Opened 3 months ago

Closed 2 months ago

#227 closed Defect (fixed)

Retrieve global ERA5 data

Reported by: ahilboll Owned by: anphi
Priority: major Milestone:
Component: flex_extract Version: flex_extract_v7.0.4
Keywords: Cc:

Description

I'm having trouble retrieving hemispheric (i.e., northern hemisphere) ERA5 data using flex_extract version 7.0.4. I'm using the GATEWAY mode, i.e., I have flex_extract installed on my local system to submit jobs to the ECGATE queue.

Installation works fine, and I can retrieve non-hemispheric ERA5 data just fine. E.g., this COMMAND file works fine:

DAY1
DAY2
DTIME 1
TYPE AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN
TIME 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23
STEP 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
ACCTYPE FC
ACCTIME 06/18
ACCMAXSTEP 12
M_CLASS EA
M_STREAM OPER
M_NUMBER OFF
M_EXPVER 1
M_GRID 250  
M_LEFT -170000
M_LOWER 10000
M_UPPER 80000
M_RIGHT 170000
M_LEVELIST 1/to/137
M_RESOL 799
M_GAUSS 0
M_ACCURACY 24
M_OMEGA 0
M_OMEGADIFF 0
M_ETA 1
M_ETADIFF 0
M_DPDETA 1
M_SMOOTH 0
M_FORMAT GRIB2
M_ADDPAR /186/187/188/235/139/39
PREFIX EA
ECSTORAGE 0
ECTRANS 1
ECFSDIR ectmp:/${USER}/econdemand/
MAILOPS hilboll@uni-bremen.de
MAILFAIL hilboll@uni-bremen.de
EOF

However, this here does *not* work:

DAY1
DAY2
DTIME 1
TYPE AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN
TIME 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23
STEP 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
ACCTYPE FC
ACCTIME 06/18
ACCMAXSTEP 12
M_CLASS EA
M_STREAM OPER
M_NUMBER OFF
M_EXPVER 1
M_GRID 250  
M_LEFT -180000
M_LOWER 00000
M_UPPER 90000
M_RIGHT 180000
M_LEVELIST 1/to/137
M_RESOL 799
M_GAUSS 0
M_ACCURACY 24
M_OMEGA 0
M_OMEGADIFF 0
M_ETA 1
M_ETADIFF 0
M_DPDETA 1
M_SMOOTH 0
M_FORMAT GRIB2
M_ADDPAR /186/187/188/235/139/39
PREFIX EA
ECSTORAGE 0
ECTRANS 1
ECFSDIR ectmp:/${USER}/econdemand/
MAILOPS hilboll@uni-bremen.de
MAILFAIL hilboll@uni-bremen.de
EOF

I tried already different left/right combinations, like -180/180, -179750/180, -180/179750, -179875/179875, and the like.

The error message I'm getting is this:

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0  0x3dd543250f in ???
#1  0x3dd54896f6 in ???
#2  0x40a1c5 in __rwgrib2_MOD_readlatlon
	at /home/ms/datex/gdmv/flexpart/scripts_gateway/flex_extract_v7.0.4/src/rwGRIB2.f90:72
#3  0x41240b in preconvert
	at /home/ms/datex/gdmv/flexpart/scripts_gateway/flex_extract_v7.0.4/src/preconvert.f90:402
#4  0x418c6d in main
	at /home/ms/datex/gdmv/flexpart/scripts_gateway/flex_extract_v7.0.4/src/preconvert.f90:53

Change History (6)

comment:1 Changed 3 months ago by anphi

  • Owner set to anphi
  • Status changed from new to accepted

comment:2 Changed 3 months ago by anphi

  • Component changed from FP other to flex_extract

comment:3 Changed 3 months ago by anphi

Hi,

interesting case!

In a first very quick test, I used my current dev version with

grid=0.25/0.25 
area=2.0/-179.75/0.0/180.0 

and this worked absolutely fine.

In a second test I used version 7.0.4 with a smaller domain as you did:

DTIME 3
TYPE AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN
TIME 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23
STEP 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
M_GRID 250  
M_LEFT -179750
M_LOWER 00000
M_UPPER 2000
M_RIGHT 180000
M_LEVELIST 130/to/137

And this also worked for me except for the following notes from the CONVERT2 program:

Note: The following floating-point exceptions are signalling: IEEE_UNDERFLOW_FLAG IEEE_DENORMAL

For the third test I used the above setup for version 7.0.4 but with the domain

M_LEFT -179750
M_LOWER 00000
M_UPPER 90000
M_RIGHT 180000

and this worked as well!

So, could you please check if you did not have a typo in the area and can you copy/check a list of the modules you used on ecgate ?!

The modules my version 7.0.4 currently uses is:

  1) version/3.2.10(default)
  2) verbose/true(default)
  3) mode/64(default)
  4) gnu/6.3.0(default)
  5) slurm/17.02.2(default)
  6) jasper/1.900.1(default)
  7) eccodes/2.9.2(nov18:default)
  8) ecfs/2.2.5(nov18:test:new:prodn:default)
  9) python/2.7.12-01(may18:new:default)
 10) perl/5.16.3-01(default)
 11) ecaccess/4.0.2(default)
 12) sms/4.4.13(default)
 13) metview/5.4.1(new:default)
 14) ecmwf/1.0(default)
 15) fftw/3.3.4(default)
 16) emos/437-r64

My dev version uses:

  1) version/3.2.10(default)
  2) verbose/true(default)
  3) mode/64(default)
  4) gnu/6.3.0(default)
  5) slurm/17.02.2(default)
  6) ecfs/2.2.5(nov18:test:new:prodn:default)
  7) python/2.7.12-01(may18:new:default)
  8) perl/5.16.3-01(default)
  9) ecaccess/4.0.2(default)
 10) sms/4.4.13(default)
 11) metview/5.4.1(new:default)
 12) ecmwf/1.0(default)
 13) jasper/1.900.1(default)
 14) eccodes/2.9.2(nov18:default)
 15) fftw/3.3.4(default)
 16) emos/455-r64

comment:4 Changed 2 months ago by ahilboll

Thanks for testing this, Anne! This is indeed curious ... yesterday, I had no troubles retrieving the data for the whole northern hemisphere. This is (part of) the CONTROL that works:

TYPE AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN AN
TIME 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23
STEP 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
M_GRID 250
M_LEFT -179750
M_LOWER 00000
M_UPPER 90000
M_RIGHT 180000
M_LEVELIST 50/to/137

I'm running this in the *gateway* mode. Everything seems to be fine now, thanks for your help, and sorry for the noise!

comment:5 Changed 2 months ago by anphi

No problem! You' re welcome.
You should keep reporting any problem you have.
This helps a lot.

Since it seems to work now, I am going to close the ticket.

comment:6 Changed 2 months ago by anphi

  • Resolution set to fixed
  • Status changed from accepted to closed
Note: See TracTickets for help on using tickets.
hosted by ZAMG