The surface and vegetation parameters are defined for 10 surface types (as discussed in the previous chapter), and anywhere from 1 to 10 exist within each 8 x 8 km grid box. A normalized surface weight (omega) is provided in the distribution for each of the surfaces within each 8 x 8 km grid box such that

(3.1)

where i represents the grid index (i=1,1471), and cls represents the surface type or class. The parameters on the distribution have been aggregated into effective parameters on the 8 x 8 km grid following the method outlined in Noilhan and Lacarrère (1995) (presented herein). The correspondence tables for various surface variables are listed in this chapter: they are provided i) as an aid for those who use vegetation parameters which are not contained in the distribution, ii) for those who wish to use their own aggregation rules to determine their effective parameters for the 8 x 8 km grid (other than the ones given on the distribution), iii) for those who need to define parameters for explicit vegetation canopy and bare soil or low ground cover fractions (energy budgets), or iv) for those who need this data to define parameter values for their mosaic tiles. All of the tables listed in this chapter are also available on the distribution.

Soils

    The soil hydrological and thermodynamic parameters are to be estimated by each modeling group based on the provided clay and sand fraction fields. In addition, we have provided the standard four hydrological parameters based on the provided soil texture using the regression relationships from Cosby et al. (1984): the slope of the retention curve (the so-called b-parameter), the porosity (w_sat: m³ m^-3), the matric potential at saturation (psi-sub-sat: m) and the hydraulic conductivity at saturation (k-sub-sat: m s^-1). These parameters are valid for the Brooks and Corey (1966) (with the residual saturation set to zero) or Clapp and Hornberger (1978) models. As an example, the hydrological parameter values for 11 standard textural classes using the regression relationships from Cosby et al. (1984) and Noilhan and Lacarrère (1995) using the Clapp and Hornberger (1978) parameter values are shown in Tables 4.2 and 4.3, respectively. The provided parameter values are optional: we only ask that the modelers determine their parameters in a manner which is consistent with the provided soil texture information.

    There are 2 sets of values of the wilting point (w_wilt) and field capacity (w_fc) volumetric water contents provided on the distribution. The first set of values corresponds to that used by ISBA which is based on regression fits to the parameter values from Clapp and Hornberger (Noilhan and Lacarrêre, 1995). The field capacity volumetric water content corresponds to a hydraulic conductivity of 0.1 mm day^-1 (Wetzel and Chang 1987), and the wilting point corresponds to a matric potential of -150 m (Jacquemin and Noilhan, 1990). The second set uses the same definitions of wilting point and field capacity, but with the Cosby et al. (1984) parameters (both sets of values are shown in Tables 4.2 and 4.3 As the definitions of these two parameters are not standard among the SVAT schemes, the provided parameters are optional and their definition is up to the discretion of the modelers. We only ask that the modelers use the provided texture (and possibly the vegetation mask in the case of the wilting point) data to define their respective parameters.

    The total soil depths (d_total) and rooting depths (d_root) are provided (Table 4.4). We would like all modelers to use the provided total soil depths unless they conflict with a particular model's assumptions (please let us know if they do). Please note that the rooting depths are simply recommended: modelers are free to define the rooting depths or vertical distributions as they see fit. The effective soil (and rooting) depth for each 8 x 8 km grid box is defined as

(3.2)

In addition, the average altitude (topography) for each grid cell is also provided.

Vegetation

    The main vegetation or surface parameters which vary only in space are listed in Table 4.5. Note that the aggregated snow-free albedo on the distribution is calculated as the average of the visible (VI < 0.7 micrometers) and near infrared (IR >= 0.7 micrometers) wavelength albedos. The albedo is aggregated in the distribution in the same manner as the total soil depth, while the effective minimum stomatal resistance (R_{s, min}) is calculated as the weighted sum of the inverse stomatal resistance:

(3.3)

    Several of the parameters also vary as a function of month: the Leaf Area Index (LAI), snow-free surface roughness length (z₀), and the vegetation cover fraction (veg). The same monthly values are to be applied to all 4 years. The correspondence table for the LAI is shown in Table 4.6. It is calculated for each grid box on the distribution as

(3.4)

Please note that any schemes which simulate vegetation evolution (biomass or LAI as prognostic variables for example) are to impose the provided parameters.

    The vegetation cover fraction is defined in the dataset following

(3.5)

where a_cls = 0.5 for forests and 0.6 for low vegetation cover/canopies (Smith et al. 1993; Roujean et al. 1997). The veg correspondence is shown in Table 4.7.

    The correspondence for the snow-free surface roughness length (z₀) is shown in Table 4.8. The effective snow-free roughness for each grid element is calculated using

(3.6)

    We request that modelers linearly interpolate the vegetation cover fraction, LAI, greenness fraction and surface roughness length in time. The convention to be used is that the monthly value corresponds to the value at the end of the month. Therefore, for interpolation purposes, we request that the value at the beginning of the first month be equal to the value at the end of the twelfth month.

Tile parameters

    The surface parameters must be calculated using a re-normalization of the type or class (cls) weights when distinct surfaces are treated within each grid box. For example, if a grid box is divided into distinct bare soil and vegetated regions, then (as an example) the LAI for the vegetated region should be calculated as

(3.8a)

where j corresponds to a sub-region of grid box i. The delta function is unity for all omega values which correspond to the surface j, and it is zero otherwise so that

(3.8b)

Snow Cover and Soil Ice

    The modelers are free to define their snow and/or frozen soil scheme parameters as they see fit (snow albedo, liquid water holding capacity, surface roughness, fractional snow covered area, etc.).

Atmospheric Forcing

    The input atmospheric forcing variables provided on the distribution CD are listed in Table 4.9. There is a separate file for each of the four years. Note that the ALMA convention is used, so that additional information can be found at http://www.lmd.jussieu.fr/ALMA/. Note that the surface pressure (Psurf) is supplied at a 3-hour time step on the distribution, however, it only varies with respect to altitude.

    Some SVAT schemes are unable to use forcing at two different levels (eg. wind at 10 m and temperature at 2 m), so we request that the groups which need forcing at one height vertically interpolate the wind speed to 2 m using

(3.11)

as in PILPS-Phase 2e where z₀ is the effective or aggregated surface roughness length. In addition, it is to be assumed that the forcing variables correspond to the given height above the displacement height.

    The participants are permitted to use the forcing from the 8x8 km grid to calculate sub-grid forcing quantities for the aggregation runs (Experiments 2 and 3) if they are required by the SVAT model, such as (for example) fractional areal precipitation coverage. We only ask that participants indicate to us if they use any sub-grid forcing variables (and then briefly describe how this is done), and that the forcing quantity is conserved: i.e. the average forcing variable for the aggregated grid box equals the corresponding 8x8 km grid average value over the same region.