WAVEWATCH III v. 3.14 errata, problems and fixes
|
This page presents errata to the WAVEWATCH III
v. 3.14 manuals, and a list of identified problems and fixes. It will be updated
as needed. Please let us know if you find any errors or problems so
that we can add them to this page.
Errata
- The example ww3_prep.inp file on page 86-88 does not include various
new options. Use the new ww3_prep.inp file from the errata list below instead.
- ....
Index of Problems and Fixes
USTAR initialization in flux computations :
Problem: In the
flux computations in the subroutines W3FLX2 and W3FLX3 the friction
velocity UST is declared with INTENT(OUT), but not initialized. In the
main wave model code this is not an issue for most compilers, as the
friction velocity is defined before the subroutine call. Some strict
compilers, however, will initialize this as 0, resulting in run time
model errors. Furthermore, these variables were not properly initialized
in the output post-processors ww3_outp and gx_outp.
Fix: In
subroutines W3FLX2 in file w3flx2md.ftn and W3FLX3 in file w3flx3md.ftn,
replace the declarations
REAL, INTENT(OUT) :: UST, USTD, Z0, CD
with
REAL, INTENT(INOUT) :: UST
REAL, INTENT(OUT) :: USTD, Z0, CD
Fix: In the
post-processor gx_outp.ftn, on lines 825 and following, add
!/ST3 TAUWY = 0.
!/ST3 LLWS(:,:) = .TRUE.
USTAR = 1.
Fix: In the
post-processor ww3_outp.ftn, on lines 1195 and following, add
!/ST2 ZWND = ZWIND
!/ST3 ZWND = ZZWND
!/ST0 USTAR = 1.
!/ST1 USTAR = 1.
!/ST2 USTAR = 1.
Identified by Jesus Portilla, Posted April 20, 2010
Back to index.
Obstruction in mosaic grids :
Problem: When using
a mosaic of two-way nested grids, artificial obstructions may appear at
the boundaries of inner grids, when obstructions are defined between
grid points (FLAGTR = 1 or FLAGTR = 3 in ww3_grid.inp).
Fix: We have not
yet pursued a software fix for this. For now, use no obstructions
(FLAGTR = 0) of obstructions defined at the grid points instead (FLAGTR
= 2 or FLAGTR = 4, the latter is the NCEP operational default approach).
Identified by Chris Bunney, Posted May 31, 2010
Back to index.
Errors in ww3_prep.inp :
Problem: The
version of ww3_prep;inp distributed with the code and included in the
manual does not include several new options.
Fix: Use the
replacement ww3_prep.inp file available here.
May 31, 2010
Back to index.
Saving data in W3IOBC :
Problem: The
subroutine W3IOBC processes boundary data files. When reading data for
which the discrete spectral grid is different than the spectral grid for
the model, this routine converts the spectra to the spectral model
grid. This works well on the first reading of the file, but
corresponding data for the input spectra are not save, and hence
subsequent reading from the file fails.
