Extremes of the Extremes in rainfall driven floods
Upcoming SlideShare
Loading in...5
×
 

Extremes of the Extremes in rainfall driven floods

on

  • 417 views

 

Statistics

Views

Total Views
417
Views on SlideShare
417
Embed Views
0

Actions

Likes
0
Downloads
6
Comments
0

0 Embeds 0

No embeds

Accessibility

Categories

Upload Details

Uploaded via as Microsoft PowerPoint

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

Extremes of the Extremes in rainfall driven floods Extremes of the Extremes in rainfall driven floods Presentation Transcript

  • www.idrologia.polito.itExtremes of the extremesStill too few to analyse their change?Pierluigi Claps Politecnico di Torino Deciphering River Flood Change Vienna, 3-5 September 2012
  • Hydrologic MonstersBlack Swans: Unexpected phenomena in hydrology RELATED MEETINGS www.idrologia.polito.it 2
  • LITERATURE DISCUSSIONS www.idrologia.polito.it 3
  • my meaning:floods caused bybig rainfall outliersRainfall observations exceedinglydifferent from the rest of the stationrecordValues almost statistically impossibleto observe, in ‘stationary’ terms, in a lifetime (if consideringonly the station record):Hovever….if we are observing them relatively frequently in a region, they should not be ashighly unlikely!Definition of REGION is crucial for a correct transfer of information inspace www.idrologia.polito.it 4
  • Some of the worst Italian Big rainfall Outliers Cetara 1910 It was not an earthquake! www.idrologia.polito.it 5
  • Molare disaster (Ortiglieto dam) 13 August 1935 554 mm in 24 h (>30% MAP) 111 victims www.molare.net www.idrologia.polito.it 6
  • 1954 – Salerno Disaster 25/26 October 1954 Worst catastrophe in Italy (318 victims) due to rainfall-induced floods 504 mm in 24 h (40% of the MAP) www.idrologia.polito.it 7
  • Genova 1970: largest daily rainfall in Italy 7/8 October 1970, 948 mm in 24 h, (90% of the MAP) 43 victims www.idrologia.polito.it 8
  • The november 2011 “black swan”Precipitation maps of the two eventsSource: Italian Civil Protection www.idrologia.polito.it 9
  • 1 Vernazza-Monterosso Oct 25-26 2011Liguria – Cinque Terre Brugnato station • 143 mm in 1 h • 328 mm in 3 hs • 469 mm in 6 hs (Salerno 1954) • 511 mm in 12 hs • 538 mm in 24 hs • 542 mm in 30 hs • Italian Record break13 victims www.idrologia.polito.it 10
  • Monterosso Vernazza http://www.flickr.com/people/chiara-sibona/ www.idrologia.polito.it 11
  • Cinque terre event: How extreme? Pignone: middle-age bridge www.idrologia.polito.it 12
  • 2 Genova - Nov 4 2011 Vicomorasso station • 181 mm in 1 h (Capoterra 2008) • 336 mm in 3 hs (Giffone, 1959 ) • 385 mm in 6 hs • 411 mm in 12 hsRadar reflectivity and Sea surface temperature anomaliesFrom Parodi et al., Eos, Vol. 93, N 6 victims www.idrologia.polito.it 13
  • 180 mm/h =50 m3/s/km2 www.idrologia.polito.it 14
  • Via Fereggiano School exit time12:30:00 12:46:41 12:58:37 www.idrologia.polito.it 15
  • Same synoptyc meteorological configuration for the 2 events• Pronounced moist air advection from the subtropical Mediterranean areas near Africa.• Significant sea surface temperature (SST) anomaly (>1°C).• Strong pressure ridge centered on eastern Europe acting as a block to the motion of the system. www.idrologia.polito.it 16
  • Meteorological “bombs” Quezzi amateur station Genova (11/4/2011) Avg hourly intensity 170 mm/h 460 mm / 6 h Total rainfall (12 h) 556 mm www.idrologia.polito.it 17
  • Social relevance of the 2011 Genoa Flash Flood• Scary event (in an advanced and big city)• Very well forecasted, except exact position, yet producing victims• Miscommunication between Weather Forecast Service and Local Civil Protection• Children and schools involved www.idrologia.polito.it 18
  • Social relevance of the Big rainfall Outliers and their consequent flash floods• Cause more and more fatalities, as compared to lowland floods• Isolated, very fast events, huge unit discharges• Very small basins, interaction with urban infrastructures (often river reach not even present)• Flood hazard evenly distributed in the space• Frequently associated to heavy debris transport (coastal hillslopes) www.idrologia.polito.it 19
  • Questions arising:• “Man in the street” question: why are we still calling them exceptional? How frequent are they REALLY?• Civil Engineer question: How do these ‘new’ observations should impact on design rainfalls?• Scientist questions: Are STAT toolkits for frequency assessment adequate? Will the frequency increase in the near future? www.idrologia.polito.it 20
  • Regional or At-Site estimation of T(h)? 2011 Kysely et al. (2011) and many other conclude with: REGIONAL! All regional models (ROI) agree on return periods in the order of several hundreds of years; although “there are no means of validating the estimates” www.idrologia.polito.it 21
