TemperatureType 5year runmean

[mariuslam]

Hi,

I post this here because I have been trying desperately to create a 5 year running mean of single year temperature minimums. But instead of being updated every year and remaining stable, I see that the 5 year running mean behaves weirdly when I plot it (see figure), it should remain the same throughout the year. When I print it in the model it seems fine, but not when I plot from the output file.

Ps: I would also like T_REF24 and T_HARD1 to not contain NaNs, instead have similar values to the ones written before.

I hope someone will be able to have a look at the lines below or at the link and give me a hint of why the output is weird in the history file.

Best regards,

Marius

CTSM_single_site_fateshydro/src/biogeophys/TemperatureType.F90 at c9a2bd4eba549e830c21cd330a2fdddadb58e737 · mariuslam/CTSM_single_site_fateshydro

Contribute to mariuslam/CTSM_single_site_fateshydro development by creating an account on GitHub.

github.com

!-------------------------------------------------------- InitHistory
if (use_fates_hardening) then !Marius
this%t_ref2m_24_patch(begp:endp) = spval
call hist_addfld1d (fname='T2m_24', units='K', &
avgflag='A', long_name='average temperature 2m (24hrs)', &
ptr_patch=this%t_ref2m_24_patch, default='active')
this%t_mean_5yr_patch(begp:endp) = spval
call hist_addfld1d (fname='T_HARD5', units='K', &
avgflag='A', long_name='5 year mean of minimum yearly 2-m temperature for hardening', &
ptr_patch=this%t_mean_5yr_patch, default='active')
this%t_min_yr_patch(begp:endp) = spval
call hist_addfld1d (fname='T_HARD1', units='K', &
avgflag='A', long_name='minimum year 2-m temperature for hardening', &
ptr_patch=this%t_min_yr_patch, default='active')
end if
!----------------------------------- InitAccBuffer
this%t_mean_5yr_patch(bounds%begp:bounds%endp) = spval
call init_accum_field (name='T_HARD5', units='K', &
desc='5 year average of yearly min 2-m temperature for hardening', accum_type='runmean', accum_period=-5, &
subgrid_type='pft', numlev=1, init_value=0.0_r8)
end if

!------------------------------------- UpdateAccVars
if (use_fates_hardening) then
! Accumulate and extract T_REF24
call update_accum_field ('T_REF24' , this%t_ref2m_patch , nstep)
call extract_accum_field ('T_REF24' , rbufslp , nstep)
do p = begp,endp
this%t_ref2m_24_patch(p) = rbufslp(p)
end do
! Start 1 year minimum temperature loop for hardening
end_yr = is_end_curr_year()
do p = begp,endp
if (rbufslp(p) /= spval) then
this%t_min_yr_inst_patch(p) = min(rbufslp(p), this%t_min_yr_inst_patch(p))
endif
if (end_yr) then
this%t_min_yr_patch(p) = this%t_min_yr_inst_patch(p)
rbufslp(p)=this%t_min_yr_inst_patch(p)
this%t_min_yr_inst_patch(p) = spval
else if (secs == dtime) then
this%t_min_yr_patch(p) = spval
endif
end do
if (end_yr) then
call update_accum_field ('T_HARD5', rbufslp, nstep)
call extract_accum_field ('T_HARD5', this%t_mean_5yr_patch, nstep)
end if
end if

 

[oleson]

Is this set correctly:

call init_accum_field (name='T_HARD5', units='K', &
desc='5 year average of yearly min 2-m temperature for hardening', accum_type='runmean', accum_period=-5, &
subgrid_type='pft', numlev=1, init_value=0.0_r8)

With accum_period=-5, this would be 5 days, not 5 years?

 

[oleson]

I implemented your code into a global SP simulation and it seems to be working as intended, at least for a few grid cells I checked. I've attached a plot of T_HARD5 for 6 years for a high-latitude point.
I'm a FATES novice but maybe there is something weird associated with patches changing weights or something? Your first few years look stable?

 

[mariuslam]

Hi Keith, Thanks a lot for having such a deep look into it. I agree that the patch averaging is likely to be the cause. Especially since I have a lot of mortality during the years where Tmean5yr starts fluctuating, then it stabilizes again. However, I think the cause of the problem is in CLM as I write the variable to the output file in the TemperatureType file straight away. Is that possible? I will try writing the variable in the output from fates as well and I expect it to be stable there

 

[mariuslam]

I now write the variable in FatesHistoryInterface and it is stable there.

 

[oleson]

Our FATES expert here advises to post this as an issue on the FATES github. It is useful to have that on the FATES forum to capture the discussion for other folks, as the patch dynamics are different from static columns.

 

[mariuslam]

Ok I will post it there :)