Quick fix: *** THIS FIX IS TO BE USED WITH ww3_shel
OR WITH ww3_multi ONLY WHEN ALL BOUNDARY DATA FILES HAVE THE SAME SPECTRAL
RESOLUTION *** In subroutine W3IOBC in file w3iobcmd.ftn replace
lines 196 - 204
!/ ------------------------------------------------------------------- /
!/ Local parameters
!/
INTEGER :: IFILE, IERR, I, J, NKI, NTHI, IX, IY,&
ISEA, IP, ISP, NPTS, ISOUT, IS, IGRD
!/T1 INTEGER :: IK, ITH
!/S INTEGER, SAVE :: IENT = 0
REAL :: XFRI, FR1I, TH1I
!/T1 REAL :: HS, HS0
with
!/ ------------------------------------------------------------------- /
!/ Local parameters
!/
INTEGER, SAVE :: IFILE, IERR, I, J, NKI, NTHI, IX, IY,&
ISEA, IP, ISP, NPTS, ISOUT, IS, IGRD
!/T1 INTEGER :: IK, ITH
!/S INTEGER, SAVE :: IENT = 0
REAL, SAVE :: XFRI, FR1I, TH1I
!/T1 REAL :: HS, HS0
Full fix: Add relevant
parameters to be saved to the data structure in w3odatmd.ftn to be saved for
each separate grid individually. In file w3odatmd.ftn, make the following
additions (green text) around lines 335,
!/
TYPE OTYPE5
INTEGER :: NBI, NBI2, NFBPO, NBO(0:9), &
NBO2(0:9), NDSL(9), NKI, NTHI
!/MPI INTEGER :: NRQBP, NRQBP2
INTEGER, POINTER :: IPBPI(:,:), ISBPI(:), &
IPBPO(:,:), ISBPO(:)
!/MPI INTEGER, POINTER :: IRQBP1(:), IRQBP2(:)
REAL :: XFRI, FR1I, TH1I
REAL, POINTER :: XBPI(:), YBPI(:), RDBPI(:,:), &
XBPO(:), YBPO(:), RDBPO(:,:), &
ABPI0(:,:), ABPIN(:,:), ABPOS(:,:), &
around line 427,
!/
TYPE OTYPE5
INTEGER :: NBI, NBI2, NFBPO, NBO(0:9), &
NBO2(0:9), NDSL(9), NKI, NTHI
!/MPI INTEGER :: NRQBP, NRQBP2
INTEGER, POINTER :: IPBPI(:,:), ISBPI(:), &
IPBPO(:,:), ISBPO(:)
!/MPI INTEGER, POINTER :: IRQBP1(:), IRQBP2(:)
REAL :: XFRI, FR1I, TH1I
REAL, POINTER :: XBPI(:), YBPI(:), RDBPI(:,:), &
XBPO(:), YBPO(:), RDBPO(:,:), &
ABPI0(:,:), ABPIN(:,:), ABPOS(:,:), &
and around line 1330
NBO => OUTPTS(IMOD)%OUT5%NBO
NBO2 => OUTPTS(IMOD)%OUT5%NBO2
NDSL => OUTPTS(IMOD)%OUT5%NDSL
NKI => OUTPTS(IMOD)%OUT5%NKI
NTHI => OUTPTS(IMOD)%OUT5%NTHI
XFRI => OUTPTS(IMOD)%OUT5%XFRI
FR1I => OUTPTS(IMOD)%OUT5%FR1I
TH1I => OUTPTS(IMOD)%OUT5%TH1I
FLBPI => OUTPTS(IMOD)%OUT5%FLBPI
FLBPO => OUTPTS(IMOD)%OUT5%FLBPO
FILER => OUTPTS(IMOD)%OUT5%FILER
Furthermore, in the file w3iobcmd,ftn, around line 170 add
USE W3ADATMD, ONLY: CG
USE W3ODATMD, ONLY: NDSE, NDST, IAPROC, NAPROC, NAPERR, NAPBPT, &
NBI, NBI2, NFBPO, NBO, NBO2, NDSL, &
NKI, NTHI, XFRI, FR1I, TH1I, &
IPBPI, ISBPI, XBPI, YBPI, RDBPI, &
IPBPO, ISBPO, XBPO, YBPO, RDBPO, &
ABPI0, ABPIN, ABPOS, FLBPI, FILER, FILEW, &
and around line 190 remove (red text)
!/ ------------------------------------------------------------------- /
!/ Local parameters
!/
INTEGER :: IFILE, IERR, I, J, NKI, NTHI, IX, IY,&
ISEA, IP, ISP, NPTS, ISOUT, IS, IGRD
!/T1 INTEGER :: IK, ITH
!/S INTEGER, SAVE :: IENT = 0
REAL :: XFRI, FR1I, TH1I
!/T1 REAL :: HS, HS0
REAL, ALLOCATABLE :: TMPSPC(:,:)
LOGICAL :: FLOK
Identified by Sander Hulst.
