Dear could you make clear what does mean by post processing in your mind? As for as I know it refers to shows/present the results/outcomes from WRF model in various ways by using some techniques. Do you mean how to improve WRF simulation for better solar radiation outcome? One thing more WRF do not simulate/compute all components of solar radiation so you have use some other techniques in combination with WRF. we need more information about your idea.

Cheers!

Good morning dear Shahzad.

I have Global horizontal irradiances resulting from WRF runs. I also have aerosol optical depth, water vapour from cimel sunphotometer and satellite instruments. 

So I was wondering if there are any suggested techniques to improve the outcome of WRF's GHIs by adding these affective factors as a post processing procedure since we didn't include aerosol data as an input in WRF.

I guess that the thing I have in mind is somehow a more advanced technique than just take the ratio measured to estimated GHIs and see it's dependence from aod, wv etc (I have already done it) and get a fit out of the data and then apply this fit to the estimations in order to get them "fixed".

I hope that I gave you the hole picture as clear as I could.

: )

Thank you!

I don't know the definite answer for your question. However according to our experience (see the attached Table) WRF-calculated solar radiation was not better (as compared to measurements taken in Ukraine through the whole year) than the results of simple parameterization implemented in RODOS Meteorological Preprocessor according to Hanna and Chang (1993) Atmospheric Environment, Vol. 27A, p. 1491, formula (9) which uses only geographical coordinates, date-time,  land use, temperature, and cloud cover. Therefore it seems to me that with data that you have it is easier to develop parameterization and use it alternatively to WRF output.

Another thing if you have data (such as aod, etc) in one point and you want to improve WRF result in other point, at some distance (say, 20 km) from your measurements. Then you have to calculate solar radiation where you have enough measurements and use this calculated value to improve wrf result in other points using some kind ot optimal interpolation e.g. https://www.researchgate.net/publication/200017141_Introduction_of_data_assimilation_procedures_in_the_meteorological_pre-processor_of_atmospheric_dispersion_models_used_in_emergency_response_systems

If you want we can discuss this subject in more detail.

Article Introduction of data assimilation procedures in the meteorol...

Thank you Ivan.

Do you think that a simple multilinear fit as a post process of the WRF outcome related to aod, water vapor etc. should do the work?

Yes, it seems to me it can work. However nonlinear fit (e.g. neural network) could be more flexible.