Welcome, Guest
Username: Password: Remember me
General topics
  • Page:
  • 1
  • 2

TOPIC: Spectral grid

Spectral grid 3 years 11 months ago #1487

  • Alvaro Subias
  • Alvaro Subias's Avatar
  • OFFLINE
  • Fresh Boarder
  • Posts: 13
  • Thank you received: 2
lisa, thank you very much for your answer. But i have the following error :( does anybody have any idea what happens


cca-login1:/scratch/ms/es/sp1p/hm_home/AIB_40h11b5cubic/bin$ ./ECTO namlist infileAIB_40h11b5cubic.fa
*******************************
* PROGRAM : ECTO
*******************************
NAMELIST FILE namlist OPENED
[Wed Feb 17 14:31:34 2016] [unknown] Fatal error in PMPI_Init_thread: Other MPI error, error stack:
MPIR_Init_thread(449):
MPID_Init(169).......: channel initialization failed
MPID_Init(547).......: PMI2 init failed: 1

Spectral grid 3 years 11 months ago #1488

  • Lisa Bengtsson
  • Lisa Bengtsson's Avatar
  • OFFLINE
  • Administrator
  • Posts: 60
  • Thank you received: 5
Hm, maybe it matters if it is compiled with MPI or not...

Did you use an ALTO from cycle 40? I can get it to work with a cycle 38h1.2 binary. If you have one from cycle 38h1.2 maybe you can try that as a first step?

Spectral grid 3 years 10 months ago #1496

  • Alvaro Subias
  • Alvaro Subias's Avatar
  • OFFLINE
  • Fresh Boarder
  • Posts: 13
  • Thank you received: 2
Dear Lisa

thanks for the info about ECTO, i run it at 38h12. I have run 10 days of July in 2013 comparing

linear step 60
quadratic step 90
cubic step 120
cubic step 60

the scores are similar except for u10m where differences appears
V_00000000_00000000_ALL_ALL_FF_0.1.png


I have run quadratic and cubic experiments in December of 2014, the cubic120 explodes after some days of integration because there appears underground semi-lagrangian trajectories, this kind of messages does not appear in the cubic90 case. Can in control this with some kind of diffusion?

i would like to update hirlam.org/portal/validation/40h1/aemet_40h1.html, how could i upload a directory tree with a verification?

best regards
Alvaro

Spectral grid 3 years 10 months ago #1498

  • Alvaro Subias
  • Alvaro Subias's Avatar
  • OFFLINE
  • Fresh Boarder
  • Posts: 13
  • Thank you received: 2
activating the semi-lagrangian horizontal diffusion for vertical divergence LSLHD_SVD=true reduces the number of SMILAG TRAJECTORY UNDERGROUND messages in the cubic grid experiment with tstep=120 but not dissappear at all, does anybody has an idea to this respect?
  • Page:
  • 1
  • 2
Time to create page: 0.077 seconds