Quick fix posted May 31, 2010
Full fix posted July 29. 2010
Back to index.
Declaration error for restart file under MPI :
Problem:
A declaration error was found in the auxiliary array VAAUX used to gather
all spectra into a single processor when writing the restart file under MPI.
On some systems, particularly with different word lengths for integers and
reals, this will break the restart file.
Fix:
In file w3odatmd.ftn replace lines around 329 and line 422. This needs
to be done twice, although at the second location the original line is
formatted a little different.
!/MPI INTEGER, POINTER :: IRQRS(:), IRQRSS(:), VAAUX(:,:,:)
with
!/MPI INTEGER, POINTER :: IRQRS(:), IRQRSS(:)
!/MPI REAL, POINTER :: VAAUX(:,:,:)
Identified by Yalin Fan, Posted June 14, 2010
Back to index.
Freezing of wind fields and wave height noise :
Problem:
The operational web sites for NOAA WAVEWATCH III have been featuring frozen
wind fields and associated noise in wave height fields in the global
multi-grid wave model starting around March 2010 and severely aggravated
as of July 27, 2010. This operational model uses a modified version
of ww3_multi.ftn that allows for computations to progress while wind files
are incrementally updated, i.e., this model can thus run side-by-side with
the weather model forcing it in a precursor to full dynamic coupling of the
weather and wave model. The spurious features in this wave model were
traced back to issues with the file system while dynamically updating the
input wind files for the wave model. This problem was solved for the Aug. 17
12z model cycle by adding various tests to the scripting and model driver
ww3_multi.ftn for this model.
Fix:
No fix needed for public distribution version of WAVEWATCH III, because the
corresponding version of the wave model driver ww3_multi is not publicly
distributed.
Identified by many, Posted Aug. 17, 2010
Back to index.
Initialization of spectra for processing of boundary data :
Problem:
For some applications of the ww3_multi.ftn wave model driver and in some
hardware/compiler combinations the model crashes in the flux computation
routines W3FLX2 or W3FLX3. This problem was traced back to NaNs being
introduced in the spectrum at boundary points of nested grids. The NaNs
were introduced due to lack of initialization of arrays, but on most
systems do no materialize as the non-initialized arrays are multiplied by
0 on first usage.
Fix:
In file w3updtmd.ftn add the following lines in subroutine W3UBPT around
line 802
!
BBPI0(:,0) = 0.
BBPIN(:,0) = 0.
ABPI0(:,0) = 0.
ABPIN(:,0) = 0.
!
DO IBI=1, NBI
ISEA = ISBPI(IBI)
Identified by several, Posted Aug. 18, 2010
Back to index.
Managing MAPSTA in inundation grids :
Problem:
When using grids with inundation capabilities in the multi-grid model driver
ww3_multi, the status map MAPSTA is not properly processed in the wave model
routine WMWAVE in wmwavemd,ftn, resulting in dry grid point accidentally
being switched back on with zero water depth. For some compilers, this will
lead to NaNs in the refraction routines and subsequent model failure,
typically somewhere in the source term or flux computations.
Fix:
In file wmwavemd.ftn add the following line around line 990 in section
9.b of subroutine WMWAVE. Beware that this line is already present several
lines earlier before the call to subroutine W3WAVE.
IF ( FLGHG1 .AND. .NOT.FLGHG2 .AND. &
GRDHGH(I,0).GT.0 ) THEN
MAPST2 = MAPST2 + 8*MAPMSK
MAPSTA = ABS(MAPSTA)
DO IX=1, NX
DO IY=1, NY
IF ( MAPST2(IY,IX) .GT. 0 ) &
Posted Aug. 27, 2010
Back to index.
Possible issue with depth treatment in W3SRCE :
Possible problem:
In this routine, the depth is allowed to go to 0, whereas in the propagation
routines and all other computations, the depth is not allowed to be smaller
than DMIN. Although we have not had any reports of this giving problems with
model runs, we suggest the following preventive correction to the code.
