David Karoly, Frank Drost and Ailie Gallant University of ... - IMAGe

David Karoly, Frank Drost and Ailie GallantUniversity of Melbourne, AustraliaUpdate of Karoly and Braganza, 2001; Braganza et al., 2003, 04Climate extremes index - Gallant and Karoly, JClim, 2010

Karoly & Braganza simple indices• Would like a small number of indices of globalsurface temperature variations that representdifferent aspects of natural climate variability butrepresent key features of patterns of anthropogenicclimate change• Want indices that show a common signal due togreenhouse climate change but are nearlyindependent for natural climate variations– Global mean surface temperature (GM)– Mean land – ocean temp difference (LO)– Mean NH -– SH temp difference (NS)– Mean magnitude of the annual cycle over land (AC)– Mean meridional temperature gradient in the NH (MTG)

Data• Gridded instrumental observations for surface airtemperature and SST for 1881-2009 fromHadCRUT3 (Met Off, UK), GHCN-ERSST (NOAA,US), and GISS Temp (NASA, US)• Coupled ocean-atmosphere climate modelsimulations, 8 different models– unforced simulations of internal climate variability (longcontrol runs, typically 500 years)– multi-member ensembles of simulations of the 20 thcentury, including variations in anthropogenic forcings(greenhouse gases and aerosols) and natural forcings(volcanic aerosols and solar irradiance changes), 30runs

Common responseSimple indices from CMIP3model simulations for C20Cand A1B emissions.Dark grey shading is ±1σabout ensemble mean

Climate variability• Compare internal variabilityin control simulations withdetrended observations• Interannual and decadalvariations• Some models (1,5, 8) havetoo small variability inseveral indices• Decadal variability of MTG istoo large in several models

Correlations of decadal variations of indices fromdetrended obs and 120 yr segments from control runs

Climate change• Trends over 1960-2009• Grey band is 90% confidinterval from control runs• Significant trends in all obsindices, except NS• General agreement of obsand model trends except formodel 5• Generally too small forcedresponse in models for AC

RLO, ratio of warming on land to ocean• RLO = Δ(land T relative to 1900-30 av) / Δ SST• Upper panel for obs, note smaller values for El Nino• Smaller ocean warming due to heat capacity andevaporationRLO av for 1990-2009HadCRUT3 = 1.3GISS = 2.0NOAA = 1.7Models = 1.5

Climate Extremes Index, CEI• Karl et al., 1996, Gleason et al, 2008 http://www.ncdc.noaa.gov/extremes/cei/• Fraction of area of country experiencing muchabove/below normal:– Mean maximum temperature– Mean minimum temperature– Soil moisture– Proportion of rain from very heavy rain days,– Number of rain days• Multivariate, non-parametric• Current use adds contribution from extremes below10th %ile and above 90th %ile, which is a problem ifthere is a trend

Mean Tmax > 90%ileHeavy 1-day precipMean Tmax < 10%ileMean Tmax, combinedFromNOAACEIwebsiteCEI, 5 comps

Data from NOAA CEI websiteComponent 1 Extreme mean max temperatureComponent 3 Extreme mean moisture (PDSI)Percentage area30%90th %ileCombined10%0%1910 1930 1950 1970 1990 2010-10%-20%-30%Percentage area20%15%10%5%0%1910 1930 1950 1970 1990 2010-5%-10%-15%90th %ileCombined-25%Component 2 Extreme mean min temperatureComponent 5 Extreme number rain daysPercentage area40%90th %ileCombined20%10%0%1910 1930 1950 1970 1990 2010-10%-20%-30%Mean Tmax < 10%ilePercentage area20%15%10%5%0%1910 1930 1950 1970 1990 2010-5%-10%-15%90th %ileCombined-25%