-Interim reanalysis: configuration and per-formance of the data assimilation system.Quarterly Journal of the Royal MeteorologicalSociety, 137, 2011.

[12] U. Ehret, E. Zehe, V. Wulfmeyer,K. Warrach-Sagi, and J. Liebert. Shouldwe apply bias correction to global and re-gional climate model data? Hydrol. EarthSyst. Sci., 16:3391–3404, 2012.

39

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

[13] N. Elguindi, A. Grundstein, S. Bernardes,U. Turuncoglu, and J. Feddema. As-sessment of CMIP5 global model simula-tions and climate change projections forthe 21st century using a modified Thorn-thwaite climate classification. ClimaticChange, 122:523–538, 2014.

[14] P. Frattini, G.B. Crosta, N. Fusi, and P. DalNegro. Shallow landslides in pyroclasticsoils: a distributed modelling approach forhazard assessment. Engineering Geology,73:277–295, 2004.

[15] C. Frei, J.H. Christensen, M. Deque, D. Ja-cob, R.G. Jones, and P.L. Vidale. Dailyprecipitation statistics in regional climatemodels: Evaluation and intercomparisonfor the European Alps. Journal of Geophys-ical Research, 108(D3), 2003.

[16] X. Gao, Y. Shi, R. Song, F. Giorgi, Y. Wang,and D. Zhang. Reduction of future mon-soon precipitation over China: comparisonbetween a high resolution RCM simulationand the driving GCM. Meteorol Atmos Phys,100:73–86, 2008.

[17] F. Giorgi, C. Jones, and G.R. Asrar. Ad-dressing climate information needs at theregional level: the cordex framework.World Meteorological Organization Bulletin,58(3):175–183, 2009.

[18] S. Gualdi, S. Somot, L. Li, V. Artale,M. Adani, A. Bellucci, A. Braun, S. Cal-manti, A. Carillo, A. DellAquila, M. Deque,C. Dubois, A. Elizalde, A. Harzallah, D. Ja-cob, B. L’Heveder, W. May, P. Oddo, P. Ruti,A. Sanna, G. Sannino, E. Scoccimarro,F. Sevault, and A. Navarra. The CIRCEsimulations: Regional climate change pro-jections with realistic representation of theMediterranean Sea. Bulletin of the AmericanMeteorological Society, 94:65–81, 2013.

[19] K. Haslinger, I. Anders, and M. Hofstatter.Regional climate modelling over complex

terrain: an evaluation study of COSMO-CLM hindcast model runs for the GreaterAlpine Region. Climate Dynamics, 40:511–529, 2013.

[20] R.M. Hirsch, J.R. Slack, and R.A. Smith.Techniques of trend analysis for monthlywater quality analysis. Water Resource Re-search, 18(1), 1982.

[21] K.L. Kapper, H. Truhetz, and A. Gobiet.Determination of the effective resolutionof regional climate models by spectral de-composition. Proceedings 11. Osterr. Kli-matag, 2010.

[22] M.G. Kendall. Rank correlation methods.Charles Griffin, 4th edition, 1975.

[23] S. Kotlarski, K. Keuler, O.B. Christensen,A. Colette, M. Deque, A. Gobiet, K. Go-ergen, D. Jacob, D. Luthi, E. van Mei-jgaard, G. Nikulin, C. Schar, C. Teich-mann, V. Vautard, K. Warrach-Sagi, andV. Wulfmeyer. Regional climate modelingon European scales: a joint standard eval-uation of the EURO-CORDEX RCM en-semble. Geoscientific Model DevelopmentDiscussions, 7(1):217–293, 2014.

[24] I. Macadam, A.J. Pitman, P.H. Whetton,and G. Abramowitz. Ranking climate mod-els by performance using actual valuesand anomalies: Implications for climatechange impact assessments. GeophysicalResearch Letters, 37(16), 2010.

[25] H.B. Mann. Non-parametric tests againsttrend. Econometrica, 13:245–259, 1945.

[26] D. Maraun. When will trends in Euro-pean mean and heavy daily precipitationemerge? Environmental Research Letters,8:014004, 2013.

[27] Malte Meinshausen, S.J. Smith, K. Calvin,J.S. Daniel, M.L.T. Kainuma, J-F. Lamar-que, K. Matsumoto, S.A. Montzka, S.C.B.Raper, K. Riahi, A. Thomson, G.J.M.

