Opened 9 years ago

Closed 9 years ago

#120 closed Support (fixed)

Simulated particles are fewer then requested in FLEX-WRF

Reported by: hbarbosa Owned by: jbrioude
Priority: minor Milestone: FLEXPART_WRF_3.2_FPbase_9
Component: FP other Version: FLEXPART-WRF
Keywords: Cc:

Description

I'm using the latest version of flexwrf (beta v3.2) to do forward simulations during 24hs. When I request 100000 particles to be released during these 24hs, only 62921 are actually sent out.

You can see this in the attached file flexwrf.output, and I copy a small part below. After 56400 seconds (16hs) the total number of particles stops increasing.

51600 SECONDS SIMULATED: 59756 PARTICLES: Uncertainty: 0.000 0.000 0.000
52800 SECONDS SIMULATED: 61142 PARTICLES: Uncertainty: 0.000 0.000 0.000
54000 SECONDS SIMULATED: 62302 PARTICLES: Uncertainty: 0.000 0.000 0.000
55200 SECONDS SIMULATED: 62901 PARTICLES: Uncertainty: 0.000 0.000 0.000
56400 SECONDS SIMULATED: 62921 PARTICLES: Uncertainty: 0.000 0.000 0.000
57600 SECONDS SIMULATED: 62921 PARTICLES: Uncertainty: 0.000 0.000 0.000
58800 SECONDS SIMULATED: 62921 PARTICLES: Uncertainty: 0.000 0.000 0.000
60000 SECONDS SIMULATED: 62921 PARTICLES: Uncertainty: 0.000 0.000 0.000
61200 SECONDS SIMULATED: 62921 PARTICLES: Uncertainty: 0.000 0.000 0.000

If I ask for 50k particles, it also stops after 16hs but with 31443 particles simulated.

And if I ask for 30k particles, it also stops after 16hs now at 18898 simulated particles.

And if I ask for 20k particles, the programs crashes with a SIGSEGV segmentation fault!!!

It should be straight forward to reproduce this error.

Attachments (2)

flexwrf.output (9.0 KB) - added by hbarbosa 9 years ago.
text output of flexwrf written to the screen
goamazon.forward (7.7 KB) - added by hbarbosa 9 years ago.
my flexwrf input file

Download all attachments as: .zip

Change History (7)

Changed 9 years ago by hbarbosa

text output of flexwrf written to the screen

Changed 9 years ago by hbarbosa

my flexwrf input file

comment:1 Changed 9 years ago by jbrioude

It is not going to be straightforward without your wrf output.
Do you have the same errors in serial mode and OMP mode?

comment:2 Changed 9 years ago by hbarbosa

Hi jbrioude,

The same error happens with

  • serial mode
  • 1 node (MPI) and 4 processors (openmp)
  • 2 nodes (MPI) and 4 processors openMP in each node

What is more weird is that it stops increasing the counter at different points: 12563, 12585, and 12607 respectively.

My input file and input data is available at:

ftp://lfa.if.usp.br/Temp/iop1_data

comment:3 Changed 9 years ago by jbrioude

  • Owner changed from somebody to jbrioude
  • Status changed from new to assigned

One possibility is that the trajectories are leaving the limited WRF domain.
Do you see trajectories leaving the domain in your FLEXPART output?
how far is the release location from a WRF domain boundary?

comment:4 Changed 9 years ago by hbarbosa

Hi JBrioude,

The particles definitely leave the domain. You can see an mp4 movie of my simulation here:

ftp://lfa.if.usp.br/Temp/png_vint.mp4

However, that number on the screen should be the number of particles released so far, right? And not the number of particles inside your domain at that particular time, right?

Thanks,
Henrique

comment:5 Changed 9 years ago by jbrioude

  • Priority changed from major to minor
  • Resolution set to fixed
  • Status changed from assigned to closed
  • Type changed from Defect to Support

No, the number of trajectories shown in the FLEXPART output is the number of trajectories used in the domain at different time of the simulation, not released from the beginning of the simulation.

Note: See TracTickets for help on using tickets.
hosted by ZAMG