Proper way to control particle numbers released

Topics about the HYSPLIT dispersion model.
Post Reply
dlu
Posts: 6
Joined: September 3rd, 2021, 5:43 pm
Registered HYSPLIT User: Yes

Proper way to control particle numbers released

Post by dlu »

I'm learning how to use the dispersion model and want to understand how to release an exact number of particles from a starting location. I am testing the starting location: (lat 31.0069, lon -88.0103, hgt 182.88). For a given day at hour 00, my goal is to follow n particles over the next 6 hours. For example, I'd like to release 2 particles at hour 00 then see where they are at hours 1, 2, 3, 4, 5, and 6. I tried two methods of accomplishing this, but got different results with each.

Could someone please explain how to properly control the number of particles released?

Method 1
Based on the tutorial (https://www.ready.noaa.gov/documents/Tu ... _base.html), I assumed that to release 2 particles, the emissions rate should be 100 and the hours of emission should be 0.02 in the CONTROL file. Then in the SETUP.CFG file, the numpar = 2 and maxpar = 10000 (the default). I was able to get 2 particles every 6 hours that I could track.

Method 2
Then, I tried changing the settings to emission rate = 1, hours of emission = 1, numpar = 2, maxpar = 2. I also was able to get 2 particles every 6 hours that I could track. (As a side note, when I set it to emission rate = 1, hours of emission = 1, numpar = 2, maxpar = 10000, I ended up with 60 particles every 6 hours instead. I'd like to also understand why this is the case).

Method 1 Results
particle_num lat lng height hour
0 00001 -88.0035 31.0446 179.0 001
1 00002 -88.0061 31.0442 166.0 001
2 00001 -88.0060 31.0968 167.0 002
3 00002 -88.0104 31.0959 160.0 002
4 00001 -88.0148 31.1638 162.0 003
5 00002 -88.0179 31.1631 166.0 003
6 00001 -88.0328 31.2448 144.0 004
7 00002 -88.0326 31.2448 174.0 004
8 00001 -88.0660 31.3371 116.0 005
9 00002 -88.0526 31.3410 171.0 005
10 00001 -88.1180 31.4403 83.0 006
11 00002 -88.0803 31.4513 166.0 006

Method 2 Results
particle_num lat lng height hour
0 00001 -88.0060 31.0443 166.0 001
1 00002 -88.0037 31.0441 179.0 001
2 00001 -88.0143 31.0953 133.0 002
3 00002 -88.0043 31.0965 169.0 002
4 00001 -88.0318 31.1607 145.0 003
5 00002 -88.0083 31.1639 179.0 003
6 00001 -88.0531 31.2410 153.0 004
7 00002 -88.0214 31.2452 167.0 004
8 00001 -88.0809 31.3368 124.0 005
9 00002 -88.0488 31.3387 130.0 005
10 00001 -88.1263 31.4440 107.0 006
11 00002 -88.0915 31.4440 117.0 006

I've attached my CONTROL and SETUP.CFG files to this post. The only difference between the control and setup files between the two methods are what I flagged above. Everything else is the same.
Attachments
SETUP_e1h1.txt
(604 Bytes) Downloaded 208 times
CONTROL_e1h1.txt
(261 Bytes) Downloaded 217 times
alicec
Posts: 411
Joined: February 8th, 2016, 12:56 pm
Registered HYSPLIT User: Yes

Re: Proper way to control particle numbers released

Post by alicec »

numpar which is number of particle's released per cycle is not strictly enforced by HYSPLIT.
However maxpar is strictly enforced.

HYSPLIT will try to interpret numpar in a way that makes sense with your other inputs.

So for instance when you set

numpar =2
hours of emission = 1

HYSPLIT 'knows' that it runs out of particles to release in two time steps.
When it gets to the third time step, it releases another particle anyways as long as maxpar is not exceeded.

When you set maxpar=10,000, then HYSPLIT will continue to release particles every time step throughout the hour
even if this ends up being larger than numpar.
dlu
Posts: 6
Joined: September 3rd, 2021, 5:43 pm
Registered HYSPLIT User: Yes

Re: Proper way to control particle numbers released

Post by dlu »

Hi, thank you for the response. I didn't receive a notification so I wasn't able to see it until now.

