Dynamical versus Statistical Projections of Ocean Wave Heights

Transcription

1 Dynamical versus Statistical Projections of Ocean Wave Heights Xiaolan L. Wang and Val R. Swail Climate Research Division, Science and Technology Branch, Environment nment Canada Andrew Cox OceanWeather Inc., USA Datasets and Methodologies Comparison - wave height climate and variability Comparison - the projected possible future changes Summary

2 Datasets and methodologies Our previous studies good relationships between SWH & atmospheric indices: ˆ ( t) = aˆ + bg ˆ + cˆ H avg t P t seasonal mean SWH Hˆ ( t) GEV ( ˆ μ = ˆ μ + ˆ ρ G + ˆ ρ P, ˆ σ, ˆ ξ 20 y t 0 1 t 2 t ) seasonal extreme SWH Predictors P t : Seasonal mean SLP anomalies G t : Seasonal anomalies of squared SLP gradient ERA40 SLP for (96x48 Gaussian grid) Observed predictors ERA40 waves (seasonal mean or max SWH) for (1.5x1.5 lat/long grid) Observed predictands Statistical projections of H avg (t) and H 20y (t) Projected predictors P t and G t Projections of SLP (96x48 Gaussian grid) CGCM2, IS92a scenario, 3 runs: each for 3x20-yr: , , Projections of surface winds ODGP-2G Dynamical projections of H avg (t), H max (t) 6-hourly wave height fields

3 Datasets and methodologies (cont d) not H max (t) Statistical projections of H avg (t) and H 20y (t) Projected P t and G t Projected surface winds ODGP-2G Dynamical projections of H avg (t), H max (t) ERA40 & dynamically projected H max(t) Compare long-term means ( t) GEV ( ˆ μ = ˆ μ + ˆ βt + ˆ λt ˆ ξ ˆ tr 20 2 y ˆ t 0, σ, ) H ˆ H20 tr y ( t) = hˆ + ˆ β t + ˆ λ 0 t 2 statistically projected H 20y (t) Will try this: 12-hly ERA40 SWH and SLP data ˆ ˆ + Statistical projections of 12-hly SWH: Ht = aˆ + bgt cˆ Pt Projected 12-hly P t & G t Statistical projections of H max (t) ~ ERA40 & Dynamical projections of H max (t)

4 Comparison of the wave height climate and variability

5 The climate of seasonal mean SWH (cm) Dyn. - JFM Dyn. - OND Dynamical overestimates better reproduce the location of the center of high waves ERA40 - JFM ERA40 - OND ERA40: Statistical Stat. - JFM Stat. - OND

6 Differences in the climate of seasonal mean SWH (cm) JFM OND Dynamical projections minus ERA40: 1 m JFM OND Statistical projections minus ERA40: 1 m Pattern correlation skill score: Pattern correlation Pattern correlation of JFM seasonal mean SWH Dyn. Stat Pattern correlation Pattern correlation of OND seasonal mean SWH Dyn. Stat Time Time

7 The std of seasonal mean SWH (cm) Dyn. - JFM Dyn. - OND Dynamical overestimates ERA40 - JFM ERA40 - OND ERA40: Statistical Stat. - JFM Stat. - OND

8 Ratios of the variance of seasonal mean SWH (cm) JFM OND Dynamical projections ~ ERA40: 4 times JFM OND Statistical projections ~ ERA40:

9 The climate of seasonal H 20y (cm) Dyn. - JFM Dyn. - OND Dynamical ERA40 - JFM ERA40 - OND ERA40: better reproduce the observed patterns Statistical Stat. - JFM Stat. - OND

10 Differences in the climate of seasonal H 20y (cm) JFM OND Dynamical projections minus ERA40 Smaller differences in the climate of seasonal H max JFM OND Statistical projections minus ERA40

11 The climate of seasonal H max (cm) Dynamical Dyn. - JFM Dyn. - OND Patterns similar to those of the corresponding H 20y little distortion ERA40 - JFM ERA40 - OND ERA40: Dynamical projections minus ERA40: Dyn. ERA40 - JFM Dyn. ERA40 - OND

12 Comparison of the projected possible future changes

13 Projected changes in seasonal H avg (cm) ( s s) Dyn. - JFM Dyn. - OND Dynamical larger larger Patterns share substantial similarity Stat. - JFM Stat. - OND Statistical larger

14 Projected changes in seasonal H 20y (cm) ( s s) Dyn. - JFM Dyn. - OND Dynamical larger Stat. - JFM Stat. - OND Statistical larger larger

15 Summary 1. Projected mean SWH climate ~ ERA40: > Dynamical projections better location of climatological center of high waves > mid-high latitude NA: dynamical ~ overestimate; statistical ~ underestimate > subtropical NA: dynamical ~ underestimate, statistical ~ very good Statistical projections ~ higher pattern correlation skill 2. Projected variability (including trends) ~ ERA40: > Statistical projections ~ underestimate everywhere but central-eastern NA in both seasons > Dynamical projections ~ notably overestimate in central-eastern NA in winter > 60N-70N, North Sea, & Labrador Sea: dynamical ~ overestimate, in both seasons statistical ~ underestimate,

16 3. Projected extreme SWH climate ~ ERA40: Summary (cont d) > dynamical ~ underestimate winter 20-yr return values everywhere ~ overestimate fall 20-yr return values in southwestern NA and underestimate them elsewhere > statistical projections ~ very good in both seasons 4. Projected possible future changes: > dynamical & statistical projections show patterns of change of substantial similarity: in the mid-latitude NA, in the North Sea in JFM & in the area west of Ireland in OND dynamical ~ larger in mid-latitude NA; statistical ~ larger in southwestern NA in JFM > 60N-70N: dynamical & statistical projections show changes of the opposite signs, especially in OND uncertainty due to different downscaling methods or predictors used Dynamical ~ winds less confident Statistical ~ SLP

17 Acknowledgement The authors are thankful to Mr. Yang Feng for his great computing support. - The End - Thank you very much!