Re: WRF-CHEM issue

[Gabriele Pfister]

Alessandro, 

for the aerosols using MOZART-MOZAIC will be better, I would be surprised if ozone changes much. There can be many reasons for the ozone performance possibly there is some seasonality in anthro emissions that EDGAR does not include, biogenic emissions, fires, ... Or maybe your model simply does not get the meteorology correct. I assume you do regular warm starts and/or use nudging and not free run the whole year. Another thing to look at might also be the total  ozone column used to calculate the overhead ozone column. If you use the new TUV (phot_opt=4) you can set the value in the namelist (e.g. you can derive an appropriate value based on OMI or GOME total ozone columns). 

The diagnostic PM species are the correct ones to compare to obs. They are the sum of the different aerosol components for the given size range. Therefore your aerosol emission mapping is also not correct since you only assign emissions to the unspeciated PM but not to the individual aerosol components. Please look at the MOZART-MOZAIC Userguide on our website for which aerosol emission species you need to include. 

Gabriele 

On Sun, May 13, 2018 at 1:53 PM, Anav, Alessandro <A.A...@exeter.ac.uk> wrote:

Hi Gabriele,

thanks for your suggestions. Actually I have completed also the JAS simulation and indeed O3 is still poorly simulated (see new figure), so, following your suggestion I am setting up the same simulation but with MOZART-MOSAIC scheme (with fires, sea and dust on).

In this regards, I have a couple of questions. Firstly, you have highlighted that "PM2.5 and PM10 are diagnostic species", so I wonder if is correct comparing these two variables from MOZART-MOSAIC with observations? If not, how can I get the total PM2.5 and PM10 concentrations from the wrfout file?

Finally, is it correct using this speciation for MOZART-MOSAIC using anthro_emis and EDGAR-HTAP/MOZART_MOSAIC data?

 src_names = 'CO(28)','NOx(30)','SO2(64)','NH3(17)','BC(12)','OC(12)','PM2.5(1)','PM10(1)',
'BIGALK(72)','BIGENE(56)','C2H4(28)','C2H5OH(46)','C2H6(30)','CH2O(30)',             'CH3CHO(44)','CH3COCH3(58)','CH3OH(32)','MEK(72)','TOLUENE(92)','C3H6(42)','C3H8(44)',
 'BENZENE(78)','XYLENE(106)'

 emis_map = 'CO->CO','NO->NOx','SO2->SO2','NH3->NH3','BIGALK->BIGALK','BIGENE->BIGENE',
'C2H4->C2H4','C2H5OH->C2H5OH','C2H6->C2H6','CH2O->CH2O','CH3CHO->CH3CHO',
'CH3COCH3->CH3COCH3','CH3OH->CH3OH','MEK->MEK','TOLUENE->TOLUENE',
'C3H6->C3H6','C3H8->C3H8','NO2->0.0*NOx','ISOP->0.0*CO','C10H16->0.0*CO',
'SULF->0.0*SO2','C2H2->0.00561790*CO','BENZENE->BENZENE','XYLENE->XYLENE',
'GLY->0.0*CO','MACR->0.0*CO','MGLY->0.0*CO','MVK->0.0*CO',
'HCOOH->0.0*CO','HONO->0.0*CO',
'ECI(a)->0.2*BC','ECJ(a)->0.8*BC','ORGI(a)->0.2*OC','ORGJ(a)->0.8*OC','PM25I(a)->0.2*PM2.5',
'PM25J(a)->0.8*PM2.5','PM_10(a)->PM10','SO4I(a)->0.0*PM10','SO4J(a)->0.0*PM10','NO3I(a)->0.0*PM10',
'NO3J(a)->0.0*PM10','NH4I(a)->0.0*PM10','NH4J(a)->0.0*PM10','NAI(a)->0.0*PM10','NAJ(a)->0.0*PM10',
'CLI(a)->0.0*PM10','CLJ(a)->0.0*PM10','CO_A->CO','CO_BB->0.0*CO','ORGI_A(a)->0.0*PM10',
'ORGI_BB(a)->0.0*PM10','ORGJ_A(a)->0.0*PM10','ORGJ_BB(a)->0.0*PM10','VOCA->0.0*PM10','VOCBB->0.0*PM10'