Could you please elaborate on what the time steps throughout the hour means? How long is each time step, and how many particles is it assuming I'm releasing per time step? I assumed that if I were to set numpar=2 and hours of emissions=1 that over the course of 1 hour only 2 particles are released.

Based on your reply, it sounds like if I want a precise amount of particles released, I should be using limiting maxpar. Is that correct?
alicec
Posts: 411
Joined: February 8th, 2016, 12:56 pm
Registered HYSPLIT User: Yes

Re: Proper way to control particle numbers released

Post by alicec »

See here for an explanation of the integration time step.
https://www.ready.noaa.gov/documents/Tu ... _eqns.html

The default is for this time step to be variable as described with a minimum of 1 minute.
If using the default variable time step, you can see what time step was used by looking in the MESSAGE file.

You can also set the timestep by setting DELT in the SETUP.CFG file.
https://www.ready.noaa.gov/hysplitusersguide/S610.htm
dlu
Posts: 6
Joined: September 3rd, 2021, 5:43 pm
Registered HYSPLIT User: Yes

Re: Proper way to control particle numbers released

Post by dlu »

Thank you for the links. After reading through them, it sounds like DELT should depend on the meteorological conditions: "The fixed time step option should only be used when strong winds in regions not relevant to the dispersion simulation (perhaps the upper troposphere) are causing the model to run with small time steps. Improper specification of the time step could cause aliasing errors in advection and underestimation of air concentrations. An alternate approach is to set the a negative value which will result in that value (abs) being used as the minimum time step."

If I don't have a sense of whether the strong winds are affection my dispersion simulation, it seems like the best choice for me to make is to leave this parameter in its default value of 0. In that case, if I wanted to get precisely n particles, then it sounds like I should be setting numpar = n and maxpar = n. If I set my hours of emission to be 1, then should the emission rate be n as well? I'm not very familiar with HYSPLIT and have had a hard time understanding these fields after reading the documentation, so would really appreciate any pointers you have.

Additionally, do you have a recommendation on the number of particles that should be released to get an accurate simulation? I know the default of numpar = 2500 and maxpar = 10000, but are these the minimum particles recommended?
alicec
Posts: 411
Joined: February 8th, 2016, 12:56 pm
Registered HYSPLIT User: Yes

Re: Proper way to control particle numbers released

Post by alicec »

viewtopic.php?f=3&t=1583&p=4581&hilit=numpar#p4581

The emission rate determines how much mass is released while numpar determines how many computational particles are released.
For example suppose you define an emission rate of 1000 (kg/hour) and you release 1000 computational particles in an hour. Then each computational particle has a mass of 100(kg/hour) / 100(particles/hour) = 1 kg/particle.

You can use the MESSAGE file to check what you are getting. Here is an example in which the emission rate was set to 1(unit mass/hour). Numpar was set to 100 particles/hour. And below is part of a MESSAGE file.

The second to last column (in bold and underlined) shows the number of computational particles. Here, you see the initial time step is 15 minutes. Each 15 minute time step, 25 computational particles are emitted until after an hour, you have 100 computational particles in the simulation.
The last column (number after the bold number) shows the total mass being carried in the simulation.
The total amount of mass emitted after an hour is 1 (or close enough). So each computational particle has 1(unit mass)/100 (particles) = 0.01 unit mass/particle.

NOTICE advrng: (kg ,xyr,xy1) - 1 10 10 147 52
NOTICE main: Initial time step (min) 15
NOTICE emspnt: emissions started
NOTICE main: 1 59620335 25 0.25000003
NOTICE main: 1 59620350 50 0.49999979
NOTICE main: 1 59620365 75 0.74999958
NOTICE main: 1 59620380 100 0.99999934
NOTICE parout: 13 5 11 1 0 Num= 100




A good rule of thumb is that the concentrations you need to resolve
should be represented by at least 10 computational particles in a grid box per time step. See section 2.4 for a more in depth explanation.
https://www.mdpi.com/2073-4433/11/12/1369/htm

You can also run the model a few times with different random seeds. (SEED in SETUP.CFG)
https://www.ready.noaa.gov/hysplitusersguide/S410.htm
The amount of variation you see in your results from using the different random seeds indicates how much
uncertainty is due to particle number. You can increase the particle number until the uncertainty is acceptable.


Hope this is helpful.
Post Reply

Return to “Dispersion Model”