Hello All,
For AerChemMIP we need a new standard name for a quantity which is the isobaric zonal mean of the eddy meridional temperature advection. That is, the mean wrt. longitude of the product of v' and T', where v' and T' are the eddy meridional velocity and air temperature respectively and "eddy" denotes the departure from the isobaric zonal mean. The units should be "K m s-1"
This is related to two existing terms:
1. northward_heat_flux_in_air_due_to_eddy_advection [W m-2]
2. product_of_northward_wind_and_air_temperature [K m s-1]
The first of these involves the eddy advection, but includes a factor of density and the specific heat constant. Since density is spatially varying, (1) can not be directly converted to the term we want, though there is an approximate relationship.
The second specifies the product, but without reference to the eddies.
The term "eddy_advection" occurs in 38 existing terms, always in the form "...._due_to_[...]eddy_advection", describing the part of some process which can be attributed to some kind of eddy advection. In this case we want a term for the eddy advection itself.
In some preliminary discussion we have the following ideas:
(1) northward_heat_flux_expressed_as_temperature_flux_in_air_due_to_eddy_advection -- staying close to the existing heat flux term;
(2) northward_temperature_flux_in_air_due_to_eddy_advection -- more descriptive
(3) Karl has suggested terms based on "product_of_...", to avoid the use of "temperature flux", because the notion of "flux" sits uncomfortably with a non-conservative quantity like temperature.
In reviewing the use of "eddy_advection" in existing terms, I can see that the nature of the eddy advection is clearly defined in 27 of them, and the heat flux term above is the only one for which the user is left to guess. The standard interpretation would be as a departure from the isobaric zonal mean, but that is not spelled out in the standard name.
Combining some of the ideas from Karl and Michaela (suggestion 2 above), I think we could use:
* northward_isobaric_zonal_mean_eddy_advection_of_air_temperature
This differs from Karl's suggestions in using "eddy_advection" rather than trying to adapt "product_of" to deal with this use case.
The phrase "isobaric_zonal_mean_eddy_advection" would be defined as the zonal mean advection by eddies defined as the departure from the instantaneous isobaric zonal mean of the velocity and the advected quantity.
I'm submitting this on behalf of the group, so please consider comments by Karl and Michaela below. The discussion is still quite open, but I think it is better to engage with the CF list at this stage (the definition of the term we want to represent is clear).
regards,
Martin
________________________________
From: Taylor, Karl E. <taylor13 at llnl.gov>
Sent: 07 May 2019 23:15
To: Michaela Hegglin; Juckes, Martin (STFC,RAL,RALSP)
Cc: Plummer, David (EC); David Neubauer; senesi at meteo.fr; RIGOUDY Gaelle; Michael Schulz
Subject: Re: vt100
I just noticed that there is a standard_name
"product_of_northward_wind_and_air_temperature". Perhaps this could be
modified to something indicating we're only considering the "eddy"
component, e.g.,
"eddy_component_of_product_of_northward_wind_and_air_temperature" or
"product_of_northward_wind_and_air_temperature_eddy_components" or
eddy_associated_product_of_northward_wind_and_air_temperature".
best,
Karl
On 5/7/19 2:23 AM, Michaela Hegglin wrote:
> Hi Martin, Karl,
>
> I got some more input from the DynVarMIP people on this too and Martin has analysed this correctly.
>
> It seems that v?T? * c_p * \rho would give us the correct units in [W/m^2]. The c_p factor is trivial, but the density isn't. Since it is requested on a specific pressure level, it would depend only on the temperature at 100 hPa: \rho = 10^4 Pa/RT. This does add some nonlinearity. To do it properly, you?d need the account for variations in \rho (T) in time and space.
>
> Hence, would it be possible to keep the calculation as it is done by most researchers as v?T? and rename the variable to
>
> northward_temperature_flux_in_air_due_to_eddy_advection with units of [m K s-1].
