Power: planner oriented watershed modeling system for environmental responses

POWER is an acronym that stands for Planner Oriented Watershed modelling system for Environmental Responses. It is a software package developed within the Department of Hydrology of LTHE (Haverkamp et al., 2003) aimed at the simulation of integrated flow systems of stream and overland flow, soil water and solute movement (e.g., fertilizer and pesticides) in the unsaturated and saturated aquifer zones combined with plant root uptake.Basically, the POWER modelling system is conceived as a combination of three innovative fully complementary flux-based methods. Starting with the theoretical modelling platform presented by Reggiani et al. (1998), the method is adapted for application to real world watershed studies without having recourse to the restrictive hypothesis used by Reggiani et al. (2000). Subsequently, the method is coupled with the flux-based equation solver newly developed by Ross (2002, 2003), using the scale invariance approach of Haverkamp et al. (1998) to describe the hydraulic soil properties.The spatial discretization geometry is based on a hierarchy of three levels, i.e., two in the horizontal (REWs and RECs) and one in the vertical, where the second and third discretization geometries are fully nested within their respective parent structures. For the first level of horizontal discretization, the watershed is subdivided into a series of representative elementary sub-watersheds (REWs) by the simple use of a digital terrain model (DTM). The REWs constitute well defined, spatial entities with irregular prismatic volumes. Each REW is considered to host the five primary components of the hydrological cycle, i.e., (i) open channel (river) flow; (ii) infiltration and/or saturation excess surface runoff; (iii) vertical unsaturated subsurface flow; (iv) lateral unsaturated subsurface flow and (v) groundwater flow, where each of the five flow processes is characterized by its own specific space and time scale. The second level of horizontal discretization consists of a disaggregation of each REW into a series of vertical columns called representative elementary columns (RECs), which involve both the unsaturated (vadose) zone and saturated zone (underlying regional water table) all the way down to the bedrock taken as the bottom boundary condition. The determination of the RECs is based on a series of superimposed GIS layers such as maps of soil texture, land-use, river network and infrastructure. Each REC is supposed to contain the sub-regions associated with the primary components (ii), (iii), (iv) and (v) of the hydrological cycle. The aggregation between the two horizontal levels of REWs and RECs is obtained by the use of a physically-based averaging procedure applied for the description of the various flow components of the water flow phenomena. The vertical discretization corresponds to the third level of discretization, consists of the subdivision of each REC column into a given number of cells.The calculation of the different flow processes is carried out following the bottom-up procedure, i.e., for each time step in a simulation, the water flow and solute transport is calculated firstly for the resolution of the RECs followed by that of the REWs and finally for the whole watershed. The 3D unsaturated/saturated water flow (e.g., infiltration, soil evaporation, water ponding soil water redistribution, lateral flow, deep drainage and capillary rise) and solute transport is calculated by the numerical solution of the Richards` (1931) equation combined with a standard convection/dispersion equation with a linear equilibrium adsorption isotherm. A simple root extraction model calculates the actual plant water uptake and transpiration as a function of root length density changing with depth and time. The coupling of a nitrogen transport and transformation with the POWER system is achieved by using the model of Birkinshaw and Ewen (2000) tracking both the carbon and nitrogen balance of three separate pools covering the soil organic matter. Possible surface run-off at the soil surface is estimated by the use of a flow-discharge equation newly developed for the case of shallow water based on a modified Manning (1891) equation. The POWER code is planned as an evolutive simulation system which may be interfaced, at a later stage, with other models such as large-scale atmospheric simulation models, plant growth models and/or economic models. The object oriented C++ coding language is used.As the reality shows that most natural watersheds are often poorly defined in input-data, the number of system parameters is reduced to a strict minimum in order to reduce the risk of over-parameterization. Considerable effort has been directed during the development of the POWER code, towards the choice of independent system-parameters. Moreover, the architecture of the POWER code has fully structural flexibility enabling the model to match the sophistication of the solution with the specific project requirements and/or the availability of data. The so-called bottom-line configuration of the POWER code is not more input-data demanding than most classical lumped modelling systems.The results obtained with POWER for two case studies of two different watersheds, are discussed. / POWER - Planned Oriented evaluative Watershed model for Environmental and socio-economic Responses) est un outil de modélisation hydrologique intégrée développé au LTHE Grenoble permettant la simulation des flux d'eau et de solutés (fertilisants et pesticides) prenant en compte les écoulements dans la zone saturée et non saturée, incluant l'extraction racinaire ainsi que les écoulements de surface et dans la rivière. La plate-forme de modélisation POWER combine trois approches novatrices. La première s'appuie sur des équations intégrées de conservation de la masse et de la quantité de mouvement intégrées à l'échelle de sous-bassins (Reggiani et al., WRR, 2000). La seconde permet une résolution robuste des transferts d'eau et de solutés dans les sols à partir d'une solution non itérative de l'équation de Richards et de convection - dispersion (Ross, Agronomy J., 2003). Enfin , la spécification des propriétés hydrodynamiques des sols repose sur les théories d'invariance d'échelle proposées par Haverkamp et al. (1998). La discrétisation spatiale de l'espace utilise un maillage représentant explicitement l'hétérogénéité des surfaces continentales au travers de maillages emboîtés: les sous-bassins représentatifs (REWs) d'une part et les colonnes élémentaires représentatives (RECs) d'autre part. Le premier niveau de découpage permet de travailler sur des unités de modélisation pertinentes pour l'hydrologie et s'appuyant sur le topographie sur lesquelles sont modélisées les écoulements dans la zone saturée et la rivière. Le second niveau de découpage essaie de dégager des unités homogènes d'un point de vue pédologie et occupation du sol, sur lesquelles sont représentés les transferts dans la zone non saturée du sol et le ruissellement de surface. S'appuyant surl des avancées technologiques en informatique la plate-forme est modulaire (modélisation orientée objet) et il est donc aisé de modifier la représentation d'un processus. Une telle structure permet d'avoir un outil assez bien adapté à la problématique du changement d'échelle.