Modelling soil water dynamics in a forest ecosystem. II: Evaluation of spatial variation of soil profiles

✍ Scribed by W. Bouten; J. V. Witter

• Publisher: John Wiley and Sons
• Year: 1992
• Tongue: English
• Weight: 696 KB
• Volume: 6
• Category: Article
• ISSN: 0885-6087
• DOI: 10.1002/hyp.3360060406

✦ Synopsis

This paper is the second in a series of three papers, dealing with the hydrology of a forested lowland catchment, within the context of soil acidification research. The hydrological behaviour of the unsaturated soil zone is described with the numerical simulation model SWIF. The first paper presents a site specific model calibration and discusses the implications of the hydrological behaviour for soil acidification (Bouten et al., 1992). The present paper deals with the extension of the model results from one specific site to a larger research area. A third paper gives a model description and discusses its numerical behaviour (Tiktak and Bouten, 1992).

In order to evaluate the effects of the field variability of soil horizon thicknesses, the model SWIF is validated by using measured groundwater table dynamics of three sites, each having different soil horizon thicknesses. Subsequently, a sensitivity analysis is carried out in order to select the location-dependent model parameters that cause the main variation in hydrological behaviour. Finally, the spatial patterns of model results and of the location-dependent model parameters are compared.

The drainage depth and the depth of the boundary between the sandy top soil and the underlying boulder clay appear to be the key parameters that cause large differences in transpiration and in the vertical distribution of root water uptake and soil water fluxes. Spatial patterns of model results, therefore, also show resemblance with the spatial patterns of these location-dependent model parameters.

Simulation results for a reference location with averaged soil horizon thicknesses turn out to be beyond the 90 per cent confidence interval of the areal mean model results. This emphasizes the necessity to simulate first and then average.