>
> Also, by ?eddy_advection?, people do mean deviations from the instantaneous zonal mean.
>
> If it doesn?t sound like changing the name is a good idea, we could simply change the units to [K m s-1], and explain that v?T? was intended?
>
> This is what people expect to deliver for this variable as Gaelle?s case shows too.
>
>
> Michaela
>
>
>
>
>> On 7 May 2019, at 09:30, Martin Juckes - UKRI STFC <martin.juckes at stfc.ac.uk> wrote:
>>
>> Hi Karl,
>>
>>
>> It would need a new standard name .. perhaps :
>>
>>
>> northward_heat_flux_expressed_as_temperature_flux_in_air_due_to_eddy_advection
>>
>>
>> -- this is the current name with "_expressed_as_temperature_flux" inserted. It would make it clear that it is closely related to northward_heat_flux_in_air_due_to_eddy_advection, but also provide the necessary level of detail for CF.
>>
>>
>> The relationship between the two is not a simple transformation of units: you need to make an assumption about the density and heat capacity of the air (if this is done after the fact it will, I believe, involve some degree of approximation).
>>
>>
>> regards,
>>
>> Martin
>>
>>
>> ________________________________
>> From: Taylor, Karl E. <taylor13 at llnl.gov>
>> Sent: 06 May 2019 16:34
>> To: Plummer, David (EC); Michaela Hegglin; Juckes, Martin (STFC,RAL,RALSP)
>> Cc: David Neubauer; senesi at meteo.fr; RIGOUDY Gaelle; Michael Schulz
>> Subject: Re: vt100
>>
>> Hi Martin,
>>
>> It appears to me that perhaps more important than the inconsistency
>> between standard_name and units, is the issue of whether or not to
>> remove an instantaneous zonal-mean or a local time mean (or perhaps
>> operate without removing any sort of mean). Since the heat flux is
>> simply the temperture flux scaled by the specific heat of air at
>> constant pressure with a value of about 1000 J kg-1 K-1, users can
>> probably guess whether heat flux or temperature flux is actually being
>> reported and correct for it. Users will not generally be able to guess
>> whether deviations from some mean or full quantities are used in
>> calculated the products.
>>
>> Since vt100 is used for a fairly specific "aerochem" diagnostic, I think
>> they should decide what they *really* want (and need for their
>> analysis). Then I think we should make an exception to our rule
>> (because the science apparently demands it) and modify the data request
>> to reflect this. If there is any way we can do this without changing
>> the current standard_name for vt100, that would certainly be best
>> because then all data would be searched for using the same standard
>> name. The units would be inconsistent for some already written data
>> sets, but errata could be recorded by es-docs for those models
>> indicating the error in units.
>>
>> best regards,
>> Karl
>>
>>
>> On 5/4/19 3:28 PM, Plummer, David (EC) wrote:
>>> Hi Michaela
>>> I'll admit to being a bit out of my depth on stratopheric dynamics, but from a practical standpoint looking at our diagnostic codes that were developed by Charles McLandress we calculate the eddy meridional heat flux as the product of the instantaneous deviations around the instantaneous zonal average of the meridional velocity and temperature. The end product is the zonal and monthly average of this product. This calculation does agree with the description of V'T' in the literature where the primes are used to denote deviations around the zonal mean.
>>> Note that the MIP table for CCMI, which derives from the CCMVal data request, has the following comment: Zonally averaged meridional heat flux at 100 hPa as monthly means derived from daily (or higher frequency) fields
>>> It is really important that the sampling be daily or better because you want to capture the meridional advection of temperature by transient eddies.
>>> It is one of the terms in the Transformed Eulerian Mean (TEM) calculation of the stratospheric residual circulation and the value at 100 hPa is usually seen as a proxy for the total amount of planetary wave drag entering the stratosphere. Chapter 4 of the CCMVal report, Stratospheric Dynamics, used the term 'eddy meridional heat flux' and referenced a paper by Newman et al. (2001) where this term is referred to similarly. I have also seen a few articles in the literature where it is referred to as the 'temperature flux', so that term is not unknown.
