Annual mean NO2 chemistry not included in the source allocation for NO2 and O3

[brucerd]

Source allocation for NO2 and O3 is carried out in 'uEMEP_source_fraction_chemistry'. Annual mean chemistry, which uses a pdf correction 'correct_annual_mean_chemistry', is not included in this source allocation, so annual mean chemistry is not getting the correct source contributions. This has little consequence for NO2 but for O3 this can lead to changes in sign, which is incorrect.

The solution is just to put this correction routine in after calling the chemistry scheme being used within the source allocation loop. This will apply for the photostationary and the phototimescale schemes but not for the other empirical schemes that are already based on annual mean concentrations.

Complication is that the source contribution calls chemistry and allocates sources within the i,j grid loop, whereas the annual correction routine has the grid loop internally in the call. The solution may be to implement the actual correction function 'uEMEP_annual_mean_pdf_correction_NO2_O3', which is inside the correction subroutine, in the source allocation routine right after the chemistry call, but this needs to be correctly initialised with data, see how this is done in 'correct_annual_mean_chemistry' . This is also slightly complicated by the variable 'y_Jd_acc' that is saved prior to running, to make it quick. This is the distribution function for Jd.

[brucerd]

Alternative solution is to do the source allocation on the uncorrected chemistry. This will give roughly the correct source allocation in relative terms, which will need to be renormalised with the corrected values. This, however, would require a bit of fiddling, keeping the uncorrected chemistry results in a temporary array up until the call for 'uEMEP_source_fraction_chemistry' is made in the 'save_netcdf_file' routine. Else, the chemistry can simply be rerun in 'save_netcdf_file' to get the values required and all the calculations made there, requiring only temporary arrays within this routine. This method is probably the least fiddly and self contained, but will require more time as the chemistry correction takes some time to run. However, this will be quicker than putting the chemistry correction within the source loop in 'uEMEP_source_fraction_chemistry' and doing the calculations there for every source, as suggested above. That will take much much much more time.

So either save the concentrations after the chemistry call 'call uEMEP_chemistry' in 'uEMEP_main', and before the 'call correct_annual_mean_chemistry', or rerun it again in 'save_netcdf_file' and save temporary array locally. The output from the uncorrected chemistry must then be used in 'uEMEP_source_fraction_chemistry' and the resulting source contributions renormalised with the annual corrected total concentrations. Remeber, source allocation is a diagnostic and source allocation for non-linear species is an indication, rather than a fact, as these are normalised within the 'uEMEP_source_fraction_chemistry' routine to always give the total concentration, even though they do not normally add up to the total concentration.