Add provenance information about BIOS ancillaries

[ccarouge]

Add information to the global attributes of the gridinfo and add variable attributes were needed.

[SeanBryan51]

Currently the appended BIOS ancillary variables in the BIOS gridinfo don't contain attributes that are particularly helpful, for example:

float rhosoil(latitude, longitude) ;
        rhosoil:long_name = "GDAL Band Number 1" ;
        rhosoil:_FillValue = -9999.f ;
        rhosoil:grid_mapping = "crs" ;

We should at least add a suitable long_name and units attribute for the BIOS ancillary variables.

[SeanBryan51]

@har917 what information do you want to convey in the global attributes in the BIOS gridinfo with regards to provenance? I can add these to the regridding script used for generating the BIOS gridinfo file.

Also if you have knowledge of the relevant units and/or long_name for the BIOS ancillary variables I can add these as well

[har917]

@SeanBryan51

Information to go in the global attributes (may need fine editing)

• data for soil parameters bch, rhosoil, clay, silt, css, hyds, sucs, sfc, ssat and swilt provided from the Atlas of Australian Soils (digital)
• source https://data.csiro.au/collection/csiro:40340
• references: doi:10.25919/5df03e991672d and https://publications.csiro.au/publications/publication/PIlegacy:889
• data for other variables determined from 0.5 degree global gridinfo file.

The units [in ()] and long name should follow the existing information in the gridinfo - however if that's not available then we can include

• bch: b-parameter in Campbell equation (-)
• rhosoil: density of dry soil (kg/m3)
• clay: clay fraction of soil matrix (-)
• silt: silt fraction of soil matrix (-)
• hyds: hydraulic conductivity of soil at saturation (m/s)
• sucs: soil suction at saturation (m)
• sfc: soil moisture content at field capacity (m3 water / m3 soil)
• ssat: soil moisture content at saturation (m3 water / m3 soil)
• swilt: soil moisture content at wilting point (m3 water / m3 soil)

[SeanBryan51]

Here are the attributes in the original gridinfo file (gridinfo_CSIRO_CRU05x05_4tiles.nc) for the appended BIOS variables:

BIOS Variable	long_name	units
silt	UM soil texture - silt fraction	unknow
clay	UM soil texture - clay fraction	unknow
sand	UM soil texture - sand fraction	unknow
css	Soi specific heat capacity	J/kg/K
sfc	VOL SMC AT CRIT PT	unknow
rhosoil	SOIL BULK DENSITY	kg/m^3
bch	CLAPP-HORNBERGER B COEFFICIENT	unknow
hyds	SAT SOIL CONDUCTIVITY	m/s
ssat	VOL SMC AT SATURATOIN	unknow
swilt	VOL SMC AT WILTING	unknow
sucs	SATURATED SOIL WATER SUCTION	m
mvg
c4frac
modis_igbp
avgannmax_fapar

Let me know if you are happy (or unhappy) with these attributes. Note: you did not specify attributes for the sand and css variables- let me know if these attributes should change.

I will also need the attributes for variables mvg, c4frac, modis_igbp and avgannmax_fapar as these were not in the original gridinfo file.

[har917]

@SeanBryan51 Some points - the sand layer is not used (by any configuration of CABLE as far as I know) since this is evaluated from the silt and clay variables. So there is a question as to whether it should be in there or not.

The silt, clay and sand variables are dimensionless (not unknown) - in effect they are fractions.

The units of the sfc, ssat and swilt variables are 'm^3 water / m^3 soil' - I would expand 'VOL SMC' to 'volumetric soil moisture content.

Apologies - yes the dimensions for css are correct. Minor typos in that it's 'soil specific heat capacity' and `VOL SMC AT SATURATION'

I will need to dig out additional information for the other variables (including sources). Long name and units are mvg: NVIS major vegetation group (categorical) c4frac: fractional cover of C4 grass (dimensionless) modis_igbp: IGBP biome (categorical) avgannmax_fapar: average-annual maximum value for FAPAR (dimensionless)

[SeanBryan51]

@har97 here are the proposed changes to BIOS variable attributes:

