[MITgcm-support] realfreshwaterflux usage in the Arctic

[Martin Losch]

Hi Jonathan,

I current don't use the ECCO2 Arctic face configuration, but something similar.

In a global model we know tha tthe net freshwater flux into the ocean is zeros (when averaged over some time, usually a full seasonal cycle, let along climate trends), and the global sea level fluctuates according to seasonal fluctuations of the freshwater flux (and the surface temperatures etc.)

With any regional configuration, you'll have to close the freshwater budget yourself. The Arctic has very little evaporation and a quite a lot of continental run-off and (I am guessing that) on average it's a source of oceanic volume rather than a sink (I may be wrong about the latter part, but it's not relevant for closing the budget). You'll have to make sure that you open boundary conditions (the volume flux through the open boundaries) actually balance the surface flux, otherwise you'll get dramatic sea level changes very quickly. It's much easier to use virtual salt flux instead, that way you just have to make sure that the net volume flux through the open boundaries is zero and surface fresh water flux does not contribute to volume changes.

I have a configuration, where I need the non-linear free surface and I also want "realFreshWaterFlux=.TRUE." (note that even with this flag, salinity is still only changed through a virtual salt flux, but the volume change is accounted from properly, only with the non-linear free surface, salinity is changed directly by changing volume). In this configuration I tune the open boundary flux to approximately balance the sea-level change, i.e. I run the model, diagnose the sea level change and modify the boundary conditions in a second run. One could try to do this "online", but you'd have to code it yourself. There is a "balanceEmPmR" flag, but that should (to my mind) only be used in global configurations, because it subtracts the mean EmPmR at every timestep, and this may  not be small in regional configurations.

So, in order turn on the "realFreshWaterFlux" in data &PARM01:
 exactConserv = .TRUE.,
 useRealFreshWaterFlux=.TRUE.,
# for the non-linear free surface parameters with rstar-coordinate I have this:
 hFacInf=0.1,
 hFacSup=5.,
 select_rStar=2,
 nonlinFreeSurf=4,

Why does freshwater lead to sea level change everywhere (even far away from the sources)? Simply fast adjustments to pressure gradients. When you add water volume in one surface grid cell, you'll immidiately have d\eta/dx and this pressure forcing drives a fast gravity wave that will lead to very fast (not instantaneous!) adjustments. In the model, you don't see all of this (although most if it, if you look at a sequence of timesteps), because the surface elevation/pressure is solved with an implicit solver (in order to reduce the time step requirements of fast gravity waves). These implicit solvers damp the fast waves a little, but the effect is preserved. [This is not the most accurate description of what's going on.]

The river runoff, it's added, where you want it to be added (where you prescribe in the appropriate file/field). Nowhere else.

Martin

On Sep 30, 2013, at 3:03 AM, Jonathan Whitefield <jwhitefield at alaska.edu> wrote:

> Hi everyone.
> 
> Dimitris Menemenlis and I were discussing the way that the model
> receives fresh water this week, and I found out that, for the ECCO2
> Arctic face regional configuration, realfreshwaterflux is disabled,
> and freshwater is input solely as a salinity anomaly, as opposed to
> "normally", where a volume of fresh water is added. I am planning on
> doing some runs that look at changes in the Arctic due to river input,
> so I have a couple of questions and concerns that I wanted to bring to
> the wider community.
> 
> First of all, has anyone out there run the Arctic with
> realfreshwaterflux enabled? If so, how do I turn it back on. If not,
> is it actually possible, or even advisable to re-enable it? What
> happens if I change the volume flow of the rivers? Does the fact that
> realfreshwaterflux is disabled mean that (for example) doubling the
> volume input of a river would have no difference?
> 
> Secondly, I was told that when realfreshwaterflux is enabled in the
> global model, it adds the volume across the entire ocean at once. This
> seems unrealistic to me - how would flow from an Arctic river be added
> to a sub-tropical ocean? I would like to be able to have the river
> volume be added only where the rivers discharge in to the ocean,
> allowing me to look at the effects of changing the discharge on
> regional buoyancy driven circulation, so is there an option to disable
> this, and have the volume only where it should be?
> 
> Apologies for the multitude of questions at once, but hopefully
> someone out there has flipped the same switches I want to! Please
> contact me if you need clarification on any of my questions, either
> through the support list, or personal email.
> 
> Thanks!
> 
> -- 
> Jonathan Whitefield, M.Sc.
> Ph.D. Student, Physical Oceanography
> 
> 110 O'Neill Building,
> P.O. Box 757220
> University of Alaska Fairbanks
> Fairbanks, AK 99775-7220
> 
> Phone: (907) 474-5184
> Fax: (907) 474-5863
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support