2
$\begingroup$

For example, I am interested in making a simulation for some chemical species for 2019. I am planning to provide NCEP/GDAS FNL meteorological data at 0.25 degree and CAM-Chem chemical boundary layer. I want to know what time period of meteorological data and chemical boundary layer should I provide for my simulation?

$\endgroup$
9
  • 3
    $\begingroup$ Depending on the initial conditions you use and the meteorological event involved, I've found it takes at least 2-3 days of spin-up for the chemical conditions to stabilize in regional CTMs. $\endgroup$
    – f.thorpe
    Jan 10 at 1:39
  • $\begingroup$ I am trying to do an year-long simulation. I do not think 2-3 days of spin-up would be enough for the chemical conditions to stabilize in regional CTMs. What I am confused about is that if I am trying to do a simulation for 2019, do I have to provide meteorological data and chemical boundary layer for 2019 and/or let us say, December 2018? $\endgroup$ Jan 10 at 7:55
  • 3
    $\begingroup$ Meteorology usually takes longer to spin up than chemistry. Once your chemical boundary conditions wash over your entire domain... it's spun. If winds are stagnant it could take more than 3 days but it doesn't really matter how long your simulation is. Since you are using WRF-CHEM and they are coupled, maybe you take 5 days (e.g. start at Dec 27) but that's more for the meteorology than the chemistry. $\endgroup$
    – f.thorpe
    Jan 10 at 8:06
  • 1
    $\begingroup$ If you are looking at typical air quality products like ozone and aerosols, etc. then the chemistry is quick and transport dominates. If you are modeling GHGs and other long-lived pollutants, then you would need a much longer chemical spin-up (or really good initial conditions). If you have a large domain (e.g. continental) you might need more days so the transport can flush out the initial conditions from the entire domain. forum.mmm.ucar.edu/phpBB3/viewtopic.php?t=9894 wiki.harvard.edu/confluence/pages/… $\endgroup$
    – f.thorpe
    Jan 10 at 22:32
  • 1
    $\begingroup$ If you really want to be conservative you could do a full week of Met/Chem spinup. Then noone can question the first dew days of your 2019 results. $\endgroup$
    – f.thorpe
    Jan 10 at 22:35
1
$\begingroup$

Generally, spin up is necessary. To give context, look at how the model initializes. Does that look anything like a real chemical field? If you don’t have the eye for such things, examine the RMSE or other error metrics and compare them to other studies. Odds are, probably not. So you have to supply the model with emissions data and give a long enough period for it to seem at least plausible.

When I was using WRF-Chem to simulate $\ce{O3}$, I used a couple days for the emission of $\ce{NO_x}$ to be somewhat representative.

$\endgroup$
1
  • $\begingroup$ I am new to running the WRF-Chem model, so this question maybe a little elementary. Therefore, I wanted to ask that to perform an year-long simulation of certain chemical species for 2019, would I have to provide the emissions for 2019 or for example, 2015 would suffice. I am assuming that the WRF-Chem model will be able to interpolate the emissions to 2019. The reason I am asking this question is because EDGAR/ECLIPSE emissions are available for 2015 and not 2019. $\endgroup$ yesterday

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.