Detection of Human Influence on Trends of North Atlantic Ocean Wave Heights and Atmospheric Storminess

Transcription

1 Detection of Human Influence on Trends of North Atlantic Ocean Wave Heights and Atmospheric Storminess Xiaolan L. Wang, Val R. Swail, Xuebin Zhang, Francis W. Zwiers Climate Research Division, Environment Canada Myles Allen Department of Physics, University of Oxford

2 Outline Data sets Methodology Results Concluding remarks

3 Data sets Seasonal means and maxima of SWH original ERA40 wave data ( observed wave heights) P t anomalies of seasonal mean SLP G t seasonal anomalies of squared SLP gradients (geostraphic wind energy) P t and G t SLP of ERA40 reanalysis for P t and G t SLP of NCEP/NCAR reanalysis for observed atmosphere (anomalies relative to the climate) Simulated P t and G t 9 ensembles of integrations of human-induced climate change (45 in total)

4 Table 1. The nine coupled ocean-atmosphere models used in this study and the number of integrations (runs) conducted with each of these models. Model ECHO-G GFDL CM2.0 GFDL CM2.1 GISS Model E H GISS Model E R MIROC 2 medres MRI CGCM2 3 2a NCAR CCSM 3 NCAR PCM 1 Number of runs Total: 45 Period of integration: (used: : 2x42 yr.) All with historical greenhouse gases and sulphate aerosols forcing Obtained from the IPCC AR4 models output archive

5 Seasonal 20-yr return values H 20 :y Methodology Wave height not directly available Statistical simulations of SWH: Parameters aˆ, bˆ, cˆ, σˆ, ξˆ ERA40 SWH, P t and G t (de-trended) Seasonal means : Ĥ avg H ˆ = aˆ + bp ˆ + cˆ G t ˆ t t Climate model simulations corresponding SWH change GEV ( ˆ = a ˆ + bp ˆ + c ˆ G, ˆ, σ ˆ) ξ µ t t t Wave height not directly available from NNR NNR, ERA40 hindcasts Linear trend patterns of P G, H, H, Hˆ, Hˆ : T, T Multi-model mean ( ) t, t avg 20 y avg 20 y o m Observed /hindcast Optimal detection approach (TLS fit): T o = β ( T m ς ) + η If scaling factor β> 0 significantly, Simulated Observed response to anthropogenic forcing is detected internal variability η~ζ consistent? using simulations for

6 Trend patterns of seasonal mean SLP anomalies ERA40 - JFM ERA40 - OND P t Simulated JFM Simulated OND Too small: NNR - JFM NNR - OND

7 Trend patterns of seasonal anomalies of squared SLP gradients ERA40 - JFM ERA40 - OND G t Too small: Simulated JFM Simulated OND NNR - JFM NNR - OND Geostraphic wind energy atmospheric storminess

8 Trend patterns of seasonal mean SWH H avg ERA40orig - JFM ERA40orig - OND Simulated JFM Simulated OND Too small: NNR hindcast - JFM NNR hindcast - OND

9 Trend patterns of seasonal 20-yr return values of SWH ERA40 orig - JFM ERA40 orig - OND H 20 y Simulated JFM Simulated OND Too small: NNR hindcast - JFM NNR hindcast - OND

10 35 H avg H 20y JFM Pt or Havg Gt or H20y βˆ Scaling factor P t G t Ĥ avg Hˆ 20y P t G t Ĥ avg Hˆ 20y The estimated scaling factor & its 5-95% uncertainty range ( ˆ ˆ ) βˆ β, β Response detected because all ˆ > l β l h 0! ERA40 original ERA40-SWH ERA-SWH NNR Observational/hindcast data set Magnitude of response underestimated for the cases with ˆ > 1. β l ζ climate models underestimate the internal var. Conservative conclusions NNR-SWH βˆ Scaling factor P t ζ Gt ζ ERA40 H avg H 20y original ERA40-SWH Ĥ avg OND ζ Hˆ 20y ERA-SWH P t NNR Observational/hindcast data set ζ Gt Pt or Havg Gt or H20y Ĥ avg ζ Hˆ 20y NNR-SWH

11 Concluding remarks In the cold seasons: The observed trend patterns of SWH and storminess: in high-latitudes, to the south ERA40orig - JFM ERA40 - JFM storminess H avg : G t : ERA40 orig - JFM H 20 y :

12 Concluding remarks In the cold seasons: The observed trend patterns of SWH and storminess: in high-latitudes, to the south Anthropogenic forcing has had a detectable influence on the trends (all observed datasets used) The climate models significantly underestimate the magnitude of the response to anthropogenic forcing The climate models also underestimate the observed internal variability in fall ( conservative detection conclusions)

13 Acknowledgement The authors are grateful to Mr. Yang Feng for his great computing support, and to Dr. Jiafeng Wang for his help in compiling the climate model outputs - The End - Thank you very much!