Three changes and one bug fix in the NSST model in the coupled frame

[XuLi-NOAA]

      Background
      A scheme to determine the foundation temperature (Tf, or tref in the code) with the NSST model T-Profile and the oceanic model (MOM6 here) top layer temperature (T1) has been developed in NCEP atmospheric model (FV3 here) years ago, but it is never tested systematically in the frame of the coupled (atmosphere-ocean) model. This scheme, Scheme-A, has been tested with the Prototype 5 (P5) of UFS-S2S version, a more reasonable scheme, Scheme-A.1, has been proposed and tested. 
     The second change is to apply a safeguard to tref to make sure it is not colder than the frozen point.
     The third change to reduce the maximum diurnal warming depth from 30 m to 20 m. The reason to do this is that it is easier to handle the necessary imported oceanic model (MOM6 here) layer temperatures (down to 20 m, instead of 30 m). It is also based on the fact the diurnal warming layer is rarely deeper than 20 m. 
     A bug has been found out in the evaluation of the impact of the NSST model on the coupled model. 

Three changes in this issue:

(1) Scheme-A to Scheme-A.1 in ccpp/physics
The cooling effect is removed from the formula to determine Tf
(2) Apply a safeguard to tref in sfc_nst_pre_run
(3) Reduce z_w_max from 30 m to 20 in ccpp/physics and FV3/io

A bug fix in ccpp

     There are 4 equations involved in the surface (nsst here, specifically) physics, a two step interaction is introduced to calculate some surface coefficients twice when the surface wind is smaller than 2 m/s in the surface physics. 

In order, the four following equations are executed in the code :

(1) tref(i) = max(tgice, tsfc_wat(i) - dtzm(i)
(2) tsfc_wat(i) = max(tgice, tref(i) + dtzm(i))
(3) tref(i) = max(tgice, tsfc_wat(i) - dtzm(i))
(4) tsfc_wat(i) = max(tgice, tref(i) + dtzm(i))

You can see tref is assigned twice (1) and (3), and tsfc_wat is assigned at (2) and (4), it explains what we saw in the figure exactly.

The solution is to introduce tsfco (the T1 from MOM6) into sfc_nst_pre_run in ccpp, and replace tsfc_wat in (1) and (3) to be tsfco.

[climbfuji]

@XuLi-NOAA thanks for reporting the bug and providing a solution for it! @SMoorthi-emc @junwang-noaa do you agree with the suggested bugfix?

[junwang-noaa]

I knew the bug fix 2) is required to fix the first step large nsst impact, not sure about 1) and 3).

[SMoorthi-emc]

I think this was done correctly in the IPD physics driver - bug happened during conversion to CCPP (I am only referring to the "tsfco" part of the bug) Other changes are up to Xu Li. I have corrected the "tsfco" error in my own branch of ccpp.
Moorthi

[climbfuji]

Closed via #589.