Indices versus Data
27
Indices versus Data Indices versus Data Indices are information derived Indices are information derived from data from data Proxy for data Proxy for data More readily released than data More readily released than data • Indices generally not of economic Indices generally not of economic value – only research value and for value – only research value and for applications applications • More on this later More on this later Are not reproducible without the Are not reproducible without the data data • A key component of science A key component of science
description
Indices versus Data. Indices are information derived from data Proxy for data More readily released than data Indices generally not of economic value – only research value and for applications More on this later Are not reproducible without the data A key component of science. - PowerPoint PPT Presentation
Transcript of Indices versus Data
Indices versus DataIndices versus Data
Indices are information derived from Indices are information derived from datadata
Proxy for dataProxy for data More readily released than dataMore readily released than data
• Indices generally not of economic value Indices generally not of economic value – only research value and for – only research value and for applicationsapplications
• More on this laterMore on this later Are not reproducible without the dataAre not reproducible without the data
• A key component of scienceA key component of science
Many different ways to Many different ways to calculate indices for calculate indices for
extremesextremes
What types of extremes?What types of extremes? Trends in extreme events characterised by Trends in extreme events characterised by
the size of their societal or economic impactsthe size of their societal or economic impacts
Trends in “very rare” extreme events Trends in “very rare” extreme events analysed by the parameters of extreme value analysed by the parameters of extreme value distributionsdistributions
Trends in observational series of phenomena Trends in observational series of phenomena with a daily time scale and typical return with a daily time scale and typical return period < 1 yearperiod < 1 year(as indicators of extremes)(as indicators of extremes)
NO
NO
YES
Motivation for choice of Motivation for choice of “extremes”“extremes”
The detection probability of trends The detection probability of trends depends on the return period of the depends on the return period of the extreme event and the length of the extreme event and the length of the observational seriesobservational series
For extremes in daily series with For extremes in daily series with typical length ~50 yrs, the optimal typical length ~50 yrs, the optimal return period is 10-30 return period is 10-30 daysdays rather rather than 10-30 than 10-30 yearsyears
ApproachApproach
Focus on counts of days crossing a Focus on counts of days crossing a threshold; either absolute/fixed threshold; either absolute/fixed thresholds or percentile/variable thresholds or percentile/variable thresholds relative to local climatethresholds relative to local climate
Standardisation enables comparisons Standardisation enables comparisons between results obtained in different between results obtained in different countries, and even different parts of countries, and even different parts of the worldthe world
Expert Team on Climate Change Detection and Indices (ETCCDI)
started in 1999
jointly sponsored by CCl, CLIVAR and JCOMM
the ETCCDI developed an internationally coordinated set of 27 climate indices
focus on counts of days crossing a threshold; either absolute/fixed thresholds or percentile/variable thresholds relative to local climate
used for both observations and models, globally as well as regionally
can be coupled with – simple trend analysis techniques– standard detection and attribution methods
complements the analysis of more rare extremes using EVT
Klein Tank, Zwiers and Zhang, 2009, WCDMP-No. 72,WMO-TD No. 1500, 56pp.
Example: Russian heat wave, July 2010
In-situNOAA
ERA-InterimECMWF
MSUUAH
Courtesy: John Christy (top), Adrian Simmons (bottom)
In-situNOAA
Example: Russian heat wave, July 2010
ETCCDI indices add relevant information
In-situNOAA
Example: Russian heat wave, July 2010
31 days withT-max > 25°C against 9.5 days in a normal July
Example: Russian heat wave, July 2010
16 nights with T-min > 20°C against 0.5 night in a normal July
Example: Russian heat wave, July 2010
Extremes Indices
upper 10-ptile 1961-1990
the year 1996
lower 10-ptile1961-1990
Indices example
upper 10-ptile 1961-1990
the year 1996
lower 10-ptile1961-1990
“cold nights”
Indices example
upper 10-ptile 1961-1990
the year 1996
lower 10-ptile1961-1990
“warm nights”
“cold nights”
Indices example
De Bilt, the Netherlands
Indices example
1) Identify heavy falls using a site specific threshold = 95th percentile at wet days in the 1961-90 period
Changes in heavy falls
1) Identify heavy falls using a site specific threshold = 95th percentile at wet days in the 1961-90 period
2) Determine fraction of total precipitation in each year that is due to these days
Changes in heavy falls
1) Identify heavy falls using a site specific threshold = 95th percentile at wet days in the 1961-90 period
2) Determine fraction of total precipitation in each year that is due to these days
3) Trend analysis in series of fractions
Changes in heavy falls
Alexander et al.,2006; in IPCC-AR4
Changes in heavy falls
Extremes table IPCC-AR4, WG1 report (IPCC, 2007)