40

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


Velders, and D.P.P. van Vuuren. The RCPgreenhouse gas concentrations and theirextensions from 1765 to 2300. ClimaticChange, 109:213–214, 2001.

[28] M. Montesarchio, M.P. Manzi, L. Cattaneo,E. Bucchignani, and P. Mercogliano. Per-formance evaluation of a regional climatesimulation with COSMO-CLM in the Alpinespace. CMCC - Research Paper, RP0173,2012.

[29] M. Montesarchio, A.L. Zollo, E. Bucchig-nani, P. Mercogliano, and S. Castellari.Performance evaluation of high-resolutionregional climate simulations in the Alpinespace and analysis of extreme events.Journal of Geophysical Research: Atmo-spheres, 119, 2014.

[30] L. Pagano, L. Picarelli, G. Rianna, andG. Urciuoli. A simple numerical proce-dure for timely prediction of precipitationinduced landslides in unsaturated pyro-clastic soils. Landslides, 7:273–289, 2010.

[31] L. Picarelli, L. Comegna, F. Guzzetti,S.L. Gariano, P. Mercogliano, G. Rianna,M. Santini, and P. Tommasi. Potential cli-mate changes in italy and consequenceson land stability Italian contribution toJTC1 - Technical Report no 3 (TR3) onSlope Safety Preparedness for Effects ofClimate Change. Technical Report draftavailable, 2015.

[32] G. Rianna, P. Tommasi, L. Comegna, andP. Mercogliano. Preliminary assessment ofthe effects of climate change on landslideactivity of Orvieto clayey slope. In Climatechange and its implications on ecosystem andsociety: Proceedings of I SISC (Societa Ital-iana di Scienze del Clima) Conference, pages507–522, 2013.

[33] G. Rianna, A.L. Zollo, P. Tommasi,M. Paciucci, L. Comegna, and P. Mer-cogliano. Evaluation of the effects of cli-

mate changes on landslide activity of Orvi-eto clayey slope. Procedia Earth and Plane-tary Science, 9:54–63, 2014.

[34] E. Roeckner, G. Bauml, L. Bonaven-tura, R. Brokopf, M. Esch, M. Gior-getta, S. Hagemann, I. Kirchner, L. Korn-blueh, E. Manzini, A. Rhodin, U. Schlese,U. Schulzweida, and A. Tompkins. Theatmospheric general circulation modelECHAM5. Part I: Model description. Tech-nical Report 349, Max-Planck-Institut furMeteorologie, Hamburg, Germany, 2003.

[35] E. Scoccimarro, S. Gualdi, A. Bellucci,A. Sanna, P. Fogli, E. Manzini, M. Vichi,P. Oddo, and A. Navarra. Effects of trop-ical cyclones on ocean heat transport ina high resolution coupled General Circu-lation Model. Journal of Climate, 24:4368–4384, 2011.

[36] W.C. Skamarock. Evaluating mesoscaleNWP models using kinetic energy spectra.Monthly Weather Review, 132:3019–3032,2004.

[37] J. Steppeler, G. Doms, U. Schattler, H.W.Bitzer, A. Gassmann, U. Damrath, andG. Gregoric. Meso-gamma scale fore-casts using the non-hydrostatic model LM.Meteorology and Atmospheric Physics, 82(1-4):75–96, 2003.

[38] C. Teutschbein and J. Seibert. Regionalclimate models for hydrological impactstudies at the catchment scale: A reviewof recent modeling strategies. GeographyCompass, 4/7:834–860, 2010.

[39] C. Teutschbein and J. Seibert. Bias cor-rection of regional climate model simula-tions for hydrological climate change im-pact studies: Review and evaluation of dif-ferent methods. Journal of Hydrology, 456-457, 2012.

41

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i

[40] S. Tibaldi, C. Cacciamani, and S. Pecora.Il Po nel clima che cambia. In P. Viaroli,F. Puma, and I. Ferrari, editors, Atti XVIIIcongresso S. It E., volume 24 of Biologiaambientale, pages 21–28, 2010.

[41] P. Tommasi, D. Boldini, G. Caldarini, andN. Coli. Influence of infiltration on the pe-riodic re-activation of slow movements inan overconsolidated clay slope. CanadianGeotechnical Journal, 50(1):54–67, 2013.