>>> David
>>>
>>> ________________________________________
>>> From: Michaela Hegglin <m.i.hegglin at reading.ac.uk>
>>> Sent: May 4, 2019 6:23 AM
>>> To: Martin Juckes - UKRI STFC; Plummer, David (EC)
>>> Cc: Taylor Karl; David Neubauer; senesi at meteo.fr; RIGOUDY Gaelle; Michael Schulz
>>> Subject: Re: vt100
>>>
>>> Hi all again,
>>>
>>> Indeed, our community was lax about the notation and commonly used 'heat flux? instead of the more correct term ?temperature flux? (K m s-1), which is a measure for wave activity. The units should remain [K m s-1]. However, how to get the variable I am still not clear about. Hence, I have copied in David Plummer who may be able to tell us what it is exactly what was calculated in CCMVal for the temperature flux (whether v?T? or v?*T? in fact as Martin says below). Then we have clear instructions and should find the right notation for it.
>>>
>>> Thanks for looking into this,
>>>
>>> Michaela
>>>
>>>
>>>> On 3 May 2019, at 15:47, Martin Juckes - UKRI STFC <martin.juckes at stfc.ac.uk> wrote:
>>>>
>>>> Hello Karl,
>>>>
>>>>
>>>> do you have an idea what to do about this?
>>>>
>>>>
>>>> The variable "vt100" has long name "Northward Heat Flux Due to Eddies",
>>>>
>>>> standard name: "northward_heat_flux_in_air_due_to_eddy_advection" and units "W m-2".
>>>>
>>>>
>>>> AerChemMIP, who requested the variable, really wanted a meridional temperature flux (units "K m s-1") and at least one group (CNRM) has submitted temperature flux data with metadata as above saying it is a heat flux.
>>>>
>>>>
>>>> At the moment only two groups have published vt100 data (CNRM and IPSL), but it is likely that many more have produced the data and are in the process of preparing it for publication.
>>>>
>>>>
>>>> Michaela has suggested an "errata" for the data request, but I don't think this helps: if we decide to make a change, the easiest approach would be to update the request. The only reason for not doing this is that we have committed to not changing the definitions of variables -- but that is a significant reason.
>>>>
>>>>
>>>> regards,
>>>>
>>>> Martin
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> ________________________________
>>>> From: Michaela Hegglin <m.i.hegglin at reading.ac.uk>
>>>> Sent: 02 May 2019 16:23
>>>> To: Michael Schulz
>>>> Cc: Juckes, Martin (STFC,RAL,RALSP); Taylor Karl; David Neubauer; senesi at meteo.fr; RIGOUDY Gaelle
>>>> Subject: Re: vt100
>>>>
>>>> Hi all,
>>>>
>>>> Thank you, Gaelle, for this helpful and important information.
>>>>
>>>> My rather limited understanding is that the fields that were defined and produced in CCMI could be converted in Wm-2 by integration of the fields involving air density and the heat capacity (cp), however, this would not be the fields that users would look for nor would know of how to use/interpret/compare to old studies easily.
>>>>
>>>> So, yes, Martin, if there is a way to remedy the problem with leaving the units but changing the standard name (or by putting a caveat around that variable used in CMIP6) that would be great?
>>>>
>>>> Michaela
>>>>
>>>>
>>>>> On 2 May 2019, at 16:12, Michael Schulz <michaels at met.no> wrote:
>>>>>
>>>>> Hi Martin,
>>>>>
>>>>> I understand, sorry that this slipped our attention.
>>>>>
>>>>> for curiosity - what happens now? Shouldnt there be some log on errors in the data request?
>>>>> Or would you add a specific clarification to the data request that this variable has been specified with an incompatible unit and standard name?
