Installation and examples for flexpart-wrf v.3.3.2

[fkleinert]

I am trying to install flexpart-wrf v3.3.2 but have some difficulties to run the examples provided in the source code. I plan to use the hybrid MPI-OpenMP version on a HPC system. To familiarise my self with wrf-flexpart (I do not have worked with flexpart-wrf nor flexpart before), I started with the serial version. After including the NetCDF-Fortran path makefilie.mom I used make -f makefile.mom serial to compile the serial version. The executable flexwrf33_gnu_serial is succeesfully created, however I got the following warnings during compilation:

netcdf_output_mod.F90:1619:70:

 1617 |         do jy=0,grid_ny-1
      |                         2                                             
 1618 |           do ix=0,grid_nx-1
 1619 |             grid3(ix,jy,kz,kp,nage)= 1.e12*grid(ix,jy,kz)/grid_vol(ix,jy,kz)/outnum*  &
      |                                                                      1
Warning: Array reference at (1) out of bounds (0 < 1) in loop beginning at (2)
netcdf_output_mod.F90:1619:67:

 1618 |           do ix=0,grid_nx-1
      |                           2                                        
 1619 |             grid3(ix,jy,kz,kp,nage)= 1.e12*grid(ix,jy,kz)/grid_vol(ix,jy,kz)/outnum*  &
      |                                                                   1
Warning: Array reference at (1) out of bounds (0 < 1) in loop beginning at (2)
random.f90:68:14:

   68 | 1   v1=2.*ran3(idum)-1.
      |              1
Warning: Missing actual argument for argument ‘inext’ at (1)
random.f90:69:14:

   69 |     v2=2.*ran3(idum)-1.
      |              1
Warning: Missing actual argument for argument ‘inext’ at (1)
verttransform.f90:769:24:

  769 |       do 100 ix=0,nxmin1
      |                        1
Warning: Fortran 2018 deleted feature: Shared DO termination label 100 at (1)
verttransform_nests.f90:442:27:

  442 |        do 100 ix=0,nxn(l)-1
      |                           1
Warning: Fortran 2018 deleted feature: Shared DO termination label 100 at (1)
readinput.f90:118:0:

  118 |   if (path(numpath+2*(i-1)+1)(1:5).ne.'=====') then
      | 
Warning: ‘__builtin_memcmp_eq’ reading 5 bytes from a region of size 0 [-Wstringop-overflow=]

Are there any fixes for this?

After compilation, I tried to run one of the examples in examples/. As I could not find any data/ directory related to v3.3.2, I downloaded the 'WRF input files' (Wrfdata/), from flexpart-wrf version 3.1.
Moreover, I adopted the following lines in flexwrf.input.forward1:

• In the 'Former Pathnames File' section (all paths are absolute paths)
  • Output directory (directory already exists)
  • Path to meteo input data (Wrfdata)
  • Location of the AVAILABLE2 file
• In the 'Former Command File' section:
  • Set SFC_OPTION to zero (as mentioned in #271)

Running ./flexwrf33_gnu_serial flexwrf.input.forward1 yields

 the input file used is flexwrf.input.forward1                                                                                                                                                                               
 Pre-generating random numbers
 Calling readinput
 Opening 'flexwrf.input.forward1                                                                                                                                                                               ' for reading
 Reading pathnames
 option_verbose=         100
              ----------------               
  INFORMATION: SUBGRIDSCALE TERRAIN EFFECT IS
  NOT PARAMETERIZED DURING THIS SIMULATION.  
              ----------------               
 SFC_OPTION =           0
 nageclass           2
 Opening file: /p/project/deepacf/intelliaq/kleinert1/flexpart-wrf_ticket-test/Src_flexwrf_v3.3.2/examples/AVAILABLE2 for reading
 readinput() calling gridcheck()

test_xyindex_to_ll_wrf calling map_set -- map_proj_id, map_set_proj_code =    1    3

test_xyindex_to_ll_wrf -- lgrid, rmserr (km) =  0  1.22E-03



# of vertical levels in WRF data
    n_bottom_top & "true" nuvz:     59     59
    nwz &     "augmented" nuvz:     60     60

    nwzmax, nuvzmax, nzmax    :     61     61     61
    nconvlevmax, nconvlev     :     60     58
    nx, ny, nxmax, nymax      :    180    180    250    190