  • However…. Traditional regional methods must face the effects of variable record lengths www.idrologia.polito.it 22
  • Another problem (specially in Italy): station birth/death ARPA (1988-2006) SIMN (1928-1996) active rain gauges over Piemonte - Valle d’Aosta 1928-2006 www.idrologia.polito.it 23
  • Spatial correlation of eventsROI or and other regional methods cannot account if similar events occur in differentyears or in the same year.(If similar events occur in different years the individual hazard is higher) De Luca et al, 2010 www.idrologia.polito.it 24
  • Why not At-Site estimation of T(h)?At site estimation dependswhether or not the lastobservation is includedin the sample(*)(T=283 vs. T=45110) (*) Answer: the outlier can be included in the sample if the Maximim-Value test is passed www.idrologia.polito.it 25
  • Maximum-Value Test Grubbs (1969) Rossi et al. (1984)Test question:Can the maximum value X(n) (of a sample of lenght n) considered asextracted from the proposed parent distribution FX(x) , with asignificance level a ? 0.999 1000 EV1 Cumulative Probability 0.995 200 0.99 100 0.95 20 0.90 10 0.80 5 0.70 0.60 0.50 2 0.40 0.30 0.20 0.10 0.05 0.02 0.01 [T] 0 500 1000 1500 2000 2500 3000 Annual maximum flood peaks [m3/s] www.idrologia.polito.it 26
  • Test AnswerYES, iféFX ( X(n) | Q)ù <1- a në ûwhere X(n) is the sample maximum value, Q is the estimatedparameter set and a is the significance level of the test. However, Q should not depend on X(n) itself! www.idrologia.polito.it 27
  • Modified maximum value Test Laio, Allamano, Claps, HESS, 2010Suggested solution: observed X(n) can be substituted by the medianof the distribution of the n-year maxima. In distributions characterized by position, scale and shape parameters: z(u,θ3) is the quantile function of the standardized variable z=(x−θ1)/θ2, which only depends on the probability level u and on the shape parameter θ3 THE USE OF L-MOMENTS ESTIMATED MAKES THE TEST EXPLICIT
  • To get the best from local information, despite data patchiness…. www.idrologia.polito.it 29
  • Year-after-Year Kriging Allamano et al., (2011). GEOPHYSICAL RESEARCH ABSTRACTS,Spatial variability of extreme precipitation in each year, using the available data1930 … 1950 … 1970 … 1990 … 2004… At-a-point Time series of the annual maximum precipitation for d=1h
  • L-moments computed on reconstructed data “local” T(h) estimation and verification possible L-CA (h1) a)Average h1 L-CV (h1) Respective of the spatial correlation of events
  • What about h(T) in ungauged sites?• “Man in the street” question: why are we still calling them exceptional? How frequent are they REALLY?• Civil Engineer question: How do these ‘new’ observations should impact on design rainfalls?• Scientist questions: Are STAT toolkits for frequency assessment adequate? Will the frequency increase in the near future? www.idrologia.polito.it 32
  • h(T) as a “man in the street” question:Given a new ‘VERY UNUSUAL’ event in a station, shouldn’t itaffect the rainfall hazard in the surroundings? Piemonte, h12, 1935 Limited response with the individual-year data reconstruction www.idrologia.polito.it 33
  • Probability distributions for h(T): mixed?
  • About the Physically-based grounds of the TCEVMeteorologists and Hydrologistscollaborations www.idrologia.polito.it 35
  • Meteorological basis for classification of events Towards a physical-meteorological identification of more hazardous areas www.idrologia.polito.it 36
  • Why Italy in general … Kouroutzoglou et al., Int. J. Climatol. (2011)… and Mar Ligure in particular average cyclogenesis of all cyclones in the Mediterranean www.idrologia.polito.it 37
  • How to identify areas with “similar” Big Rainstorm distribution? • Accurate selection of event occurrences • More effort to select and verify very old events • Interaction with meteorologists as regards statistical analysis www.idrologia.polito.it 38
  • European-wide collaboration www.idrologia.polito.it 39
  • Issues related to changeARCHIVAL SOURCES >300 mm in 24 h Archivio S.I.C.I.Annuario dati Ambientali Archivio S.C.I.A. Progetto AnnaliRapporti evento A.R.P.A. Pubblicazioni locali Siti Web amatoriali Archivio “La Stampa” Macchia, 2011 www.idrologia.polito.it 40
  • 0 1 2 3 4 5 6 7 8 1923 1925 1927 1929 1931 1933 1935 1937 1939 1941 1943 1945 1947 1949 1951 1953 1955 1957 1959 1961 1963 1965 1967 1969 1971 1973 1975 1977 1979www.idrologia.polito.it 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 Number of observing stations 2003 2005 2007 2009 201141 stations (ths) Events in Italy
  • Explosive cyclones in the Mediterranean (1962-2001) NO TREND SO FAR!www.idrologia.polito.it 42