>>>>>
>>>>> best wishes
>>>>> Michael
>>>>>
>>>>>> On 30 Apr 2019, at 13:44, Martin Juckes - UKRI STFC <martin.juckes at stfc.ac.uk> wrote:
>>>>>>
>>>>>> Hello Michaela, Michael,
>>>>>>
>>>>>>
>>>>>> I'm afraid that is not consistent with the CF Convention. The standard name "northward_heat_flux_in_air_due_to_eddy_advection" is for a heat flux, so it needs to have units of "W m-2".
>>>>>>
>>>>>>
>>>>>> It looks as though the term was declared in the CCMI tables with an invalid combination of "units" and "standard_name". In the CMIP6 request, the "units" have been modified to be consistent with the "standard_name".
>>>>>>
>>>>>>
>>>>>> The variable "vt100" has already been published in CMIP6 (with units "W m-2") by IPSL and CNRM, St?phane Senesi or Ga?lle Rigoudy (copied in) might be able to comment on how they did this (the question is: what factor is used to convert v*t into a heat flux? (see, for example, https://esg1.umr-cnrm.fr/thredds/dodsC/CMIP6_CNRM/CMIP/CNRM-CERFACS/CNRM-ESM2-1/amip/r1i1p1f2/AERmonZ/vt100/grz/v20181205/vt100_AERmonZ_CNRM-ESM2-1_amip_r1i1p1f2_grz_197901-201412.nc.html).
>>>>>>
>>>>>>
>>>>>> If this issue had come up earlier we could have changed the standard name and the units ... but we have promised modeling centres not to change the definitions of variables (beyond clarifications) at this stage.
>>>>>>
>>>>>>
>>>>>> regards,
>>>>>>
>>>>>> Martin
>>>>>>
>>>>>> ________________________________
>>>>>> From: Michaela Hegglin <m.i.hegglin at reading.ac.uk>
>>>>>> Sent: 30 April 2019 10:30
>>>>>> To: Michael Schulz
>>>>>> Cc: Juckes, Martin (STFC,RAL,RALSP); Neubauer David
>>>>>> Subject: Re: vt100
>>>>>>
>>>>>> Hi again,
>>>>>>
>>>>>> Here more information from David Plummer, I got the units wrong in my previous email.
>>>>>>
>>>>>> Hope this helps,
>>>>>>
>>>>>> Michaela
>>>>>>
>>>>>>
>>>>>> variable_entry: vt100
>>>>>> !============
>>>>>> modeling_realm: atmos
>>>>>> !----------------------------------
>>>>>> ! Variable attributes:
>>>>>> !----------------------------------
>>>>>> standard_name: northward_heat_flux_in_air_due_to_eddy_advection
>>>>>> units: K m s-1
>>>>>> cell_methods: time: mean longitude: mean
>>>>>> long_name: Meridional Heat Flux
>>>>>> comment: Zonally averaged meridional heat flux at 100 hPa as monthly means derived from daily (or higher frequency) fields.
>>>>>> !----------------------------------
>>>>>> ! Additional variable information:
>>>>>> !----------------------------------
>>>>>> dimensions: longitude latitude time pr100
>>>>>> type: real
>>>>>> valid_min: -1000.0
>>>>>> valid_max: 1000.0
>>>>>> !----------------------------------
>>>>>>
>>>>>> It is the zonal average of the product of V' x T', where the prime is the deviation around the zonal average. So the units you get are just windspeed (m s-1) times temperature (K) -> K m s-1.
>>>>>>
>>>>>>
>>>>>>
>>>>>> ********************************************
>>>>>> Michaela I. Hegglin
>>>>>> Associate Professor in Atmospheric Chemistry
>>>>>> Department of Meteorology
>>>>>> University of Reading
>>>>>> Lyle Building, 302A
>>>>>> Reading, RG6 6BX, UK
>>>>>>
>>>>>> email m.i.hegglin at reading.ac.uk<mailto:m.i.hegglin at reading.ac.uk>
>>>>>> phone +44 (0)118 378 6693
>>>>>> fax +44 (0)118 378 8905
>>>>>> ********************************************
>>>>>>
>>>>>> On 29 Apr 2019, at 12:56, Michael Schulz <michaels at met.no<mailto:michaels at met.no>> wrote:
>>>>>>
>>>>>> Hi Michaela,
>>>>>>
>>>>>> Can you explain how this vt 100 is calculated?