Fix:
In the subroutine W3SRCE in the file w3srcemd.ftn apply the following
changes. In the header of the routine, replace
SUBROUTINE W3SRCE ( IX, IY, IMOD, SPEC, ALPHA, WN1, CG1, DEPTH, &
with
SUBROUTINE W3SRCE ( IX, IY, IMOD, SPEC, ALPHA, WN1, CG1, D_INP, &
In the use statement for the grid data structure replace
USE W3GDATMD, ONLY: NK, NTH, NSPEC, SIG, TH, DTMAX, DTMIN, &
FACTI1, FACTI2, FACSD, FACHFA, FACP, &
with
USE W3GDATMD, ONLY: NK, NTH, NSPEC, SIG, TH, DMIN, DTMAX, &
DTMIN, FACTI1, FACTI2, FACSD, FACHFA, FACP, &
In the declaration of the parameter list, replace
REAL, INTENT(IN) :: WN1(NK), CG1(NK), DEPTH, U10ABS, &
with
REAL, INTENT(IN) :: WN1(NK), CG1(NK), D_INP, U10ABS, &
In the declaration of variables internal to the subroutine add
REAL :: DTTOT, FHIGH, DT, AFILT, DAMAX, AFAC,&
HDT, ZWND, FP, DEPTH
Finally, around line 367, add the lines
!
!/T FLTEST = .TRUE.
!
DEPTH = MAX ( DMIN , D_INP )
!
!/LN0 VSLN = 0.
!/SEED VSLN = 0.
!/ST0 VSIN = 0.
Posted Aug. 27, 2010
Back to index.
Bug in combining wind seas in the partitioning routine :
Problem: In w3partmd.f90, the subroutine PTMEAN returns NP, the number of the NP_MAX partitions
for which HS >= HSPMIN, but it doesn't return information about *which* partitions were selected.
The code that combines wind-seas assumes that the NP partitions selected are the *first* NP of
the NP_MAX partitions, which is not necessarily the case.
Fix:
- In subroutine W3PART, declare a new local integer array:
INTEGER :: PMAP(DIMXP)
- Modify calls to PTMEAN from
CALL PTMEAN(NP_MAX,IMO,ZP,DEPTH,UABS,UDIR,WN,NP,XP,DIMXP)
to
CALL PTMEAN(NP_MAX,IMO,ZP,DEPTH,UABS,UDIR,WN,NP,XP,DIMXP,PMAP)
- Replace lines 236-242:
IPW = INDEX(1)
DO IP=2, NWS
IPT = INDEX(IP)
DO ISP=1, NSPEC
IF ( IMO(ISP) .EQ. IPT ) IMO(ISP) = IPW
END DO
END DO
with
IPW = PMAP(INDEX(1))
DO IP=2, NWS
IPT = PMAP(INDEX(IP))
DO ISP=1, NSPEC
IF ( IMO(ISP) .EQ. IPT ) IMO(ISP) = IPW
END DO
END DO
- In subroutine PTMEAN, add an OUT parameter PMAP as above:
INTEGER, INTENT(OUT) :: PMAP(DIMXP)
- and modify lines 1064-1065:
NPO = NPO + 1
XP(1,NPO) = HS
to
NPO = NPO + 1
IF (IP.GT.0) PMAP(NPO) = IP
XP(1,NPO) = HS
Identified by Mark Szyszka
Posted December 2, 2010
Back to index.
Bug: Out of bounds error message for point data :
Problem: When point output for the global grid (grid that wraps around)
lies between NX and 1, we get an out of bounds error in the initialization
routine.
Fix: On line 313 in wmiopomd.ftn add the following line
IF ( GLOBAL .AND. IX1 .EQ. NX ) IXS = 1-NX
Posted January 17, 2011
Back to index.
|
www.nws.noaa.gov 
