77 NOTE Mesh size selection in a soil-biosphere- atmosphere transfer model A. Mangeney, D. Aubert, J. Demarty, C. Ottlé, and I. Braud Abstract: The aim of this paper is to show the impacts of the vertical discretization in a physical soil numerical model on the calculation of the heat and mass transfer equations. The Simple Soil PlantAtmosphere Transfer (SISPAT) model was used in this study. It solves the coupled equations of mass and energy transfers in the soil and can deal with several horizons for vertically non-homogeneous soils. A series of numerical experiments have been performed to assess the influence of the vertical resolution grid on the simulation of the heat and water transfers using the SISPAT model in a one horizon configuration (homogeneous soil). In the studied case, a minimum of 20 layers has been found for a single 1.4 m thick horizon. Based on this analysis, numerical tests have been performed using SISPAT in a four horizons configuration. Key words: soil–atmosphere exchanges, numerical model, vertical resolution, heat and mass transfer. Résumé : Le but de cet article est de montrer les impacts d’une discrétisation verticale dans un modèle numérique de sol physique lors du calcul des équations de transfert de chaleur et de masse. Le modèle Simple Soil Plant Atmosphere Transfer (SISPAT – modèle vertical des transferts couplés d’eau et de chaleur entre le sol, la végétation et l’atmosphère) a été utilisé dans cette étude. Il résout les équations couplées de transferts de masse et d’énergie dans le sol et peut traiter plusieurs horizons pour les sols verticalement non homogènes. Une série d’expériences numériques ont été effectuées afin d’évaluer l’influence d’une grille de résolution verticale sur la simulation des transferts de chaleur et d’eau en utilisant le modèle SISPAT dans la configuration de un horizon (sol homogène). Dans le cas à l’étude, un minimum de 20 couches a été trouvé dans un seul horizon d’une épaisseur de 1,4 m. En se basant sur cette analyse, les tests numériques ont été effectués utilisant le modèle SISPAT en configuration de quatre horizons. Mots clés : échanges sol-atmosphère, modèle numérique, résolution verticale, transfert de chaleur et de masse. [Traduit par la Rédaction] Introduction Water resources monitoring for riverbasin management re- quires the use of models describing the partition of the precipi- tation into evapotranspiration, water storage, runoff and infiltra- tion at the soil surface, and the transfers in the underground to the water tables and the rivers. The operational management is generally provided using conceptual hydrological models Received 26 February 2002. Revision accepted 23 December 2002. Published on the NRC Research Press Web site at http://jees.nrc.ca/ on 10 February 2003. A. Mangeney, 1,2 D. Aubert, J. Demarty, and C. Ottlé. Centre d’Etude des EnvironnementsTerrestre et Planétaires, 10–12 avenue de l’Europe, 78140 Vélizy, France. I. Braud. Laboratoire d’étude des Transferts en Hydrologie et En- vironnement, B.P. 53, 38041 Grenoble CEDEX 09, France. Written discussion of this article is welcomed and will be received by the Editor until 31 May 2003. 1 Corresponding author (e-mail: mangeney@ipgp.jussieu.fr). 2 Present address: Département de Sismologie, IPGP, 4 Place Jussieu, 75005 Paris, France. calibrated on long-term series of precipitation and streamflow data. Such data are not always available and physical mod- els are sometimes needed to understand the soil water pro- cesses from the knowledge of the soil pedology and geology. The SImple Soil PlantAtmosphere Transfer (SISPAT) model (Braud et al. 1995; Braud 2000) may be used for this purpose. This 1-D model solves the complete set of coupled mass and heat equations, both for vapor and liquid exchanges in the soil– vegetation–atmosphere continuum provided the precise knowl- edge of the hydraulic and thermal soil properties and the phys- ical variables initial states. Such a model is able to deal with nonhomogeneous soils made of several horizons with different hydraulic and thermal properties. It has the potential to predict the soil temperature and water content profiles as well as the surface water and energy fluxes with a very good accuracy, as shown in the framework of different experiments by different authors (Boulet et al. 1999; Gonzales et al. 1999). Compared to conceptual models, mechanistic models are computer time intensiveand need the specification of many pa- rameters. Consequently, solutions are seeked to simplify their use. In this objective, different numerical tests have been per- J. Environ. Eng. Sci. 2: 77–81 (2003) doi: 10.1139/S03-004 © 2003 NRC Canada