Many thanks,

Alessandro

-------------------------
Alessandro Anav
University of Exeter
College of Engineering, Mathematics & Physical Sciences
Laver Building, North Park Road, Exeter EX4 4QE, UK

---

Da: Gabriele Pfister <pfi...@ucar.edu>
Inviato: lunedì 7 maggio 2018 21:18:47
A: Anav, Alessandro
Oggetto: Re: WRF-CHEM issue

 

HI Alessandro, 

interesting results. The mismatch in AMJ ozone is puzzling when JFM is fairly simulated (albeit also on the high side). I wonder if the too low aerosols in the model cause too high J values and too active photochemistry? Though of course, only having ozone and PM to compare to tells only a very small part of the story....

PM2.5 and PM10 are diagnostic species and they are not in the MOZART global model (as are not the unspeciated PM species but they make generally a small part of the total aerosols). However, if you map all the individual aerosol species in mozbc, you should have the majority covered. Also I assume most of the PM comes from local emissions, not the boundaries.  

As for the emissions, you have only emissions for OC and BC, so you definitely miss a number of others. By using GOCART you are also not having nitrate aerosols or SOA. Switching to MOZART-MOZAIC would help in this regard. I cannot help you with mapping the EDGAR emissions to the extra species since I am not really using them in my work. Sorry. 

Also I would suggest to make sure you run WRF-Chem with dust emissions active. 

Hope this helps some 

Gabriele

On Mon, May 7, 2018 at 9:41 AM, Anav, Alessandro <A.A...@exeter.ac.uk> wrote:

Dear Gabriele,

following our last discussion, I wish share some results and thoughts with you.

Looking at the figures in attachment it seems to me that WRF-CHEM (MOZCART, chem_opt=112) well reproduces the NO2 in all the two analyzed seasons and O3 is fairly well simulated during JFM, while in AMJ there is a large mismatch, up to 40 ppb at some locations.

The comparison of O3 with MOZART shows that in general the model follows its boundary conditions, even in this comparison the height is not exactly the same (I compared the lowest model layers rather the concentrations at 992 hPa).

Finally, the pattern of PM2.5 and PM10 is completely wrong; for the anthropocentric emissions I used the anthro_emis utility using the namelist mozcart.inp included in the tool:

 src_names = 'CO(28)','NO(30)','BIGALK(72)','BIGENE(56)','C2H4(28)','C2H5OH(46)',
             'C2H6(30)','C3H6(42)','C3H8(44)','CH2O(30)','CH3CHO(44)','CH3COCH3(58)',
             'CH3OH(32)','MEK(72)','SO2(64)','TOLUENE(92)','NH3(17)',
             'OC(12)','BC(12)'
 emis_map = 'CO->CO','NO->NO','BIGALK->BIGALK','C2H4->C2H4','C2H5OH->C2H5OH',
            'C2H6->C2H6','C3H6->C3H6','C3H8->C3H8','CH2O->CH2O','CH3CHO->CH3CHO',
            'CH3COCH3->CH3COCH3','CH3OH->CH3OH','MEK->MEK','SO2->SO2','TOLUENE->TOLUENE',
            'NH3->NH3','OC(a)->OC','BC(a)->BC'

I noted that PMs are not included and speciated here and this likely produces the underestimation of PMs. Now I wonder if MOZCART (chem_opt=112) can be used for PMs estimation (and if so, I just need to add PMs in the mozcart.inp?) or I have to use the MOZART-MOSAIC scheme?

Many thanks,

Alessandro

-------------------------
Alessandro Anav
University of Exeter
College of Engineering, Mathematics & Physical Sciences
Laver Building, North Park Road, Exeter EX4 4QE, UK

---

Da: Anav, Alessandro
Inviato: lunedì 8 gennaio 2018 20:10:43
A: Gabriele Pfister
Oggetto: Re: WRF-CHEM issue

 

I will remove these entries from namelist.

Thanks a lot for help,

Alessandro

-------------------------
Alessandro Anav
University of Exeter
College of Engineering, Mathematics & Physical Sciences
Laver Building, North Park Road, Exeter EX4 4QE, UK

---

Da: Gabriele Pfister <pfi...@ucar.edu>
Inviato: lunedì 8 gennaio 2018 19:58:13
A: Anav, Alessandro
Oggetto: Re: WRF-CHEM issue

 

Alessandro, 

why not just take out the namelist entries to make sure?  

