Page - 69 - in Loss and Damage from Climate Change - Concepts, Methods and Policy Options

3 ObservedandProjected Impacts fromExtremeWeatherEvents… 69 For rainfall related extremes, including droughts and river flooding, findings regarding the detection of changes is more mixed, as is the attribution of these changes to human greenhouse gas emissions. Forwindstorms, in the tropical and extra-tropical regions,both thechangesand theprecisehumancontribution to these changes are evenmore uncertain. In addition, there is an extensive literature that has looked at how the likelihood of individual extremeweather events has possi- bly changed due to anthropogenic forcing (see James et al. 2018). In addition to monotonic changes from anthropogenic forcing, the role of natural variability in shaping the impacts fromnatural disasters can be very large. This is an important reasonwhy evenwhen trends in extremes are found, related to large decadal vari- ability in theoccurrenceof extremes related tonatural variability, the attributionof smaller changesover time toanthropogenic emissions.This is for instance thecase fortropicalcyclones,wherelargenaturalvariabilitycomplicatesthedetectionofany remaining trend (Knutsonet al. 2010). 3.3 ObservedImpactsBasedonDisasterLossRecords 3.3.1 LossDataandNormalisation Several records are available of disaster losses. Themost notable global databases consist of those managed by CRED (EM-DAT database,1 Munich Reinsurance Company(NatCatSERVICEdatabase),2 andSwissReinsuranceCompany(SIGMA database).3 Besides these global databases, several combined are available under Desinventar.4 While these databases provide a good overall understanding of loss frequency and trends, several other records of natural hazard impacts exist that are moredetailed, includingnational accounts of disaster losses andnational and local insurance records.Someof thesearealsoassessed in the studies reportedhere. Several researchers have analysed disaster loss records, to assess the frequency and size of impacts from these hazards. In addition, many have analysed which drivers(hazard,exposure,orvulnerability)mayhaveledtochangesintheseimpacts over time.Anoften-usedapproachisso-callednormalisation,whichtries toaccount for changes in exposure over time, by applying correction factors to the observed loss record. These factors are based on the total size of the exposed assets and their value (see Pielke and Landsea 1998). This is also common practise in the insuranceindustryinordertoarriveatacommonreferencebaselineofhistorical loss events that can be compared to catastrophemodels that simulate risks for today’s exposureandvulnerabilitiesorforaspecificbaselineyear(Pielkeetal.1999).Many 1http://www.emdat.be. 2https://www.munichre.com/en/reinsurance/business/non-life/natcatservice/index.html. 3http://institute.swissre.com/research/overview/sigma_data/. 4https://www.desinventar.org/ andhttp://www.desinventar.net/.