Questions for TREFHT values at coastlines and T difference between history and restart files.

[chunhw@gmail_com]

Dear CESM users,   While running cesm 1.1.1 with dynamic core of HOMME, I found several points that I cannot understand.  Please find the attached figure.  We have chosen the option of "instantaneous" for printing variables in the history file.  In Figure (b), you can find spurious values of 2m temperature, as a result of history file (TREFHT in the file of *.h0.*.nc) along with coastlines (below 200K), which are not reasonable.  Values over other areas look reasonable though, when compared with Figure (a) TS and (c) T from history file.  Is there anyone that can answer why it does not have physically reliable values over the coastlines?   Next, we printed out the same variable of air temperature into the restart file (*.r.*.nc) and the history file (*.h0*.nc).  Interestingly and unexpectedly, the values are not same and the difference between two does not seem negligible as you can see in Figure (d).  It seems to me that the difference looks significant especially over the edges of the cube.  Is this caused by some post-processing (or remapping) between the cubed sphere grid and a lat-lon grid, which are not unified between restart file and history file?  What do you think about this?    Your comments would help my work a lot.  Thank you very much in advance.

 

[fsche@mech_kth_se]

So I join the club on this problem - I have the same issue. My default output files for TREFHT show temperatures down to 10 K. For TS, deviations are less severe but still up to 20 K compared to neighboring grid points. Affected grid points are along the coastlines only. It looks comparable to the figure attached in the previous posting above.I'm using the F_1850 case (CESM 1.0.5) but with paleo topography and ice sheets. So I created new land-sea masks etc and my paleo SST and icefrac were just created by interpolating and replacing the standard climatology file for HadSST.Could it be that the grid points with land fraction being not 0 or 1 at the coasts are calculted by weighted averaging (based on the land-ocean fraction) of the SST file (in deg. Celsius) with the atmosphere (in K)? Looking e.g. at total cloud cover, the effect is clearly visibile also here with much higher cloud cover over the inconsistent grid points. So the problem is also seen by the model and not only by the post-processing...In the user guide is written for DOCN:SSTDATA mode assumes the only field in the input stream is SST. It also assumes the SST is in Celsius and must be converted to Kelvin. The files from HadSST are in Celsius and so are my paleo SST.

 

[fsche@mech_kth_se]

So I join the club on this problem - I have the same issue. My default output files for TREFHT show temperatures down to 10 K. For TS, deviations are less severe but still up to 20 K compared to neighboring grid points. Affected grid points are along the coastlines only. It looks comparable to the figure attached in the previous posting above.I'm using the F_1850 case (CESM 1.0.5) but with paleo topography and ice sheets. So I created new land-sea masks etc and my paleo SST and icefrac were just created by interpolating and replacing the standard climatology file for HadSST.Could it be that the grid points with land fraction being not 0 or 1 at the coasts are calculted by weighted averaging (based on the land-ocean fraction) of the SST file (in deg. Celsius) with the atmosphere (in K)? Looking e.g. at total cloud cover, the effect is clearly visibile also here with much higher cloud cover over the inconsistent grid points. So the problem is also seen by the model and not only by the post-processing...In the user guide is written for DOCN:SSTDATA mode assumes the only field in the input stream is SST. It also assumes the SST is in Celsius and must be converted to Kelvin. The files from HadSST are in Celsius and so are my paleo SST.

 

[fsche@mech_kth_se]

Was that problem ever solved or reported elsewhere?

 

[fsche@mech_kth_se]

Was that problem ever solved or reported elsewhere?

 

[hannay]

I haven't been able to reproduce the problem. I output instantaneous TREFHT and everything looks fine around coastlines. 

 

[hannay]

I haven't been able to reproduce the problem. I output instantaneous TREFHT and everything looks fine around coastlines. 

 

[fsche@mech_kth_se]

Also TS is affected, TREFHT is just much worse. I get the problem for paleo tropographies using F1850 compset and prescribed SST (I use TS rather than SST) from TraCE. 

 

[fsche@mech_kth_se]

Also TS is affected, TREFHT is just much worse. I get the problem for paleo tropographies using F1850 compset and prescribed SST (I use TS rather than SST) from TraCE.