BIOS Variable	long_name (before)	long_name (after)	units (before)	units (after)
silt	UM soil texture - silt fraction	Silt fraction of soil matrix	unknow	1
clay	UM soil texture - clay fraction	Clay fraction of soil matrix	unknow	1
sand	UM soil texture - sand fraction	Sand fraction of soil matrix	unknow	1
css	Soi specific heat capacity	Soil specific heat capacity	J/kg/K	Unchanged
sfc	VOL SMC AT CRIT PT	Volumetric soil moisture content at critical point	unknow	m3 m-3
rhosoil	SOIL BULK DENSITY	Soil bulk density	kg/m^3	Unchanged
bch	CLAPP-HORNBERGER B COEFFICIENT	Clapp-Hornberger b coefficient	unknow	1
hyds	SAT SOIL CONDUCTIVITY	Hydraulic conductivity of soil at saturation	m/s	Unchanged
ssat	VOL SMC AT SATURATOIN	Volumetric soil moisture content at saturation	unknow	1
swilt	VOL SMC AT WILTING	Volumetric soil moisture content at wilting	unknow	1
sucs	SATURATED SOIL WATER SUCTION	Soil suction at saturation	m	Unchanged
mvg		NVIS major vegetation group
c4frac		Fractional cover of C4 grass		1
modis_igbp (TO BE REMOVED)		IGBP biome
avgannmax_fapar (TO BE REMOVED)		Average-annual maximum value for FAPAR		1

Some reasons/proposals regarding the above changes:

1. Remove "UM soil texture" for silt, clay and sand variables as the UM is not relevant in the context of BIOS ancillaries.
2. Keep sand variable for now for backwards compatibility.
3. Expand acronyms and be verbose in long name description where possible for clarity.
4. Set units to be 1 for dimensionless variables as per the cf-conventions.
5. Categorical data such as mvg, and modis_igbp can be self describing using the flags_* attributes (see cf-conventions#flags). Is this of interest?

[AlisonBennett]

Great to see units going into the variable attributes!

Is it possible to also add the version numbers to the NVIS and MODIS vegetation classifications? I'm not sure what they are, but @har917 may.

---

From: Sean Bryan @.> Sent: Wednesday, 25 September 2024 14:50 To: CABLE-LSM/CABLE @.> Cc: Subscribed @.***> Subject: Re: [CABLE-LSM/CABLE] Add provenance information about BIOS ancillaries (Issue #406)

@har97https://github.com/har97 here are the proposed changes to BIOS variable attributes:

BIOS Variable long_name (before) long_name (after) units (before) units (after) silt UM soil texture - silt fraction Silt fraction of soil matrix unknow 1 clay UM soil texture - clay fraction Clay fraction of soil matrix unknow 1 sand UM soil texture - sand fraction Sand fraction of soil matrix unknow 1 css Soi specific heat capacity Soil specific heat capacity J/kg/K Unchanged sfc VOL SMC AT CRIT PT Volumetric soil moisture content at critical point unknow m3 m-3 rhosoil SOIL BULK DENSITY Soil bulk density kg/m^3 Unchanged bch CLAPP-HORNBERGER B COEFFICIENT Clapp-Hornberger b coefficient unknow 1 hyds SAT SOIL CONDUCTIVITY Hydraulic conductivity of soil at saturation m/s Unchanged ssat VOL SMC AT SATURATOIN Volumetric soil moisture content at saturation unknow 1 swilt VOL SMC AT WILTING Volumetric soil moisture content at wilting unknow 1 sucs SATURATED SOIL WATER SUCTION Soil suction at saturation m Unchanged mvg NVIS major vegetation group c4frac Fractional cover of C4 grass 1 modis_igbp IGBP biome avgannmax_fapar Average-annual maximum value for FAPAR 1

Some reasons/proposals regarding the above changes:

1. Remove "UM soil texture" for silt, clay and sand variables as the UM is not relevant in the context of BIOS ancillaries.
2. Keep sand variable for now for backwards compatibility.
3. Expand acronyms and be verbose in long name description where possible for clarity.
4. Set units to be 1 for dimensionless variables as per the cf-conventionshttps://cfconventions.org/cf-conventions/cf-conventions.html#units.
5. Categorical data such as mvg, and modisigbp can be self describing using the flags* attributes (see cf-conventions#flagshttps://cfconventions.org/cf-conventions/cf-conventions.html#flags). Is this of interest?

— Reply to this email directly, view it on GitHubhttps://github.com/CABLE-LSM/CABLE/issues/406#issuecomment-2372951802, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AIL5VYWNYHZXAFQJTLK624TZYI6KBAVCNFSM6AAAAABOXM6HFOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNZSHE2TCOBQGI. You are receiving this because you are subscribed to this thread.Message ID: @.***>

[AlisonBennett]

Decided to remove MODIS from the GridInfo file as it is just for BLAZE. Can we add provenence (ie. data source and version ) into the global attributes?

[SeanBryan51]

Require version for NVIS major vegetation group classification to add to global attributes - potentially investigate use of flags_* attributes to self describe the data

[har917]

Quick updates:

• on the issue of long_name I would retain the ones that are in the current default gridinfo file (with the possible exception of the layer that we had to change the sign of)
• the NVIS layer would appear to be a processing (GeoTransform and format conversion) of the v5 edition of the National Vegetation Information System. In the absence of further information I think this is v5.0 (not v5.1)