Hi Team and @cbengel
I am running ACCESS-rAM3 for simulating the extreme precipitation events. Could you please let me know where the definitions of low, mid, and high level cloud are implemented in ACCESS-rAM3? Specifically, I am looking for the model levels or vertical thresholds or pressure levels used for these diagnostics. Could you point me to the relevant source file or documentation?
Thanks in advance!
Kind regards,
Leena
I’m not too familiar with the cloud levels, but try https://github.com/MetOffice/um/blob/23785d7b7fc65364d4160e211f3593d5ec45a08c/src/control/top_level/setcona_4A.F90#L1440 as a starting point
Hi @Scott and @Leena, I found specific divisions of low/mid/high cloud based on pressure levels in the COSP (Satellite Simulation) package.
@bethanwhite do you know if the COSP definitions match the normal low/mid/high cloud definitions?
Hi @cbengel, I don’t know if COSP matches the low/mid/high cloud levels. I think it would be safest to assume it doesn’t for now.
The answer is probably inside the sub-grid cloud scheme code. I’ve just been removed overnight from the Met Office github accounts because the Microsoft Met Office recertification link wouldn’t work for me (thanks Microsoft…) - which meant I couldn’t re certify my access by yesterday’s deadline, so my Met Office account has been removed from everything 
But I think I have some local code somewhere. Do you know off the top of your head if rAM3 uses PC2? (I can fire up a suite and check if you don’t know).
@Leena would you share the science question you’re trying to answer with these sub-grid cloud amounts? There might be a way to use other model output to diagnose the same information but with better provenance.
According to [the RAL3 paper( GMD - Peer review - The third Met Office Unified Model–JULES Regional Atmosphere and Land Configuration, RAL3 ) (Table 1) RAL3 is not using the PC2 scheme anymore.
Ah it’ll be bi-modal then. But I will double check that that’s what it’s set to use in the suite.
Hi @bethanwhite , @cbengel and @Scott , thank you for your response.
@bethanwhite , We are comparing sub-hourly extreme precipitation simulations from multiple models. To ensure a fair and physically consistent comparison, we need to know how ACCESS-RAM3 defines low-, medium-, and high-level cloud (i.e., the pressure, height, or model-level thresholds), so that equivalent cloud layers can be computed from other models (eg WRF).
Thank you.
@Leena why don’t you use the 3D total cloud field (add liquid cloud mixing ratio and ice cloud mixing ratios), define a threshold in g/kg for ‘cloud’ (e.g. 0.1 g/kg for a cloud that is likely to be precipitating), and then you have a 3D mask for cloud that you can use to define your own height/pressure-level analysis.
This will make things much more consistent with the precipitation analysis because the cloud mixing ratio is calculated by the microphysics scheme (== ‘large scale precipitation scheme’ in the UM), which is also where the precipitation is calculated.
The cloud fractions / amounts are sub-grid fractions calculated by a different physics parameterisation (the cloud scheme) to account for sub-grid cloud fraction variability. It’s an important interface to the radiation scheme, but for resolved precipitation analysis like you are doing, I would use the full cloud mixing ratio fields that are calculated in the microphysics scheme.
Also, I’m pretty sure that WRF doesn’t have a sub-grid cloud fraction scheme in the way that the UM does, so you won’t have the equivalent information from WRF.
Hi @Leena and @bethanwhite,
Thank you for your advice @bethanwhite – much appreciated.
@Leena, if you want you can also reach out directly to the Met Office for advice:
Thanks @bethanwhite , this is an excellent suggestion. I will let the other modelling teams know as well. Thank you again for your time.
Also,@cbengel thank you for sharing the Met Office contact channel and for all your help.
No worries @Leena!
If you want to chat further or have any questions about what I’ve suggested, send me an email and I’m happy to jump on a Zoom call.
@cbengel just to tie up some loose ends for anyone whose Hive search for the rAM3 cloud scheme brings up this thread in the future:
– rAM3 uses the bimodal sub-grid cloud scheme.
If you just use the GUI (or look in /app/um/rose-app.conf for the value of i_cloud_vn) it appears to be set to the Smith scheme.
But actually the ral3p3 optional conf overrides this and sets it to the bimodal scheme.
Thank you @bethanwhite.
I had assumed that RAL3P3 would over-ride any settings so it is great to have that confirmed.
And to complete the loose ends, I have had a trawl through the UM code and found out where these levels are set.
They’re set inside:/src/control/top_level/cderived_mod.F90 in an array called h_split, defined as heights and then converted to model levels by the model when they are used:
“The ranges are determined by heights held in array h_split by comparison with heights of model levels (above surface) in r_ref_theta.”
This height-model level comparison and subsequent calculation of cloud amounts in these specified levels is done in /src/atmosphere/large_scale_cloud/diagnostics_lscld.F90.
They are given as lower / upper bounds for each level type, i.e. low cloud is defined between [h_split(1),h_split(2)], medium cloud is defined between [h_split(2),h_split(3)] and high cloud between [h_split(3),h_split(4)].
( ==> in theory a user could change these definitions, if required).
! low:middle:high cloud model levels =(1)->(2):(2)->(3):(3)->(4)
h_split(1) = 111.0 ! ICAO 1000mb height (m)
h_split(2) = 1949.0 ! ICAO 800mb height (m)
h_split(3) = 5574.0 ! ICAO 500mb height (m)
h_split(4) = 13608.0 ! ICAO 150mb height (m)
So that’s the answer to Leena’s original question 
But my advice for the science question being addressed is to use the 3D cloud field out of the microphysics scheme to use for resolved precipitation analysis, especially across multiple models!
Thanks @bethanwhite.
TO add to that, VERY LOW CLOUD amount is anything below 1000mb.
I agree though that your original advice sounds best, but it will affect the amount of data being written out for @Leena’s experiments. Things to balance out.
Thanks for all your help.
Amazing @bethanwhite …just excellent. Thank you so much for locating it. Thanks