Data convention for forcing data

The table below provides the list of variables needed to force land-surface schemes. It is expected that all land-surface scheme inter-comparisons within GLASS will provide these forcing variables in a netCDF file. The data sets of past projects will also be archived at a central location in this format.This should ensure that the work needed to run the land-surface scheme for an inter-comparison will be kept to a minimum and that past expertise is preserved.

It was chosen in this list of variable to separate the Northward and Eastward wind as in future we my encounter parameterization of surface turbulence which take wind direction into account. To avoid problems of past inter-comparison projects with the separation of precipitation into rainfall and snowfall it was chosen to provide these two variables separately instead of precipitation only.

The choice of units, sign convention and rank of the data are described above.

As a general rule fluxes are averaged over the time-step of the forcing data. The fluxes need to be conserved when a time interpolation is applied before the data is fed to the land-surface scheme. Other variables on the contrary should be instantaneous samples in order to capture well the diurnal cycle. This obviously only applies if the time sampling is high enough for the instantaneous variables to be representative of the time interval. For this table, variables which will be provided as averages are in red

The expected range of values for the forcing variables is provided and will be checked when the netCDF is produced.

Variable	Description	Definition	Units	Sign (direction of positive values)
Wind_N	Near surface northward wind component	Northward wind measured at reference levels near the surface (3D variable).	m/s	Northward
Wind_E	Near surface eastward wind component	Eastward wind measured at reference levels near the surface (3D variable).	m/s	Eastward
	A simplification is possible if only the module of the wind is available.
Rainf	Rainfall rate	Average total rainfall over a time step of the forcing data.	kg/m²s	downward
	Sub-grid rainfall information
Snowf	Snowfall rate	Average total snowfall over a time step of forcing data.	kg/m²s	downward
	Sub-grid snowfall information
Tair	Near surface air temperature	Temperature measured at reference levels near the surface (3D variable)	K
Qair	Near surface specific humidity	Specific humidity measured at reference levels near the surface (3D variable)	kg/kg
PSurf	Surface pressure	Pressure measured at the surface	Pa
SWdown	Surface incident shortwave radiation	Incident radiation in the shortwave part of the spectrum averaged over the time step of the forcing data	W/m²	downward
LWdown	Surface incident longwave radiation	Incident longwave radiation averaged over the time step of the forcing data	W/m²	downward
CO2air	Near surface CO₂ concentration	The partial pressure of CO₂ concentration at the atmospheric reference level (3D variable).	ppmv

Sub-grid variability of atmospheric forcings

Information on the sub-grid variability of atmospheric forcing needs to be provided to the land-surface schemes if possible. These variables pose problems as they are viewed differently in the observational and modelling communities. For instance, rainfall can be separated into a convective and a large scale components in atmospheric models. Some land-surface schemes will use this information. On the other hand from observations it is easier to obtain spatial variance. As atmospheric models will refine their physical processes they will evolve towards quantities much closer to the ones obtained from observations.

It is thus proposed here to allow for both solution and to encourage modelling centers to improve their description of the sub-grid variability of atmospheric variables.

Convective vs large-scale precipitation

The variables Rainf_LS (Snowf_LS) and Rainf_C (Snowf_C) allow specification of the sub-grid characteristics of both liquid and solid precipitation. The distinction between stratiform and convective precipitation is somewhat arbitrary and is no longer valid in many GCMS. However, this distinction serves as a standard until a consensus can be reached on a more representative approach. The total precipitation (Rainf, Snowf) is equal to the sum of the large-scale and convective components and can be specified in lieu of these components.

Variable	Description	Definition	Units	Sign (direction of positive values)
LSRainf	Large-scale rainfall rate	Average large-scale rainfall over a time step of the forcing data.	kg/m²s
CRainf	Convective Rainfall rate	Average of convective rainfall over a time step of the forcing data.	kg/m²s
CSnowf	Convective snowfall rate	Average of the convective snowfall over a time step of forcing data.	kg/m²s
LSSnowf	Large-scale snowfall rate	Average of the large-scale snowfall over a time step of forcing data.	kg/m²s

Rainfall and snowfall spatial variance

Until a better solution is found the spacial variance is proposed.

Variable	Description	Definition	Units	Sign (direction of positive values)
SVRainf	Rainfall rate	Spatial variance of average total rainfall over a time step of the forcing data.	(kg/m²s)²
SVSnowf	Snowfall rate	Spatial variance of average total snowfall over a time step of forcing data.	(kg/m²s)²

If only the module of the wind is available

The are data sets where only the module of wind is available. In this case one could set one of the wind components to zero and write the data into the other. This can generate a significant volume of unneeded data. To solve this problem it is proposed to introduce a wind module variable which would replace Wind_W and Wind_E.

In order to avoid having to change the code from one case to another a sample code section is proposed which allows to keep all the flexibility in the software.

Variable	Description	Definition	Units	Sign (direction of positive values)
Wind	Near surface module of the wind	Wind speed measured at a reference levels near the surface (3D variable).	m/s

Ancillary data

Inter-comparison projects may also wish to provide to the participating land-surface schemes variables which describe the physical characteristics of the surface. It is not as easy to provide a convention for them as they may come with different time sampling rates. This list of variables will probably also change from one inter-comparison to the other as they depend on the scientific objectives of the project.Indeed as all schemes can determine the values for these variables by themselves (At least when they are running coupled to a GCM) it is a legitimate position not to provide them at all. Nevertheless, it is recommended that if they are provided, they be provided in the netCDF format using the meta-data convention described above.The variables which could be included are the following :

Vegetation type(s) (possible 3D variable)
Vegetation cover fraction
Vegetation Albedo
Leaf Area Index
Height of vegetation
Surface roughness length (momentum, heat and moisture)
Greenness or green leaf fraction
Soil information (texture, composition, depth, albedo, etc.)
Solar zenith angle (It is expected that a standard program to compute it will be provided in the bazaar).
Upper drainage area
Fractional area of grid cells (i.e.area within a hydrologic unit; land/sea boundary)
Irrigation fluxes

Last modified: Thu May 4 23:38:11 WEST 2000