Analysis of a long record of annual maximum rainfall in Athens, Greece, and design rainfall inferences

• Published in: Natural hazards (Dordrecht) 2000-07, Vol.22 (1), p.29-48
• Main Authors: KOUTSOYIANNIS, Demetris, BALOUTSOS, George
• Format: Article
• Language: English
• Subjects: Earth sciences; Earth, ocean, space; Engineering and environment geology. Geothermics; Exact sciences and technology; Extreme weather; Flood protection; Greece, Athens; Hydrology; Hydrology. Hydrogeology; Natural hazards: prediction, damages, etc; Rainfall; Statistical analysis
• ISSN: 0921-030X; 1573-0840
• DOI: 10.1023/a:1008001312219

An annual series of maximum dailyrainfall extending through 1860-1995, i.e., 136 years,was extracted from the archives of a meteorologicalstation in Athens. This is the longest rainfall recordavailable in Greece and its analysis is required forthe prediction of intense rainfall in Athens, wherecurrently major flood protection works are under way.Moreover, the statistical analysis of this long recordcan be useful for investigating more generalisedissues regarding the adequacy of extreme valuedistributions for extreme rainfall analysis and theeffect of sample size on design rainfall inferences.Statistical exploration and tests based on this longrecord indicate no statistically significant climaticchanges in extreme rainfall during the last 136 years.Furthermore, statistical analysis shows that theconventionally employed Extreme Value Type I (EV1 orGumbel) distribution is inappropriate for the examinedrecord (especially in its upper tail), whereas thisdistribution would seem as an appropriate model iffewer years of measurements were available (i.e., partof this sample were used). On the contrary, theGeneral Extreme Value (GEV) distribution appears to besuitable for the examined series and its predictionsfor large return periods agree with the probablemaximum precipitation estimated by the statistical(Hershfield's) method, when the latter is consideredfrom a probabilistic point of view. Thus, the resultsof the analysis of this record agree with a recently(and internationally) expressed scepticism about theEV1 distribution which tends to underestimate thelargest extreme rainfall amounts. It is demonstratedthat the underestimation is quite substantial (e.g.,1 : 2) for large return periods and this fact must beconsidered as a warning against the widespread use ofthe EV1 distribution for rainfall extremes.[PUBLICATION ABSTRACT]