The soil-vegetation system behaves as a water reservoir whose content varies in response to fluctuating supplies and demands, and a critical parameter that affects evaporation is the effective water-holding capacity of the soil (Milly and Dunne, 1994). For this reason, the time required to achieve the steady state (in which the balance is not affected by the initial conditions) can be considerably longer than the time scale used for weather forecasts.
According with the results of previous tests (Cassardo et al., 1997), in which it was verified that an integration period of about 6 months seemed to produce a final result independent from the initial conditions, it has been decided to run the LSPM for 5 months.
To verify that the obtained results did not depend on the initial values selected for soil properties, a longer run (9 months) was performed. By comparing the differences in the monthly mean values of September, we founded as maximum difference 0.1 °C for the soil and canopy temperatures (Figure 10) and 0.05% (expressed as fraction of the soil porosity) for the soil moisture (Figure 11) was found in all soil layers.
 

	Figure 10 - Differences between the September 1995 montly mean temperatures in the soil and canopy calculated by LSPM in the 9-month simulation and in the 5-month simulation: (a) soil temperature at 5 cm; (b) soil temperature at 20 cm; (c) soil temperature at 50 cm; (d) soil temperature at 1 m; (e) soil temperature at 2 m; (f) canopy temperature.
	Figure 11 - Differences between the September 1995 montly mean soil moistures (expressed in units of the soil porosity) calculated by LSPM in the 9-month simulation and in the 5-month simulation: (a) soil moisture at 5 cm; (b) soil moisture at 20 cm; (c) soil moisture at 50 cm; (d) soil moisture at 1 m; (e) soil moisture at 2 m.

Then, by running a long-lasting simulation with LSPM starting from 1st May and lasting until 30th September 1995, the following outputs were calculated: temperature and moisture in nine soil layers (whose total extension was about 7 m), net radiation, turbulent (sensible and latent) and conductive (soil-atmosphere) heat fluxes, evaporation, runoff and drainage. The database of soil moisture was used as initial conditions for LAMBO. The same procedure previously described in the CLIPS experiment was used to spatially interpolate the synoptic stations on the LAMBO regular grid, whose mesh size was 15x15 Km.
In Figure 12, the plots of initial soil moisture content in the first 10 cm of soil (in m) taken from ECMWF (control plot) and calculated with LSPM in the frame of the CLIPS experiment were shown. As regards to the Figure 12b, it is necessary to underline that the soil moisture field was calculated on the entire domain of LAMBO, even if the coverage of the synoptic stations was not homogeneous on that area. For graphic reasons, this interpolated field is displayed in Figure 12b on the entire domain, but for sake of comparison it is better to focus the attention on the soil moisture values located inside the rectangular area of longitude 6-12° W and latitude 43-48° N. In this zone, the CLIPS soil moisture field appeared to be more structured, and the differences between the driest and the wettest regions were more evident that in the ECMWF field, that seemed practically constant throughout the whole domain (excepting for the relatively dry tongue on the Western Alps). In particular, the numerical values of the soil moisture evaluated by LSPM are lower than those coming from ECMWF analyses.
 

	Figure 12 - Initial field of soil moisture content (in m) on 12nd September 1995 at 06 UTC: (a)  ECMWF; (b) LSPM.