Unreasonable high NO2 and ozone

[HJORQUER]

Hello everybody

I am using WRF-Chem 4.2 to simulate gas and aerosol chemistry in two different regional domains in South America (tropical and subtropical), using the RADM2/MADE/SORGAM mechanism (chem_opt=2).

I both cases I get unreasonably high NO2 and O3 mixing rations, well above the measured in several monitoring sites, likewise NO is inreasonably low.

I do not understand why I am getting these anomalous results.

Any suggestion is wellcome

Below I am attaching the physics and chem options set in the namelist.input file (tropical case)

Regards

Hector

%%%%%%%%%%%%%%%%%% namelist.input%%%%%%%%%%%

&physics
progn = 0, 0, 0, 0,
naer = 1e9, 1e9, 1e9, 1e9,
sst_update = 0,
rdlai2d = .false.,
usemonalb = .false.,
mp_physics = 6, 6, 6, 6,
ra_lw_physics = 4, 4, 4, 4,
ra_sw_physics = 4, 4, 4, 4,
radt = 12, 12, 12, 12,
sf_sfclay_physics = 91, 91, 91, 91,
sf_surface_physics = 2, 2, 2, 2,
bl_pbl_physics = 1, 1, 1, 1,
ysu_topdown_pblmix = 0,
topo_wind = 1, 1, 1, 1,
bldt = 0, 0, 0, 0,
isfflx = 1,
ifsnow = 1,
icloud = 1,
surface_input_source = 1,
slope_rad = 0, 1, 1, 1,
topo_shading = 0, 1, 1, 1,
cu_physics = 16, 16, 0, 0,
cu_diag = 0, 0, 0, 0,
cudt = 0, 0, 0, 0,
isfflx = 1,
ifsnow = 1,
icloud = 1,
surface_input_source = 1,
num_soil_layers = 4,
num_land_cat = 20,
maxiens = 1,
maxens = 3,
maxens2 = 3,
maxens3 = 16,
ensdim = 144,
cu_rad_feedback = .false., .false., .false., .false.,
mp_zero_out = 2,
mp_zero_out_thresh = 1.e-8,
/

&chem
kemit = 15,
chem_opt = 2, 2, 2, 2,
bioemdt = 30, 30, 30, 30,
photdt = 30, 30, 30, 30,
chemdt = 0, 0, 0, 0,
io_style_emissions = 2,
emiss_inpt_opt = 1, 1, 1, 1,
emiss_opt = 3, 3, 3, 3,
emi_inname = "wrfchemi_d<domain>_<date>",
phot_opt = 1, 1, 1, 1,
gas_drydep_opt = 1, 1, 1, 1,
aer_drydep_opt = 1, 1, 1, 1,
bio_emiss_opt = 3, 3, 3, 3,
ne_area = 500,
opt_pars_out = 1,
dust_opt = 0,
dmsemis_opt = 0,
seas_opt = 2,
gas_bc_opt = 1, 1, 1, 1,
gas_ic_opt = 1, 1, 1, 1,
aer_bc_opt = 1, 1, 1, 1,
aer_ic_opt = 1, 1, 1, 1,
gaschem_onoff = 1, 1, 1, 1,
aerchem_onoff = 1, 1, 1, 1,
wetscav_onoff = 0, 0, 0, 0,
cldchem_onoff = 0, 0, 0, 0,
vertmix_onoff = 1, 1, 1, 1,
chem_conv_tr = 0, 0, 0, 0,
biomass_burn_opt = 0, 0, 0, 0,
aer_ra_feedback = 0, 0, 0, 0,
chem_in_opt = 0, 0, 0, 0,
input_chem_inname = "wrfinput_d<domain>",
have_bcs_chem = .false., .false., .false., .false.,

/

 

[jordanschnell]

Hi Hector,

When you say "unreasonably high", does that mean orders of magnitude differences? An initial step you can try is to turn off your emissions (set emiss_opt = 0) to make sure those are not the issue.

Also, are you simulating over a nested domain? If so, please also try setting feedback = 0. You should also verify that you are setting the correct number of values for the namelist options (check the Registry, and look for the term "max domains").

Jordan

 

[HJORQUER]

Hi Jordan

Thanks for the suggestions.

The NO2 and O3 model overestimation is by a factor of 4 - 6 (with simulated peaks in the hundreds of ppb); nigthtime NO is consistently low near 1 ppb.

Simulated CO and PM2.5, on the other hand, are similar in magnitude to the observed values, although with temporal mismatchs, simulation does not capture all temporal variability.

Regarding nested domains, feedback=0 has already been set.

We have run the model with zero emissions, and in that case all gaseous species have near zero modeled concentrations, as expected.

We have tried halving the emissions of NOx and VOC, and the problem persists, although the ozone overestimation halves, the NO2 overestimation is still 4-5 times above observed NO2, and the NO concentrations at night remain low about 1 ppb.

Apparently, the model is simulating too much oxidation capacity with the baseline emissions (taken from HTAPv2 and spatially allocated with local emission inventories and population data from GHSL). So it seems that urban emissions need to be scaled further down for NOx and VOC.

I will try a new simulation with further reductions in NOx and VOC emissions.

Regards

Héctor

 

[jordanschnell]

Hi Héctor,

Thanks for the information. One additional thing you can try is to compile the model with KPP, and use chem_opt = 106, which is the equivalent mechanism you are using. Or try using another RADM/RACM mechanism. I am not aware of problems with chem_opt = 2, but it would may be good idea to make sure. Please let me know and we can iterate further if necessary.

Jordan

 

[HJORQUER]

Hi Jordan

Thanks for your last suggestion.

I have found that by using chem_opt=106 (RADM/MADE/SORGAM with KPP) the problem is solved and now results make sense for the NO and NO2 diurnal profiles; this has happened for all monitorig sites compared and both South American domains. I do not know what happened with chem_opt=2 though. Maybe the TI fellows who compiled WRF Chem did it so with the KPP option only (but I'm not sure about this).

Héctor