Near-wall SGS model

[cbegeman]

@vanroekel @xylar I'm noticing that the modeled SGS momentum fluxes do not smoothly vary below the surface layer (SGS model is Moeng-Wyngaard). Saiki et al. (2010) discuss an issue with SGS models in the vicinity of boundaries, which is that the model can fail to produce a smooth transition between surface fluxes (prescribed) and vertical fluxes though grid cells near the boundary. As the resolved fluxes tend toward 0 near the boundary, there should be a concomitant increase in SGS fluxes to the prescribed surface flux. The fix that Saiki et al. propose is supposed to be included in PALM (Maronga et al., 2015) but I've noticed that it is not implemented in the version we've been working with. Do you mind taking a look at Saiki et al. and letting me know if you think this approach would be worth implementing? I'd be using the MOST shape functions from McPhee's work in equations 11 and 14. Interestingly, I'm not seeing the same discontinuities in SGS fluxes of pt and sa in the cases I'm running. I'm not sure why this is yet.
Thanks!

[vanroekel]

sure I'll take a look. I skimmed very briefly just now and this seems like a continuation/extension of Sullivan et al 1994. So whether or not we need to implement this is completely dependent on use case. I think for you this would be an important addition. But for Qing or LeAnn it would be less so. In fact for larger turbulence problems in the boundary layer Peter told me not to use the scheme. This scheme is only important if your problem depends strongly depends on getting MOST right.

Is this in newer versions of PALM? Would it be easy to cherry pick from a newer version?

[cbegeman]

The newer version also does not include this change. It only seems to treat MOST surfaces separately in RANS mode.
I think that for our problem we do need to get MOST right.