Mother domain:
  east-west   range:        0.0  2148000.0   Grid distance:    12000.0
  south-north range:        0.0  2148000.0   Grid distance:    12000.0

 readinput() calling gridcheck_nests()
 Reading OUTGRID

readoutgrid diagnostics
xtmp1, ytmp1 (in) -1.2300000000E+02  3.8000000000E+01
xtmp2, ytmp2 (in) -1.2000000000E+02  4.1500000000E+01
out_xm0, out_ym0   3.6903406250E+05  1.0307651250E+06
dxout,   dyout     1.4537512695E+04  1.0897026367E+04
 Reading OUTGRID nest

readoutgrid_nest diagnostics
xtmp1, ytmp1  (in) -1.2300000000E+02  3.8000000000E+01
xtmp2, ytmp2  (in) -1.2000000000E+02  4.1500000000E+01
out_xm0n, out_ym0n  3.6903406250E+05  1.0307651250E+06
dxoutn,   dyoutn    1.4537512695E+04  1.0897026367E+04
 Reading RECEPTORS
 Reading SPECIES
 option_verbose:         100
 option_verbose:         100
 Reading RELEASES
 nspec =            2
 emitvar =            0
 Reading species:           1 of           2
 Reading species:           2 of           2
  Releasepoints allocated:            2
 numpoint =           2
 release location=           1
 COMPOINT = "box1              NAME OF RELEASE LOCATI
 release location=           2
 COMPOINT = "box2                                    
 maxpart2      110000           0
 alloc part      110000      110000           0
 Main: Old IGBP data will be used for land use categories
  #### FLEXPART ERROR! FILE CONTAINING          ####
  #### RELATION LANDUSE,z0 DOES NOT EXIST       ####
  #### IT IS NEEDED FOR DRY DEPOSITION          ####
  #### COPY DATA FROM src_flexwrf_v3.0/data     ####
  #### IN YOUR CURRENT DIRECTORY                ####
  #### OR USE A DIFFERENT SPECIES               ####

which seems to be related to the dry deposition of Cs-137. The IGBP_int1.dat file is located in the Wrfdata/ directory (LU_OPTION is set to zero as given by default). I also copied this file as well as the executable into the examples/dir.

Is this error related to some changes between flexpart-wrf versions 3.1 and 3.3.2?
Are the example input data from version 3.1 outdated, if so are there new example data?

For simplicity, I replaced the Cs-137 with a second air tracer to omit the deposition. This, however, results in a Richardson failure. In #242 daliaga supposed to uncomment some lines in readwind.f90, and readwind_nests.f90 which finally resulted in a complete serial forward model run after recompilation (make -f makefile.mom [clean, serial]).
Both, the forward1 and die backward1 example showing the following warning WARNING: No (or incomplete) flux data contained in WRF output file wrfout_d02_2010-05-18_00.nc but did not crash. I am therefore unsure if this is related to #233.

Can you confirm that the steps mentioned above are adequate to test the installation of flexpart-wrf v3.3.2 for the serial and omp mode?

Before continuing with the hybrid-mode, I thought it might be good to clarify all the issues mentioned above to ensure that they won't cause any additional issues later on.
I already saw that I might have to include the MPI_ROOT path into makefile.mom for compilation.

Thank you in advance for your answer and best regards!

[pesei]

As a first quick answer, I think the compiler warnings are nothing to worry (haven't investigated them now, though).

[fkleinert]

Thanks for your first answer. Do you have some additional information on the remaining questions?
Best regards.

[ignacio]

LU_OPTION = 1 was an experimental feature that didn't improve the results as much as expected.
LU_OPTION = 0 is recommended.

The "Richardson failure:" error often appears from malformed text input not detected at reading time. A better handling of this error is intended in future versions. For now, please double check the input file format.

As for "WARNING: No (or incomplete) flux data contained in WRF output file wrfout_d02_2010-05-18_00.nc", it only indicates that the fluxes are calculated internally and not read as input form WRF winds. This is the default mode also in the main FLEXPART version. The resulting simulations are still valid.

[pesei]

I assume that this is solved.