The comparison looks fairly good to me. 

Gabi

On Mon, Jan 8, 2018 at 11:24 AM, Anav, Alessandro <A.A...@exeter.ac.uk> wrote:

Dear Gabi,

thanks for the quick reply.

Indeed, the feeling of "too low" values is based both on comparison with data (see attachment) and the expectation of O3 > 0 during daylight hours.

Anyway, considering the overall good agreement at regional scale, I will continue the simulation to see if [O3] increases with warmer temperatures and radiation.

Regarding the namelist, auxinput8 and auxinput13 shouldn't be used as I set io_form_auxinput8=0 and io_form_auxinput13=0; or am I wrong?  

Best Regards,

Alessandro

-------------------------
Alessandro Anav
University of Exeter
College of Engineering, Mathematics & Physical Sciences
Laver Building, North Park Road, Exeter EX4 4QE, UK

---

Da: Gabriele Pfister <pfi...@ucar.edu>
Inviato: lunedì 8 gennaio 2018 17:07:01
A: Anav, Alessandro
Oggetto: Re: WRF-CHEM issue

 

Alessandro, 

it is not possible to from afar analyze a model simulation. I cannot really make out the time stamp in your simulation, but low ozone with near-zero values at night is certainly not out of the question in areas of high NOx and high particle pollution. When you say "too" low - is this based on comparison to observations or just based on your feeling? I might help to take a look at the PANDA forecasts to get a feeling for ozone levels in Chinese cities: http://www.marcopolo-panda.eu/forecast/

I took a quick look at your namelist but something indeed there is not correct. When you run with chem_opt =112, why do you set 

 auxinput8_inname                    = 'wrfchemi_gocart_bg_d<domain>',

or  

auxinput13_inname                   = 'wrfchemv_d<domain>',

Please look at out Website for information on example namelists and required settings. https://www2.acom.ucar.edu/wrf-chem

Gabi

On Mon, Jan 8, 2018 at 3:25 AM, Anav, Alessandro <A.A...@exeter.ac.uk> wrote:

Dear Dr. Pfister,

I am sorry to bother you, but I am running WRF-CHEM V3.9 (MOZCART configuration) and I am not very confident in results, so I hope you could help me to understand what is going on with the simulation.

Specifically, the O3 concentration is often zero over the polluted mega-cities of China (see animation in attachment). I wonder whether this pattern is expected because of high titration over the polluted regions or might be related to a too high update time interval for photolysis (30 min, see attached namelist) or a wrong configuration of the model? 

I used anthro_emis (with EDGAR-HTAP data) for the anthropogenic emissions, fire_emis for the fire emissions, while for biogenic emissions I used the megan and wesely tools and mozbc for initial and boundary conditions.

Thanks into advance for help,

Best Regards,

Alessandro

-------------------------
Alessandro Anav
University of Exeter
College of Engineering, Mathematics & Physical Sciences
Laver Building, North Park Road, Exeter EX4 4QE, UK

--

================================
Gabriele Pfister

Atmospheric Chemistry Observations & Modeling

National Center for Atmospheric Research

Email: pfi...@ucar.edu

Phone: +1 303 497 2915

Web: https://staff.ucar.edu/users/pfister

--

================================
Gabriele Pfister

Atmospheric Chemistry Observations & Modeling

National Center for Atmospheric Research

Email: pfi...@ucar.edu

Phone: +1 303 497 2915

Web: https://staff.ucar.edu/users/pfister

--

================================
Gabriele Pfister, Deputy Director

Atmospheric Chemistry Observations & Modeling

National Center for Atmospheric Research

Email: pfi...@ucar.edu

Phone: +1 303 497 2915

Web: https://staff.ucar.edu/users/pfister

--

================================
Gabriele Pfister, Deputy Director

Atmospheric Chemistry Observations & Modeling

National Center for Atmospheric Research

Email: pfi...@ucar.edu

Phone: +1 303 497 2915

Web: https://staff.ucar.edu/users/pfister