Problems with ARG activation for multimode lognormal aerosols

[Tianning-Zhao]

Dear Daniel,

This is Tianning Zhao. I'm currently using PYRCEL for a course project. Thanks a lot for all the efforts to put together parcel model in python, it was really helpful. I was having trouble to run ARG activation for customized multimode lognormal aerosols. The error I was getting is like this: 'MultiModeLognorm' object has no attribute 'mu'. Could you please help me fix this? Thank you.

[darothen]

Hi @Tianning-Zhao, thanks for using my code - I hope it's proving useful for your studies!

The issue you're encountering is that the arg2000() function expects a list of aerosols with LogNorm distributions; with a little bit of hacking it should be totally possibly to make a copy of the function which instead reads in a MultiModeLognorm, but I think it's probably easier and faster to just model your aerosol distribution as a collection of LogNorms and send them to arg2000() directly.

If you're able to copy/paste a code example, I might be able to really quick show you how to change your code to do this.

[Tianning-Zhao]

Dear Daniel,

Thanks for your reply. I looked into the code activation.py and now I’m able to run arg2000 and mbn2014 with multimode lognormal distribution.

With that, I did a comparison of the three parameterizations’ performance on seven Jaenicke distributions, which gives me interesting result, where the MBN predicts much higher activated fraction for background, maritime and dust. I’m new to this field and still working on finding a reasonable explanation for this. I was wondering if you could kindly provide some insights/hints that I can look into. I would really appreciate it.

Thanks for your time and consideration.

Best, Tianning

[darothen]

Hi @Tianning-Zhao, that's a really interesting question with a lot of details to consider. The biggest thing to note, though, is that the parameterizations have different assumptions and simplifications which have implications for estimating activation characteristics for different combinations of input aerosol distributions. If you haven't already reviewed the background information / original papers on each parameterization, I'd definitely recommend doing so. You can find some links to the literature in the Scientific Description portion of pyrcel's documentation.

Beyond that, I'd recommend you read Steve Ghan's 2011 paper on droplet activation parameterizations, which contains a very thorough review of their origination and what causes differences between some of them. My own 2013 paper also engages in some lengthy discussion about the ARG and MBN schemes.

Usually the cause for difference is in how the MBN scheme treats the separation between particles whose growth is kinetically limited versus those that rapidly equilibrate with the environmental supersaturation. There are both kinetic effects here - the actual uptake of water vapor - as well as dynamic ones - how much of that uptake can actually be sustained by the ambient water vapor availability. Probably a bit in the weeds for your class project, but a great topic for a Master's thesis!

I hope this is helpful!