past and future changes in extratropical storms...graham&diaz 6116 3.40 -0.24 0.93 0.26 0.79...

Past and future changes in winds and waves in extratropical stormsSofia Caires Deltares | Delft Hydraulics, The Netherlands

Val Swail and Xiaolan Wang Climate Research Division, Environment Canada

and many others at Oceanweather, KNMI, etc.

27-29 May 2008OGP/JCOMM/WCRP workshop 2

Outline

1. Quality of available present climate datasets 2. Past/present changes3. Future changes4. Concluding remarks 5. Recommendations


The study of wave climatology and climate variability requires good quality data with a reasonable time and space resolution and extent.

ObservationsTime and space coverage

Compilation of analysed winds/wavesAnalysis techniqueQuality coverage and resolution of the

observed data usedReanalysis

Quality coverage and resolution of the observed data used

How to get such data?

The need for reanalysis data

sudden and creeping inhomogeneities remain a problem!


Reanalysis data quality issues

Wind Speed• ERA-40, 1.5x1.5, global • NCEP/NCAR, 1.25x2.5 , global• Swail and Cox (2000), kinematically improved winds, 0.625x0.833, NA• (Swail et al (2006), further improvements (TC) .5x.5 (.1x.1), NA)

Significant wave height• ERA-40,couple WAM, 1.5x1.5, global• Cox and Swail (2001) , NCEP/NCAR winds, ODGP2 wave model,

1.25x2.5, global • Swail and Cox (2000), OWI 3-G wave model, 0.625x0.833, NA• (Swail et al (2006), OWI 3-G +finer shallow grid, .5x.5 (.1x.1), NA)

Considered datasets

Due to resolution, are the extratropical storms considered here mostly extratropical cyclones


Region Reanalysis n x Bias RMSE SI ρ ERA-40 4769 8.12 -0.72 1.64 0.18 0.92

NCEP/NCAR 4707 8.12 -0.40 1.73 0.21 0.89 20ºN-80ºN

Swail&Cox 1810 8.07 -0.37 1.50 0.18 0.92 ERA-40 7326 6.60 -0.39 1.35 0.20 0.86

NCEP/NCAR 7142 6.62 -0.45 1.77 0.26 0.76 20ºS-20ºN

Swail&Cox 767 6.42 -0.29 1.51 0.23 0.80 ERA-40 11468 9.66 -0.89 1.79 0.16 0.91 80ºS-20ºS

NCEP/NCAR 11427 9.69 -0.86 2.33 0.22 0.82

Region Reanalysis n x Bias RMSE SI ρ ERA-40 4769 2.66 -0.22 0.45 0.15 0.96

Cox&Swail 4707 2.66 -0.01 0.59 0.22 0.91 Swail&Cox 1810 2.54 -0.04 0.40 0.16 0.95

20ºN-80ºN

Graham&Diaz 2517 2.78 -0.23 0.76 0.26 0.89 ERA-40 7326 2.06 -0.06 0.24 0.11 0.93

Cox&Swail 7142 2.07 -0.12 0.40 0.18 0.80 Swail&Cox 767 1.81 0.09 0.29 0.15 0.85

20ºS-20ºN

Graham&Diaz 4041 2.17 -0.26 0.50 0.20 0.82 ERA-40 11468 3.41 -0.25 0.47 0.12 0.95

Cox&Swail 11427 3.42 0.00 0.72 0.21 0.86 80ºS-20ºS

Graham&Diaz 6116 3.40 -0.24 0.93 0.26 0.79

Wind speed (m/s)

Significant wave height (m)

Validation using altimeter data


Monthly means

Comparison of wind speed monthly means


Comparison of significant wave height monthly means


8-year trends

Comparison of wind speed trends


8-year trends

Comparison of significant wave height trends


Num

ber o

f grid

poin

ts

Year of changepoint

Monthly mean wind speedHistogram of WsMean

Fre

quen

cy

1960 1970 1980 1990 2000

010

2030

40

1966

Histogram of HsMean

Fre

quen

cy

1960 1970 1980 1990 2000

050

100

150

Monthly mean sig. wave height

Num

ber o

f grid

poin

ts

Year of changepoint

1966

Number of gridpoints of a significant changepoint in the indicated yearWind speed – locations of changepoint in Nov. 1966 Sig. wave height – location of changepoint in Nov. 1966

Grid-boxes of significant changepoint are shown in black

xx

Problems with inhomogeneities (MSC50)


Monthly mean wind speed (m/s) at (40.5°N, 40.5°W)

Nov. 1966 Dec. 1997

Monthly mean sig. wave height (m) at (40.5°N, 40.5°W)

Nov. 1966 Dec. 1997

See: http://cccma.seos.uvic.ca/ETCCDMI/software.shtml

Data homogenization (MSC50)


Remarks on the quality of different reanalysis

• Swail and Cox (2000) has the best statistics. ERA-40 compare well with the observations, and have in general better statistics than the other reanalyses results, especially as regards the significant wave height (Swail et al (2006) even better).

