bourassa@met.fsu.edu 4th SEAFLUX WorkshopSept. 27, 2007

A Fear-Inspiring Intercomparison of

Monthly Averaged Surface Forcing

Mark A. Bourassa, P. J. Hughes, and S. R. SmithCenter for Ocean-Atmospheric Prediction Studies, and

Department of Meteorology

Florida State University

The Florida State University 2 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Why Fear?

I get many phone calls and emails asking which product should be used. The nine, easily obtained, products that we examined are inconsistent.

Different magnitudes, patterns, and distributions Different input data

Different spatial and temporal sampling characteristics Different constraints, quality control, and bias correction

Different flux parameterizations

The Florida State University 3 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Products in Our Comparison

Product LHF SHFStress (x,y)

Wind Speed

u wind

v wind Tair Qair SST

Product Type Grid Spacing

NCEPR2 x x x x x x x x ReanalysisGaussian (T62, 194x94)

JRA25 X x x x x x x x Reanalysis (T106) ~120kmERA40 x x x x x x x x Reanalysis 1 1/8 degreesWHOI x x Hybrid 1 x 1 degreeGSSTF2 x x x x x Hybrid 1 x 1 degreeIFREMER x x x x x x x x x Satellite 1 x 1 degreeHOAPS2 x x x x x Satellite 0.5 x 0.5 degreeFSU3 x x x x x x x x x In-situ 1 x 1 degreeNOC1.1 x x x x x x In-situ 1 x 1 degree

All these products are regridded (if necessary) on to a similar 1x1 grid.

A common land mask is applied.

A common period: March 1993 to Dec. 2000

The Florida State University 4 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Land Mask

In our comparison, we mask out all values within 2 of land.

Bia

s re

lativ

e to

QS

CA

T S

peed

s (m

s-1

)SSM/I NOGAPS ECMWF NCEPR1

Figures show differences in 10m wind speed, relative to QSCAT, adjusted to non-neutral values.

The Florida State University 5 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

JRA25 – ERA40 Root Mean Square differences (using only 00Z and 12Z times) for Calendar year 2001: 500 mb temperature (K)

Langland, Maue, and Bishop, (submitted to Science) – Temperature Uncertainties Collaboration part of NRL summer internship (2007).

The Florida State University 6 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Atlantic Meridional Stress

FSU3

The Florida State University 7 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Atlantic Zonal Stress

FSU3

The Florida State University 8 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Curl of the Stress

Ship tracks are apparent in many products, as are TAO buoys

NCEPR2 IFREMER

FSU3 NOCv1.1

The Florida State University 9 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Standard Deviation of the Curl of the Stress

Ship tracks are apparent in many products, as are TAO buoys

NCEPR2 IFREMER

FSU3 NOCv1.1

The Florida State University 10 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Atlantic Latent Heat Flux

FSU3

The Florida State University 11 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Atlantic Sensible Heat Flux

FSU3

The Florida State University 12 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Latent Heat FluxJRA

ERA40

NCEPR2

WHOI GSSTF2

The Florida State University 13 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Latent Heat Flux

Satellite and In situ products

HOAPS2IFREMER

NOC FSU3

The Florida State University 14 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Reanalysis Products Have Differing Means and Distributions- Even Zonal Stress on a Monthly Scale -

Tropical Atlantic

99th

95th

90th

75th

50th

10th

5th

25th

1st

Tropical Atlantic Southern Ocean

The Florida State University 15 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Southern Ocean qair and LHF

The Florida State University 16 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

March average salinity difference between a model forced by climatology wind stress and a model forced by 12-hourly scatterometer-derived wind stresses. The high-frequency winds force intermittent offshore transport of low salinity water which results in freshening over the middle shelf.

The Florida State University 17 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

4 m ADCP (red) and NCOM ETA (black) Velocity

8/ 1 8/ 8 8/15 8/22 8/29 9/ 5 9/12 9/19 9/260

20

40

60

80

100

Sca

le c

m/s

Complex Correlation: Magnitude = 0.52, Phase Angle = -6.3°, Speed Correlation = 0.32

4 m ADCP (red) and NCOM Qscat/ETA (black) Velocity

8/ 1 8/ 8 8/15 8/22 8/29 9/ 5 9/12 9/19 9/260

20

40

60

80

100

Sca

le c

m/s

Complex Correlation: Magnitude = 0.84, Phase Angle = 5.5°, Speed Correlation = 0.84

4 m ADCP (red) and NCOM Qscat (black) Velocity

8/ 8 8/15 8/22 8/29 9/ 5 9/12 9/19 9/260

20

40

60

80

100

Sca

le c

m/s

Complex Correlation: Magnitude = 0.61, Phase Angle = 1.0°, Speed Correlation = 0.46

8/ 1

NCOM vs. COMPS ADCP 4m VelocityT.S. Harvey

Eta

QuikSCAT

QuikSCAT/Eta

Modeled

Modeled

Modeled

Observed

Observed

Observed

The Florida State University 18 bourassa@met.fsu.edu 4th SEAFLUX Workshop

Sept. 27, 2007

Conclusions

Each flux product has it’s strengths and weaknesses. The choice of which one is best is dependent on the application.

There are large differences in atmospheric humidity and scalar wind speed (and air temperature in one product) that contribute to differences in surface fluxes.

There are many applications for which temporal resolution must be much better than monthly. NWP products do not assimilate satellite data as well as would be

desired, and In situ products do not have sufficient spatial/temporal resolution

over much of the global oceans.