bulk_flux decomposition of Tau into u,v components

[wilkin]

In bulk_flux.F, once the stress is calculated according to the Fairall et al. COARE algorithm the scalar stress needs to be broken out into its u- and v-direction components.

This occurs in code ...

Code: Select all

...
          Cd=Wstar(i)*Wstar(i)/(Wspeed*Wspeed+eps)
...
!  Compute wind stress components (N/m2), Tau.
!
          cff=rhoAir(i)*Cd*Wspeed
          Taux(i,j)=cff*Uair(i,j)
          ...
          Tauy(i,j)=cff*Vair(i,j)
          

where, for clarity, I have removed the added momentum due to the impact of rain falling on the sea surface.

The intent here is that we have ...

Taux = STRESS * Uair/WindSpeed
Tauy = STRESS * Vair/WindSpeed

because Uair/WindSpeed and Vair/WindSpeed are unit vectors in the direction of the wind.

However, here Wspeed is wind speed augmented by a factor for wind gustiness. In earlier code ...

Code: Select all

          Wspeed=SQRT(Wmag(i)*Wmag(i)+Wgus(i)*Wgus(i))

Thus, the denominator in the unit vector normalization, which should have been simply Wmag, is too large and the output sustr,svstr are slightly less than the magnitude from the COARE calculation.

For a simple 1-D simulation forced by ERA5 fluxes in the Mid-Atlantic Bight the error incurred is at most 6% (see plot below). Typically, it's no more than 1%. The error is greatest at lowest wind speeds when the gustiness can become an appreciable fraction.

So - no cause for alarm! But this will be corrected in a soon to be updated release of bulk_flux.F that implements the COARE 3.5 version from Edson et al. (2013) that performs better at strong wind speeds.