• The wind speed comparisons show large differences in the tropics and differences usually larger in the Southern than in the Northern Hemisphere, testifying to the present limitations of modelling those regions.

• At a synoptic time scale the differences between the various reanalysis winds and waves are large.

• In terms of monthly means the differences in wind fields of Swail and Cox (2000) and Cox and Swail (2001) are almost nowhere significant and the ERA-40 monthly means differ from those datasets mainly south of 30°

N. The various significant wave height datasets differ at monthly mean time scales.

• The longer term behaviour of both winds and waves in the various datasets analysed is however quite similar, an indication that the NH large time scale features are equally present in all datasets.

• Many inhomogeneities still remain and the detection/removal of creeping inhomogeneities is a challenge!


Description of the past/present climate


Monthly means and standard deviations


Trends in the monthly means

January

July

SWH U10


Trends in the 99th percentiles

January

July

SWH U10


Trends in exceedances

SWH above 6 m U10 above 17 m/s


ERA-40 vs buoy data 100-yr rv

40100100 30.152.0 −+= ERAbuoy XX

(m)


Assessment against Topex data

( ) log( )m

uux u mσ λ= +


Assessment against Topex data

( ) log( )m

uuux mσ λ= +


STATFJORD

K-13N. CORM

AUKEKOFISK

FORTIES

FRIGG

GULFAKSIreland

U.K.

NorwayIceland

Year

99th

per

cent

iles

of S

WH

(m)

1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005

23

45

67

89

1113

1517 (54N, 13.5W)

Year

99th

per

cent

iles

of S

WH

(m)

1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005

23

45

67

89

1011

1213

14 (61.5N,1.5E)

Year

99th

per

cent

iles

of S

WH

(m)

1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005

23

45

67

8 (54N, 3E)

Time series of monthly 99th percentiles of SWH – MSC50(dashed line ~ mean; blue curve ~ 10yr moving average)


ξξσσμμ ==+= )()()( 1 tttrt o

A non-stationary extreme value analysis was carried out to determine trends in the extremes of the 1958-2001 ERA-40 SWH dataset.

The covariates used were P(t)=t and G(t)=t2 and the likelihood ratio shows that trends are only present in the location parameter and are linear:

The trends are therefore independent of the return period considered.

Trends in the return values


Trends in the return values

JFMAMJ

JAS OND


Future climate


Projection of extremes

What are the effects of future climate changes in return values of waves?

Future climate scenarios predicted by models do not contain SWH, but do contain SLP/U10 with which SWH/Tm relates.

1. Estimate polynomial relations between the parameters of a non-stationary extreme value model and SLP/U10 dependent covariates (ERA- 40).

2. Assume that the relationships found apply in the future.

3. Use future estimates of the SLP/U10 covariates to estimate future SWH/Tm extremes.


Percentual changes in the seasonal SWH 20-yr RV from 1990 to 2089

IS92a

B2

JFM JAS


The present climate of SWH extreme values is non-stationary with linear trends in the location parameter of the extreme value distribution.

Decadal variability affects the SWH extreme values in the Northern Hemisphere.

The future climate of SWH extreme values is characterized by non-linear trends in the location parameter of the extreme value distribution.

The higher changes and more severe extremes are to occur in the more serious future emission scenarios.

The ocean basin to be more seriously affected by climate changes is the North Pacific.

Summary of results based on global models

Caveats: Rather limited future climate analysis! Check Xiaolan’s talk for uncertainties and extended future climate changes description.


High resolution North Sea hindcasts


Wind and waves: ERA-40 (6-hourly)

•boundary waves and input winds from ERA-40•local model with 1kmx1km resolutions

Shallow water wave model schematization

Output location


Extreme wave climate at MP1 (North Sea coastal location)

•Significant trend in the current climate (1958-2001) extremes of significant wave height of about 9 mm/yr •A trend in the projections from 2001 to 2100 of 1 mm/yr•Characteristics of the wave period extremes depend on whether swell or wind- sea events are considered. •If both types of events are considered, the extremes are dominated by swell events and no present or future changes are identified.• Considering the wind-sea events only, a trend of less than 0.01s/yr in the present climate wave periods and a trend an order of magnitude smaller in the projections from 2001 to 2100 were detected.

A1b


Concluding remarks

• Issues in determining past/present climates (inhomogeneities, lack of proper measurements of extremes)

• NH extratropical storms are what can be better described• Changes in NH extratropical storms in past/present and future

climates are significant• Decadal and inter-annual variability larger than climate change

signal• Need for downscaling and including other wave parameters


Recommendations

• Inhomogeneities in present reanalysis datasets need to be addressed.

• There is a lack of reliable measurements of extreme sea states.• The coarse resolution of global climate models is an issue. We

need to look further at regional climate models. • In order to improve non-stationary extreme value fits, covariates

that are more closely related to the considered extremes need to be used. Need to look at the reliability of GCM winds and higher percentiles.

• In shallow waters the sea level rise will also contribute to changes in the sea state extremes and need to be accounted for.

past and future changes in extratropical storms...graham&diaz 6116 3.40 -0.24 0.93 0.26 0.79...

Documents