[42] P. Tommasi, P. Pellegrini, D. Boldini, andR. Ribacchi. Influence of rainfall regime onhydraulic conditions and movement ratesin the overconsolidated clayey slope ofthe Orvieto hill (central Italy). CanadianGeotechnical Journal, 43:70–86, 2006.

[43] R. Tomozieu, V. Pavan, C. Cacciamani,and M. Amici. Observed temperaturechanges in Emilia-Romagna: mean val-ues and extremes. Climate Research,31:217–225, 2006.

[44] G. Tselioudis, C. Douvis, and C. Zere-fos. Does dynamical downscalingintro-duce novel information in climate modelsimulations of precipitation change overcomplex topography region? NASA Pub-blication, 124, 2011.

[45] R. Vezzoli, M. Del Longo, P. Mercogliano,M. Montesarchio, S. Pecora, F. Tonelli,and A.L. Zollo. Hydrological simulationsdriven by RCM climate scenarios at basinscale in the Po river, Italy. In A. Castel-larin, S. Creola, E. Toth, and A. Monta-nari, editors, Evolving Water Resources Sys-tems: Understanding, Predicting and Manag-ing Water-Society Interactions Proceedings ofICWRS2014, volume 364 of IAHS Red Book,pages 128–133, 2014.

[46] R. Vezzoli, P. Mercogliano, S. Pecora,M. Montesarchio, and C. Cacciamani. Hy-drological modelling of po river (north italy)using the rcm cosmo-clm: validation. Sub-mitted to Climate Research.

[47] R. Vezzoli, P. Mercogliano, S. Pecora,M. Montesarchio, A.L. Zollo, M. Del Longo,and F. Tonelli. Evaluation of climate drivensimulations of Po river flow from 1971 to2000 through flow-duration curve indices:preliminary results. CMCC Research Paper,RP0186, 2013.

[48] D. Zanchettin, P. Traverso, andM. Tomasino. Po river discharge: apreliminary analysis of a 200-year timeseries. Climatic Change, 89:411–433,2008.

[49] A.L. Zollo, P. Mercogliano, M. Turco,R. Vezzoli, G. Rianna, E. Bucchignani,M.P. Manzi, and M. Montesarchio. Archi-tectures and tools to analyse the impactsof climate change on hydrogeological riskon Mediterranean area. CMCC - ResearchPaper, RP0129, 2012.

[50] A.L. Zollo, M. Montesarchio, M.P. Manzi,L. Cattaneo E. Bucchignani, and P. Mer-cogliano. Assessment of COSMO-CLMperformances in simulating the past cli-mate of Italy. CMCC - Research Paper,RP0145, 2012.

[51] A.L. Zollo, G. Rianna, P. Mercogliano,P. Tommasi, and L. Comegna. Valida-tion of a simulation chain to assess cli-mate change impact on precipitation in-duced landslides. In K. Sassa, P. Canuti,and Y. Yin, editors, Landslide Science fora Safer Geoenvironment, volume 1 of Pro-ceedings of World Landslide Forum 3, pages287–292, Beijing, 2014.

[52] A.L. Zollo, V. Rillo, E. Bucchignani,M. Montesarchio, and P. Mercogliano.Temperature and precipitation extremeevents over italy: assessment of high res-olution simulations with COSMO-CLM andfuture scenarios. Submitted to InternationalJournal of Climatology.

42

Cen

tro

Eur

o-M

edite

rran

eosu

iCam

biam

enti

Clim

atic

i


[53] A.L. Zollo, M. Turco, and P. Mercogliano.Assessment of hybrid downscaling tech-niques for precipitation over the Po Riverbasin. In G. Lollino, A. Manconi, J. Clague,W. Shan, and M. Chiarle, editors, Engineer-

ing Geology for Society and Territory, vol-ume 1, pages 193–197. Springer Interna-tional Publishing, 2015. doi:10.1007/978-3-319-09300-0 37.

c© Centro Euro-Mediterraneo sui Cambiamenti Climatici 2014

Visit www.cmcc.it for information on our activities and publications.

The Euro-Mediteranean Centre on Climate Change is a Ltd Company with its registered office andadministration in Lecce and local units in Bologna, Venice, Capua, Sassari, Viterbo, Benevento and Milan.The society doesn’t pursue profitable ends and aims to realize and manage the Centre, its promotion, andresearch coordination and different scientific and applied activities in the field of climate change study.

high resolution climate scenarios on mediterranean test ... detected for statistical downscaling...

Documents