[RegCNET] column-level soil/lake energy conservation error (W/m**2)

[Felipe Roque]

Thanks for your reply, Travis.

I asked that because I´m searching possible errors in my runs. My
simulations using MIROC dataset  as initial and boundary conditions always
finish with CFL violation at specific days. I put timestep with 15 seconds,
but CFL violation continues. I´m using dx with 50 km. I´m looking for
errors in MIROC dataset, but no success until now.

An important aspect considering MIROC dataset is that it has many missing
values in the bottom of atmosphere (1000, 975, 950 mb) of South America
(because of Andes mountains). This occurs because MIROC is a spectral
model. Is there any special treatment for avoiding missing values in input
dataset? I´ve made a basic interpolation based on 4 neighborhood points,
but I´m really don´t know if it has some influence on CFL violation.

I made the following in mod_miroc.F90 (I created it) after reading netcdf
file.

! INTERPOLATION TO ELIMINATE MISSING VALUES (-999.)
! klev = 23 | form 1 to 1000 mb
!  ilon = 320 | ilon = 1 (0.00 E) | ilon = 320 (358.875 E)
!  jlat = 160 | ilat = 1 (89.1415 N) | ilat = 160 (89.1415 S)

          do k = 1 , klev
            do j = 2 , jlat-1
              do i = 2 , ilon-1
                if (work(i,j,k)==-999.0000)then
                   contador=0
                   if (work(i-1,j,k)==-999.0000) then
                      temp1=0.
                      contador=0
                   else
                      temp1=work(i-1,j,k)
                      contador=1
                  endif
                   if (work(i+1,j,k)==-999.0000) then
                      temp2=0.
                   else
                      temp2=work(i+1,j,k)
                      contador=contador+1
                   endif
                   if (work(i,j-1,k)==-999.0000) then
                      temp3=0.
                   else
                      temp3=work(i,j-1,k)
                      contador=contador+1
                   endif
                   if (work(i,j+1,k)==-999.0000) then
                      temp4=0.
                   else
                      temp4=work(i,j+1,k)
                      contador=contador+1
                   endif
                work(i,j,k)=(temp1+temp2+temp3+temp4)/contador
                endif
              end do
            end do
          end do

Anyone can give an opinion about this method? Is there anything better to
solve missing value points?

Thanks to everybody!
Felipe.

2012/8/15 Travis O'Brien <travis.obrien at gmail.com>

> Hi Felipe,
>
> This should not be anything to worry about.  This is a typical occurrence
> in CLM output.  To be certain, I have asked a CLM-developer colleague of
> mine (Dr. Jinyun Tang), and he indicated that this message often appears
> because of rounding errors.
>
> Kind Regards,
> -Travis-
>
> On Mon, Aug 13, 2012 at 8:00 AM, Felipe Roque <felipenrs at gmail.com> wrote:
>
>> Hi Friends,
>>
>> I´m using RegCM-4.1. Anyone knows what could cause an error in subroutine
>> BalanceCheck (~/Main/clm3.5/src/biogeophys/BalanceCheckMod.F90)? This
>> warning doesn´t stop regcm, but is it a critical error? Should I worry
>> about it?
>> This subroutine accumulates the numerical truncation errors of the
>> water and energy balance calculation. It is helpful to see the performance
>> of the process of integration.
>> The error for energy balance = abs(Net radiation - change of internal
>> energy - Sensible heat - Latent heat)
>>
>>
>>  soil balance e nstep =     49500 point =  4105 imbalance =    0.00 W/m2
>>  soil balance e nstep =     49500 point =  5456 imbalance =    0.00 W/m2
>>  soil balance e nstep =     49500 point =  2304 imbalance =    0.00 W/m2
>>      at day =   68.7497, ktau =     198000 :  1st, 2nd time deriv of ps =
>>  0.10248E-04 0.39686E-07,  no. of points w/convection =    3309
>>             *** Day      69.7500 solar declination angle =  -4.23964630
>> degrees.
>>   dectim =    397440.000000000
>>  ATM variables written at   1989031018  0.000000000000000E+000
>>  SRF variables written at   1989031018  0.000000000000000E+000
>> SEARCH BC data for 1989031006
>> READY  BC from     1989031000 to 1989031006
>>  soil balance e nstep =     49501 point =   614 imbalance =    0.00 W/m2
>>  soil balance e nstep =     49501 point =  4743 imbalance =    0.00 W/m2
>>   soil balance e nstep =     49501 point =  2304 imbalance =    0.00 W/m2
>>  soil balance e nstep =     49501 point =  5456 imbalance =    0.00 W/m2
>>  soil balance e nstep =     49501 point =  4599 imbalance =    0.00 W/m2
>>  soil balance e nstep =     49501 point =  2716 imbalance =    0.00 W/m2
>> .......
>> .......
>> .......
>>
>> Thanks, Felipe.
>>
>> _______________________________________________
>> RegCNET mailing list
>> RegCNET at lists.ictp.it
>> https://lists.ictp.it/mailman/listinfo.cgi/regcnet
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ictp.it/pipermail/regcnet/attachments/20120815/7f3ca6cb/attachment-0001.htm>