Hi Elodie,
This looks like a problem with skipping days at the change over at
the end of the year which would occur if the file is created by
sampling some other dataset on the same day of the year. The fact
that your dates are exactly a multiple of 8 days of days into the
year make me think there is a problem with the sampling strategy.
! Days from start of year
yes? list `days1900(1997,9,6)-days1900(1997,1,1)`
!-> list 248
VARIABLE : constant
248.0
yes? list `days1900(1998,12,27)-days1900(1998,1,1)`
!-> list 360
VARIABLE : constant
360.0
! Difference between date should be 10587-10110 = 477
yes? list `days1900(1998,12,27)-days1900(1997,9,6)`
!-> list 477
VARIABLE : constant
477.0
I bet that you will find a discontinuity in the time step either
at the start of 1998.
plot tbox[gt=chl]
should do the trick.
If you break the netcdf file at the discontinuity then a tool like
mkdes should create a correct descriptor file.
Cheers,
Russ
On 28/03/14 08:53, Elodie Martinez wrote:
Dear Ferreters,
I have data that spans multiple files. I have used many times mc
files without any probles till now.
The data are every 8days so the s_delta =8.
However, my mc file only works when I enter s_delta=1. (which is
wrong of course because it should be 8 days)
when s_delta is something else than 1, I have the following
message error:
At line 228 of file mc_init_dset.F
Fortran runtime error: End of file
My simplified code to read only 1 file at the moment is below:
&FORMAT_RECORD
D_TYPE = ' MC',
D_FORMAT = ' 1A',
D_SOURCE_CLASS = 'multi satellite OUTPUT',
/
&BACKGROUND_RECORD
D_EXPNUM = '0001',
D_TITLE = 'Globcolour Chl',
D_T0TIME = '1-Jan-1970 12:00:00',
D_TIME_UNIT = 86400.0,
D_TIME_MODULO = .FALSE.,
/
&MESSAGE_RECORD
D_MESSAGE = ' ',
D_ALERT_ON_OPEN = F,
D_ALERT_ON_OUTPUT = F,
/
&EXTRA_RECORD
/
&STEPFILE_RECORD
s_filename = 'PolF-1997-1998_Glob_Wei_4km-8d.nc',
S_START = 10110,
S_END = 10587,
S_DELTA = 8.0,
S_NUM_OF_FILES = 1,
S_REGVARFLAG = ' ',
/
&STEPFILE_RECORD
s_filename = '**END OF STEPFILES**'
/
When I simply load my initial file I have the following
informations :
./PolF-1997-1998_Glob_Wei_4km-8d.nc (default)
name title I J K L M N
CHL
1:720 1:840 ... 1:61 ... ...
units on
grid GAM1 with -999 for missing data
X=160W(-160):130W(-130) Y=30S:5N
time range:
06-SEP-1997 00:00 to 27-DEC-1998 00:00
and if I list t[gt=chl]:
VARIABLE
: T
axis TIME2
FILENAME :
PolF-1997-1998_Glob_Wei_4km-8d.nc
SUBSET : 61 points (TIME)
06-SEP-1997 00 / 1: 10110.
14-SEP-1997 00 / 2: 10118.
22-SEP-1997 00 / 3: 10126.
.......
11-DEC-1998 00 / 59: 10571.
19-DEC-1998 00 / 60: 10579.
27-DEC-1998 00 / 61: 10587.
I am wondering if it could come my time axis which has values
about 10000. (some other files I read with mc files too, and also
starting in september 1997, have time values around 730000 )
Has any one already encounter this problem?
thanks a lot for your help!!!!
Elodie
|