Page - 267 - in Loss and Damage from Climate Change - Concepts, Methods and Policy Options

11 TheRoleof thePhysicalSciences inLossandDamage… 267 (simulating featureswith scales as small as a fewkilometres).Downscaling canbe accomplished throughoneof two techniques: ‘dynamical’or ‘statistical’downscal- ing (Wilby et al. 2009). ‘Dynamical’ downscaling refers to the process of nesting high resolutionRegionalClimateModels (RCMs)withinaglobalGCM(Hewitson et al. 2014;Giorgi et al. 2015)while ‘statistical’ downscaling relies on using sta- tistical relationshipsbetweenlarge-scaleatmosphericvariablesandregionalclimate (oftenatmeteorologicalstationlevel) togenerateprojectionsoffuture localclimatic conditions. Statistical methodsmay also includeweather generators that simulate weathereventsandtheirextremes.Downscalingapproachesdonotprovidemagical fixestopossiblelimitationsinthedatabeingdownscaled(Kerr2011).Incaseswhere thelargescaleGCMsignalaccuratelyrepresents theobservedone,downscalingcan add value by incorporating features that are absent inGCMs, such as the effect of coastlines and complexorography (Hall 2014).However,when for instancediffer- entRCMsdrivenbythesameGCMshowawiderangeof responses inprecipitation (Hewitsonetal.2014), thegenerationofclimateprojectionsusingdownscalingtech- niqueswill often increase the level of uncertainty in theoriginalGCMprojections, havingsignificanteffectsintheestimationofprobabilitiesofoccurrenceofdamaging events inDRRmodels andclimatechange riskassessments. Climatemodel projections (and their downscaledversions) provide information about climate variables such as temperature, precipitation, sea level, etc. The next step inaclimate riskassessment involvesunderstandinghowchanges in theclimate variables will affect natural or human systems. Hazardmodels are computational models that takeas inputsobservedor simulatedclimatevariables suchas tempera- ture,precipitation, soilmoisturecontent,windspeed, etc., anduse themtosimulate thevariablesthatarerelevanttoanalyseaparticularweatherorclimatehazard(IPCC 2012, 2014a, b). For instance, extreme rainfall events cancausefloods.But to esti- mate the extent of theflooded area, hydrological andhydraulicmodels are used to generate theflood footprint for eachparticular event (Ranger et al. 2011; Jha et al. 2012).Someofthelimitationsofhazardmodelsaresimilar to thoseofclimatemod- els: poor representation of the physical processes involved, calibration issues and computationalconstraintsallcontribute tocompounding theuncertainties in thecli- mate inputswith the uncertainties in the hazardmodel outputs. This is illustrated, for example, bymulti-model assessments ofwater availability andfloodpotential, where a large ensembleof global hydrologicalmodels is forcedbyanensembleof GCMs to estimate climate change impacts onwater resources. These studies show that climate andhydrologicalmodels contribute to a similar extent to the spread in relative riverflows’changesglobally (Scheweet al. 2014;Dankerset al. 2014). Analternativeapproach toestimate thephysical impactsof climatechangeused whenmodel projections are not available, is the use of ‘analogies.’ Two types of analogies are possible: spatial analogieswhereby another part of theworld experi- encingsimilar conditions to thoseexpected tooccur in the future isusedasaproxy toestimate future impacts in the regionof interest; and temporal analogieswhereby changesinthepast (sometimesobtainedfrompaleo-records)areusedtomakeinfer- encesaboutchanges in the future.Thisapproachhas two limitations.Firstly, expert