Description of land surface schemes

Differences in soil characteristics

Overview

Three schemes participated in ELDAS: the Tiled ECMWF Surface Scheme for Exchange over Land (TESSEL), the Interface Soil-Biosphere-Atmosphere (ISBA) scheme and the TERRA scheme. ISBA is applied in slightly different forms by CNRM and INM.

All models rely on the well-known resistance analogue to compute the turbulent fluxes. For evapotranspiration, E,

(1)

where c_veg is some measure of the vegetation cover, Dr_v is the difference (D) in water vapour density (r_v) between the effective source height of water vapour and a reference level in the air, and r_a and r_s are the aerodynamic and surface resistance, respectively. For vegetation, the latter quantity is usually computed as:

(2)

where r_s,min is the minimum stomatal resistance under optimal conditions, LAI is the leaf area index and f(x_i) are empirical functions reaching values between 0 and 1, to account for the effect of suboptimal environmental conditions on stomatal aperture. In the present context, it is important to realise that differences in f(x_i) between the models will cause the main difference in the sensitivity of screen level parameters to soil moisture conditions, and are therefore important to the performance of the DA schemes. Furthermore, these differences will cause the main difference in the behaviour of the modelled E. Also, these functions may obscure well-known relations between evapotranspiration and environmental factors, such as the one between E and shortwave radiation for vegetated surfaces.

In all cases, soil moisture is a relatively slowly varying variable. Of paramount importance is the water-holding capacity, defined as the difference between field capacity and wilting point for a soil layer with depth 1 m. The water holding capacity depends on the soil texture and differs considerably among the models, as shown below.

Differences in soil characteristics

Table 1 shows the The largest range in soil water holding capacity appears to be contained in TERRA. Although ISBA computes wilting point and field capacity from the textural composition of the soils, the actual range of water holding capacity (~80 mm) is small. TESSEL-SCM defines one soil type only. Note that the amount of water available for evapotranspiration is not only given by the water holding capacity as defined in Table 1, but also by rooting depth.

Soil

ISBA

TERRA

TESSEL-SCM

Sand

Sandy loam

Loam

Loamy clay

Clay

154

160

230

185

206

152

Table 1. Water holding capacity (mm) for different soil types in ISBA, TERRA and TESSEL-SCM, defined as the difference between field capacity and wilting point the for a 1-m deep layer of soil.

Application during ELDAS

An overview of the SVAT schemes and the main layout of the DA experiment is given in Table 2. For each centre and SVAT-scheme the table shows the main land-surface database that was used for this validation study. While ISBA and TERRA have been run in a fully coupled mode, TESSEL has been run in a single-column mode (TESSEL-SCM). ISBA and TERRA construct their land-surface properties from the Ecoclimap database (Masson et al., 2003), while TESSEL utilizes GLCC (Loveland et al., 2000). For the forcings of the land-surface part, ISBA and TERRA rely on their model-derived precipitation (P), shortwave and longwave radiation (SW and LW, respectively). TESSEL uses the special ELDAS forcing databases for these quantities described elsewhere. The DA systems of the models use screen-level observations to diagnose deviations in the soil moisture fields. ISBA and TESSEL used temperature (T) as well as relative humidity (RH), while TERRA used T only. In the case of ISBA, an additional correction to soil moisture was applied to account for the difference between the model precipitation and the ELDAS precipitation..

An overview of the SVAT schemes and the main layout of the DA experiment is given in Table 2. For each centre and SVAT-scheme the table shows the main land-surface database that was used for this validation study. While ISBA and TERRA have been run in a fully coupled mode, TESSEL has been run in a single-column mode (TESSEL-SCM). ISBA and TERRA construct their land-surface properties from the Ecoclimap database (Masson et al., 2003), while TESSEL utilizes GLCC (Loveland et al., 2000). For the forcings of the land-surface part, ISBA and TERRA rely on their model-derived precipitation (P), shortwave and longwave radiation (SW and LW, respectively). TESSEL uses the special ELDAS forcing databases for these quantities described elsewhere in this volume. The DA systems of the models use screen-level observations to diagnose deviations in the soil moisture fields. ISBA and TESSEL used temperature (T) as well as relative humidity (RH), while TERRA used T only. In the case of ISBA, an additional correction to soil moisture was applied to account for the difference between the model precipitation and the ELDAS precipitation. The present validation study is restricted to the period May-October 2000, for which output from all models was available.

Centre	SVAT	Land-surface database	Forcings	Soil Moisture Assimilation
*CNRM*	ISBA	Ecoclimap	Model P, SW, LW	(P), T, RH
*DWD*	TERRA	Ecoclimap	Model P, SW, LW	T
*ECMWF*	TESSEL-SCM	GLCC	ELDAS P, SW, LW	T, RH

Table 2. Experiment setup for the present validation study.