>>>>>>
>>>>>> Its a ccmi variable in AerChemMIP...
>>>>>>
>>>>>> See below,
>>>>>> Thanks
>>>>>> Michael
>>>>>>
>>>>>> ------- Weitergeleitete Nachricht ------
>>>>>> Von: Neubauer David <david.neubauer at env.ethz.ch<mailto:david.neubauer at env.ethz.ch>>
>>>>>> Datum: Mo. 29. Apr. 2019 um 11:11
>>>>>> Betreff: RE: vt100
>>>>>> An: Martin Juckes - UKRI STFC <martin.juckes at stfc.ac.uk<mailto:martin.juckes at stfc.ac.uk>>, Michael Schulz <michaels at met.no<mailto:michaels at met.no>>
>>>>>> Cc: Ferrachat Sylvaine <sylvaine.ferrachat at env.ethz.ch<mailto:sylvaine.ferrachat at env.ethz.ch>>
>>>>>>
>>>>>>
>>>>>> Hi Michael and Martin,
>>>>>>
>>>>>> we would like to start our simulations soon. Therefore we would need to know possible pre-factors of "vt100". Thank you for any help you can provide.
>>>>>>
>>>>>> Kind regards,
>>>>>> David
>>>>>>
>>>>>>>> -----Original Message-----
>>>>>>>> From: Martin Juckes - UKRI STFC [mailto:martin.juckes at stfc.ac.uk<mailto:martin.juckes at stfc.ac.uk>]
>>>>>>>> Sent: Freitag, 12. April 2019 17:10
>>>>>>>> To: Michael Schulz
>>>>>>>> Cc: Neubauer David
>>>>>>>> Subject: Fw: vt100
>>>>>>>>
>>>>>>>> Hello Michael,
>>>>>>>>
>>>>>>>>
>>>>>>>> Can you help with David's question below about the precise definition of
>>>>>>>> the "vt100" variable requested by AerChemMIP?
>>>>>>>>
>>>>>>>>
>>>>>>>> regards,
>>>>>>>>
>>>>>>>> Martin
>>>>>>>>
>>>>>>>>
>>>>>>>> ________________________________
>>>>>>>> From: Neubauer David <david.neubauer at env.ethz.ch<mailto:david.neubauer at env.ethz.ch>>
>>>>>>>> Sent: 12 April 2019 15:39
>>>>>>>> To: Juckes, Martin (STFC,RAL,RALSP)
>>>>>>>> Cc: Ferrachat Sylvaine
>>>>>>>> Subject: vt100
>>>>>>>>
>>>>>>>>
>>>>>>>> Dear Martin,
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> In the CMIP6 data request it is written that the units for the variable vt100
>>>>>>>> (AERmonZ) should be [W m-2]. However, computing the zonal average of
>>>>>>>> the product of the zonal average departures of v and t has the units [m s-1
>>>>>>>> K-1]. By which factors should this product be multiplied to obtain the zonally
>>>>>>>> meridional averaged heat flux?
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Thank you in advance for any help you can provide.
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Kind regards,
>>>>>>>>
>>>>>>>> David
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> --
>>>>>>>> Dr. David Neubauer
>>>>>>>> ETH Zurich
>>>>>>>> Institute for Atmospheric and Climate Science Universitaetstrasse 16, CHN
>>>>>>>> P17.2
>>>>>>>> 8092 Zurich, Switzerland
>>>>>>>> phone: +41 44 632 74 26
>>>>>>>>
Received on Wed May 08 2019 - 02:55:35 BST
