I followed the instructions in the CMAQ v5.0 Adjoint User’s Manual and run the model successfully.
After when I analyzed the results, I found that the sensitivity of ISOP(isoprene), which was supposed to be the precursor of the ozone, had a large number of negative values. It is well known that ISOP is an important substance to promote the formation of ozone. The sensitivity of ISOP to ozone should be positive. This means that the
I want to find some ways to solve question. May I ask how to solve the problem?
Sensitivity of isoprene was negative in your run to ozone emissions or ozone initial conditions or ozone boundary conditions? What exactly where you trying to calculate?
According to the accompanying model user manual, using the reverse running model will result in two documents, which are respectively sensitive to concentration and emission. When I run the program, I use the ozone concentration in the last hour of a day in the forward simulation as a forced file.
Then I started to analyze the simulation results. I found negative values when drawing the time series diagram of the sensitivity of ISOP emissions to ozone concentration (Figure 1).
You are correct to expect for Isoprene, as an ozone precursor, to have positive impact on ozone (i.e. positive adjoint sensitivity if your cost function is ozone-based). However, while overall this is correct, Isoprene like many other VOCs, may have negative impact on ozone (i.e. negative ozone sensitivities) in low NOx environments. This is consistent with findings from other forward based studies such as Hakami, Bergin, and Russell (2004) (ES&T, 38, 6748). Dunker, Koo, and Yarwood (2016) (Atmos. Environ., 145, 326) discuss the specific case of Isoprene in more details.
What is important to note is that in forward sensitivity studies (such as DDM or brute-force), negative ozone to VOC sensitivities are less likely to appear, as the sensitivity is often to larger scale changes in VOC emissions (like across a sector in entire country or region), so the negative impact from low NOx regions where emissions are smaller than industrial or populated areas is masked by positive sensitivities in more polluted areas. In adjoint analysis though, those low NOx locations are represented on their own, so you tend to see larger areas of negative sensitivities when you look at things spatially.
Hi, amir_hakami, Thank you very much for your answer.
I added NOx concentration and ISOP emission data (Figure 2) on the basis of Figure 1. I think this is consistent with your answer by analyzing the picture, but I found a new problem: the low concentration of NOx leads to the negative sensitivity of ISOP to ozone concentration. If I want to reduce it, what can I do? At present, I think high NOx regions and time periods should be selected.
However, I found from Figure 2 that NOx concentration and ISOP emissions seem to have the opposite trend, that is, high NOx time corresponds to lower ISOP emissions, but low NOx corresponds to higher ISOP emissions. May I ask how to solve the problem?
