how about nudging SST to observation?

[liangailin]

dear all:
we use 'SRELAXATION' restores the modeled SSS to the SSS in forcing file, how about SST? If I want restore the model SST to the SST in forcing files, what should I do?
regards

[m.hadfield]

Use QCORRECTION (net heat flux correction as a function of model SST minus forcing SST). You will need to specify a dQdSST. I normally put this in its own forcing file with a single value for the whole domain (though it can vary spatially). Here's the CDL for one I just created:

Code: Select all

netcdf roms_frc_dqdsst {
dimensions:
        time_fixed = 1 ;
variables:
        float time_fixed(time_fixed) ;
        float dQdSST(time_fixed) ;
                dQdSST:time = "time_fixed" ;
                dQdSST:long_name = " surface net heat flux sensitivity to SST" ;
                dQdSST:units = "Watt metre^-2 Celsius^-1" ;

// global attributes:
                :type = "ROMS forcing file" ;
                :history = "Created 2009-04-23T00:31:24" ;
data:

 time_fixed = 0 ;

 dQdSST = -40 ;
}

For salinity relaxation you can also use SCORRECTION. I forget right now how this differs from SRELAXATION, but a quick scan through the source code will tell you.

[liangailin]

hi,m.hadfield; thanks for your reply!
but I still have a question: how can I calculate dQdSST or any website to download this data??

[m.hadfield]

The justification for using QCORRECTION is that the SST dataset you're forcing towards lacks the mesoscale structure that the model produces and that in reality, there will be a tendency for those mesoscale SST features to be damped out by the atmosphere.

It's a bit of a fudge really, so you can use whatever value you want, within reason, and the value of -40 that I normally use is as good as any.

That said, it is possible to calculate the sensitivity of net heat flux to SST, assuming that temperature and humidity at 1.5 m above the ocean are fixed, as is wind speed at 10 m. Values of this quantity were included in the COADS dataset at one stage, and around New Zealand it tends to be around -40, so that's why I chose this value.

This value almost certainly overestimates the damping effect, as in reality warm SST anomalies will have warmer air over them. The only way to calculate this effect properly is to include a model of the atmospheric boundary layer. I believe some ROMS users do this.

A Google search on dqdsst throws up quite a few hits.

[kate]

Are you using BULK_FLUXES? If so, it is already computing a sort of nudging to SST via the computation of the heat fluxes as a function of SST. The justification for nudging to SSS is that there is no feedback to SSS in the bulk flux routine.

[lmkli]

Are you using BULK_FLUXES? If so, it is already computing a sort of nudging to SST via the computation of the heat fluxes as a function of SST. The justification for nudging to SSS is that there is no feedback to SSS in the bulk flux routine.

Hi, Kate,

Do you mean BULK_FLUXES has the nudging to SST or Tair(air temperature in BULK_FLUXES forcing file)? I'm a little confused because I don't remember BULK_FLUXES needs SST as input in forcing file.

[kate]

Bulk_fluxes uses Tair, which is related to SST. If the Tair is from a model, it will respond to the SST that was provided to the atmospheric model. What I meant was that SST has a negative feedback to keep it in line while SSS does not. To first order, the feedback can be linearized to look like nudging.

[lmkli]

Thanks a lot Kate.

What if I want to relax the model SST to another SST dataset which has finer resolution than Tair?
From ROMS cppdefs I can only find QCORRECTION to do it. Is there any other way to do this like SCORRECTION
or SRELAXATION?

[Pysh]

And my question. If in the upper mixed layer temperature of the water can be considered equal SST, in the case of LMD parameterization of vertical mixing (and determination in a model of the depth of the upper mixed layer Hsbl) can I use nudging the temperature of the water to the measured SST in the full upper mixed layer? I've done this experiment, and I think that in this case a temperature field quickly approaching observed than in the case of QCORRECTION. In addition to the inertia of the method of adaptation of the fields of temperature by correction of the heat flux this method can have such a significant disadvantage, as not sufficient precision of the calculation (due to the diversity of conditions in the lower layer of the atmosphere, water types, wave mixing, etc). Without doubt correctness of the LMD-parameterization and calculation of Hsbl requires verification, but it is another matter.