Page 1 of 1


Posted: July 24th, 2018, 12:43 pm
by lefebvre
I am currently running multiple pollutants with several locations per pollutant via CONTROL+EMITIMES.

My typical run contains 100 pollutants with ~17 emissions for each pollutant.
Each location has a different emission, so I'm required to have 1700 locations for the EMITIMES format.
I need a vertical line emission, with a zero-emission separator because some of these overlap. This amounts to 3 emissions for each location.
So my ~1700 emissions turns into 100*1700*3=510,000 emissions.
As a result, I'm regularly running 5million NUMPAR and 50million MAXPAR.

Is it possible to use the PARINIT capability to simplify this workflow?

It looks like I can write an ASCII or binary PARINIT containing each location and the emission for all pollutants in the CONTROL file per location.
Should the emissions in the CONTROL file should be zero, since they are know accounted for in the PARINIT?
Do I need the vertical line source definition or can I use the PARDUMP SIGMA-U,SIGMA-V, SIGMA-X to define this pollutant's height?
The SETUP.CFG needs, NINIT=1 to initialize with PARINIT and PARINF needs to point to PARINIT.

I have tried creating a binary PARINIT for this, but HYSPLIT is segfaulting.
So I now seek assistance.
Thank you in advance for your help.


Posted: July 31st, 2018, 10:20 am
by lefebvre
I have resolved my issue.
It was two fold.
The first was big endian vs little endian.
The second was I misunderstood the number of pollutants on each particle.

I have this working now.


Posted: November 19th, 2018, 6:23 am
by Renee.Degutis
I also routinely use MATLAB, HYSPLIT and MELTS and it runs just fine. It will get hot at times, so invest in one of those stands you can put your laptop on with a fan maybe. I’m in the last two years of my PhD and use a MacBook Pro with 8GB RAM. I think what is most important is having high RAM capacity, so you can run these programs at the same time.

I'm having issues installing HYSPLIT. Do I need Ghostscript for Mac OS for research in STEM?

Renee Degutis


Posted: November 19th, 2018, 4:28 pm
by ariel.stein
No, Macs should be able to handle the poscript files.


Posted: August 1st, 2019, 7:24 am
by Smaug
lefebvre wrote:
July 24th, 2018, 12:43 pm
The SETUP.CFG needs, NINIT=1 to initialize with PARINIT and PARINF needs to point to PARINIT.

Is there any other requirments in the CONTROL file when using the particle dumps? My runs doesn't seems to be reading in the particle dump and I have set ninit =1 and pinpf = 'PARINIT' in the SETUP.CFG file. I beleive the time of the release in the CONTROL has to match the timestap in the binary but apart from I am not too sure what information has to match.



Posted: August 5th, 2019, 10:29 am
by barbara.stunder
There does not need to be a release given in the CONTROL file for the model to be initialized by the particles in the PARINIT file as long as the time of the particle position is during the run. See the description of NINIT at


Posted: August 7th, 2019, 2:46 pm
by Smaug
So when using this particle dump to start a simulation will it always be treated as a simultaneous release?

I believe the manual says when NINIT=1 the particle time stamp must match the model initialization time.


Posted: August 9th, 2019, 10:12 am
by alicec
The PARINIT file basically contains a list of computational particles with their relevant characteristics.
(Time, location, mass, etc.)

With NINIT=1, HYSPLIT opens the file at the beginning of the simulation and looks for particles with the same time as the run start. If it finds some, then it reads them in and initializes them with the information from the file. Then HYSPLIT closes the file and never opens it again.
It ignores any particles with times that do not match the simulation time (the beginning time in this case).

If NINIT=0, then HYSPLIT will never open the file.

If NINIT=2 then HYSPLIT will open the file at every hour (of the simulation) and look for particles with a time that matches the simulation time.
If it finds particles with a matching time, then it will initialize them and run them in the simulation from that point forward.