New type of Level: departure level of most unstable parcel of air

[sebvi]

Branch

https://github.com/wmo-im/GRIB2/tree/issue-16

Summary and purpose

ECMWF is requesting a new type of level to properly encode 2 new parameters: Most Unstable Convective Available Potential Energy (MUCAPE) and Most Unstable Convective INhibition (MUCIN)

Action proposed

ECMWF kindly asks the team to review the proposal and accept it during the next session.

Discussions

In a coming release of its IFS model, ECMWF will make available 2 new parameters, the most unstable CAPE and the most unstable CIN. These 2 new parameters require to locate in the vertical air column the most unstable parcel of air (as shown in the attached pdf). The location of this parcel of air is grid dependent, which means it can't be encoded at a fixed height, an isobaric or isothermal level. Following a recent proposal to introduce the entries 14-16 in the Code Table 4.5, we propose to define a new type of level called "Departure Level of the most unstable parcel" and assign it the code 17 (since this new type of level complements the newly created entries 14-16).

mucape.pdf

Detailed proposal

add a new entry 17 in Code Table 4.5:	Octet	Name	Units	Description
17	Departure level of the most unstable parcel of air (MUDL)	-	This represents the vertical level from which the most unstable parcel of air (the parcel with the largest CAPE) starts rising

[sebvi]

@wmo-im/tdcf : any feedback would be much appreciated!

[efucile]

Do we also need a parameter for the departure level? With the definition of the level I can have a field providing the value of the highest value of CAPE, but it might be important to know at which level it is. I see two fields one with the value of CAPE at the "Departure level of the most unstable parcel of air (MUDL)" and one with the level (model level number) for that value of CAPE

[sebvi]

Yes we are already producing a CAPE parameter (and a CIN parameter) at each model level, this is how the most unstable CAPE is computed: CAPE is computed for all model level and at each grid point the highest value a CAPE is retained.

We will produce a CAPE50 (and CAPE100) which is calculated this way: A mean CAPE is computed between surface and a level differing by 50hPa (and 100hPa) from the surface. I am planning to ecode these this way:

Discipline=0 Category=7 Number=6 First fixed surface = 1 First scale factor=missing First scaled value=missing Second fixed surface = 108 Second scale factor=0 Second scaled value=50

although I don't know how to specify that the layer represented by surface to P0-P=50 is an average, not an integration, a max, a min, or anything else. All the statistical transformations are specific to Time, I am not aware of a way to specify this for an horizontal grid and a vertical column.

[tomkralidis]

As discussed with our subject matter experts here at MSC/CCMEP, we are fine with supporting this field as part of our model output for future implementation.

[tomkralidis]

Note we are also looking into CAPE and CIN and happy to bring in our subject matter experts to discus further.

[SibylleK]

I got comments from our research colleagues. They support the proposal and they provide following comments with an extension of the proposal:

Comment I: In FT2016-1 there already was the attempt of BoM (Weiqing Qu) to establish "air parcel definitions" in Code Table 4.5. The proposal was not accepted mainly for being too prescriptive (over-specifing of thresholds and layer thicknesses).

Comment II: DWD CAPE and CIN products destinguish between "most unstable air parcel" and a "mixed layer parcel" with local level definitions. Therefore DWD supports this proposal and wants to extend it: in addition to the "most unstable parcel", we like to define the "mixed layer parcel" and accordingly the corresponding departure level for the determination of mixed layer CAPE and CIN. In the mean time Sebastien came over with definition of CAPE50 and CAPE100; CAPE50 seems to be similar to DWDs mixed layer CAPE. So our first proposal with dimensionless "18" is changed to unit "Pa" to be able to distinguish between ECMWF CAPE50 and CAPE100 and to overcome the missing "average over a layer".

New entry 18 in Code table 4.5	Octet	Name	Units	Description
18	Departure level of a mixed layer parcel of air with specified layer depth	Pa	This represents the vertical level from which the mixed layer parcel of air starts rising. NOTE: A mixed layer parcel is a parcel of air from a (well-mixed) near-surface atmospheric layer of a specified depth, whose properties represent the mean conditions of the layer. The layer being averaged (mixed) extends from the surface to a depth defined by the level coding.

[sebvi]

Dear @SibylleK , thank you for your coontribution. Do I understand correctly that if I associate a value of 50 to this new type of level 18, it means that my mixed air parcel is computed as the mean value of CAPE/CIN from surface to a difference from surface of 50hPa? If yes, then I support this solution.

More generally for GRIB3, it would be better to have a way to specify if a layer is an average, an accumulation, etc. within that layer.

[chenxiaoxia2019]

@sebvi Hi, Sebastien, I created a new branch for this issue. And I added entries 17 and 18 to the Code Table 4.5. Could you please check it? Thanks.

[SebastianB2]

Dear colleagues,

Please find below a GRIB file (zipped for uploading) which might serve as one example for the newly proposed fixed-surface types. It contains two messages: 1st a constant (nonsensical) field of convective available potential energy for an air parcel with the mean properties of a near-surface layer of thickness 50 hPa. Its DWD short name is "CAPE_ML" and it probably corresponds to "CAPE50" of ECMWF mentioned by Sebastien above. 2nd a constant (nonsensical) field of convective inhibition for the most unstable air parcel with short name "CIN_MU". The meaning of this product is probably the same as ECMWF's "MUCIN" (see above). At the moment the two products are coded at DWD using local definitions for two fixed-surface types: "192 ML start Level of mixed layer parcel" and "193 MU start level of most unstable parcel". Both could be replaced by the newly proposed types 17 and 18 as shown with the GRIB file below.

We used ecCodes (version 2.14.1) to interpret the GRIB file. A short overview of the metadata is shown in grib_ls_of_test_new_levelType_17_and_18.pdf below (it's the output of the command: grib_ls -P levelType:l,levelType:s test_new_levelType_17_and_18.grb),
followed by a complete list of the metadata in grib_dump_of_test_new_levelType_17_and_18.pdf (the output of gribdump -O test...grb). We extended our local definitions for ecCodes by the newly proposed fixed-surface types in order to get the desired interpretation by grib_ls and grib_dump. Please let us know if there are questions, comments or remarks.

test_new_levelType_17_and_18.grb.zip grib_ls_of_test_new_levelType_17_and_18.pdf grib_dump_of_test_new_levelType_17_and_18.pdf

[sebvi]

@chenxiaoxia2019 the branch looks ok to me as well.

I could also decode the samples provided by DWD.

[amilan17]

SUMMARY: Add entries, 17 & 18, to code table 4.5.

[amilan17]

Approved by FT-2020-2.