CMAQ ISAM ozone source apportionment weird situation

Hello everyone,

I am running CMAQ ISAM for ozone source apportionment. My study domain is whole US, some sourthern part Canada, and northern part Mexico. My study source region is Connecticut State. The weird thing is there is a high ozone contribution in Califonia area. The following map is three month average Connecticut source contribution result. Would you tell me why this happen? Thank you.

This is a weird situation, I agree. What is the scale on these values? ISAM can sometimes behave strangely when its blowing around small values. Connecticut State is a pretty small source of emissions in the context of the US. Did you notice this behavior from the beginning of the simulation or was this something that accumulated further into the 3-months run?


Thank you Sergey,

I also run other Eastern States. The June July August average ISAM source apportionment result for Connecticut and New York States are (ppb):
image image
image image

This behavior appears from the begining of the simulation time and over the whole simulation time.


We have done simulations for New York for 2016 and have seen impacts over the Great Lakes, but not 1.0ppb impacts in California. If you take these maps and divide each one by the bulk concentration (filtering out the divide-by-zero situations), how does that look? It is possible that it is still in the realm of computational noise, since both of your maps look the same for both states. Unfortunately, I am really not sure what is going on.


Thank you Sergey. What do you mean by “filtering out the divide-by-zero situations”? I don’t understand.

I mean that in some places, however rare, it is theoretically possible to have zero O3 bulk concentration. You can not divide by zero, so those place will have strange behavior. Therefore, you need to screen them out in the analysis I suggested (dividing by bulk concentration).


Thank you Sergey. I have tried making fractional source apportionment maps ant the issue is still not fixed. But I find out the Initital Condition result is related to the computational noise.

June July August average ICON source apportionment map:


If I do maps for CT-ICON and NY-ICON, the computational noise disappear(same scale as before):

Why this happen?

Can you post your fraction maps for CT tag and ICON tag. I still not really sure what’s going on.


All have the same legend.
fractional ICON:
image image
fractional CT:
fractional NY:

I think this is probably just noise, if I am reading your legend correctly. The fractions over CA are yellow, which makes them 0.000041-.000050. Even if you consider a high ozone value (100ppb for example), that is still close to nothing. Your original map has 1.000ppb attribution in CA to northeastern source, which doesn’t seem consistent with the fractions you post here.

Either way, the fact that all 3 fractional maps are virtually identical and so tiny, possibly confirms its just manifestation of some numerical artifacts from the cloud and/or chemistry modules.

I will try to run more test with our data here to see if I can replicate this issue. Meanwhile, you might consider just screening out these low values.

Thank you Sergey. The fractional value is the concentration value of each grid divided by the sum of all grids’ concentration. I am not sure if my fractional is the same as your fractional definition.

First of all, I have never seen such a weird issue.
Did you exclude CMAQ spin-up runs which includes significant IC contribution when you calculate average for 3-month CMAQ outputs? Because your ICON source apportionment map shows that O3 (grid concentration/sum of all grids’ concentration) over California has the same magnitudes as your CT-O3 and NY-O3 ( the second and third maps), that implies that your results do include strong ICON memory.
You can try to run 2-month average (July and August), then you may see more reasonable results if you run CMAQ continuously from June 1.
CMAQ outputs will lose ICON memory after15 - 20 days for this case. For long-term CMAQ runs, in my opinion, the impact of ICON on surface O3 can be ignored.

Did you take a look at 3-month average of O3 concentration instead of the concentration divided by domain sum of all grids concentration?

Otherwise you may need to double check your masked region files (CT and NY) used in your emission control file.

I hope it is helpful to fix the issue.

Thank you Feng. But my run is from March 20th to October 31st. My data analysis is for June to August only.

I have checked my masked region file and I don’t see any issue.