Claudio Paniconi

EVALUATION OF A DISTRIBUTED CATCHMENT SCALE WATER BALANCE MODEL

The validity of some of the simplifying assumptions in a conceptual water balance model is investigated by comparing simulation results from the conceptual model with simulation results from a three-dimensional physically based numerical model and with field observations. We examine, in particular, assumptions and simplifications related to water table dynamics, vertical soil moisture and pressure head distributions, and subsurface flow contributions to stream discharge. The conceptual model relies on a topographic index to predict saturation excess runoff and on Philips infiltration equation to predict infiltration excess runoff. The numerical model solves the three-dimensional Richards equation describing flow in variably saturated porous media, and handles seepage face boundaries, infiltration excess and saturation excess runoff production, and soil driven and atmosphere driven surface fluxes. The study catchments (a 7.2-km2 catchment and a 0.64-km2 subcatchment) are located in the North Appalachian ridge and valley region of eastern Pennsylvania. Hydrologic data collected during the MACHYDRO 90 field experiment are used to calibrate the models and to evaluate simulation results. It is found that water table dynamics as predicted by the conceptual model are close to the observations in a shallow water well and therefore, that a linear relationship between a topographic index and the local water table depth is found to be a reasonable assumption for catchment scale modeling. However, the hydraulic equilibrium assumption is not valid for the upper 100 cm layer of the unsaturated zone and a conceptual model that incorporates a root zone is suggested. Furthermore, theoretical subsurface flow characteristics from the conceptual model are found to be different from field observations, numerical simulation results, and theoretical baseflow recession characteristics based on Boussinesqs groundwater equation.

Newtonian nudging for a Richards equation-based distributed hydrological model

Abstract The objective of data assimilation is to provide physically consistent estimates of spatially distributed environmental variables. In this study a relatively simple data assimilation method has been implemented in a relatively complex hydrological model. The data assimilation technique is Newtonian relaxation or nudging, in which model variables are driven towards observations by a forcing term added to the model equations. The forcing term is proportional to the difference between simulation and observation (relaxation component) and contains four-dimensional weighting functions that can incorporate prior knowledge about the spatial and temporal variability and characteristic scales of the state variable(s) being assimilated. The numerical model couples a three-dimensional finite element Richards equation solver for variably saturated porous media and a finite difference diffusion wave approximation based on digital elevation data for surface water dynamics. We describe the implementation of the data assimilation algorithm for the coupled model and report on the numerical and hydrological performance of the resulting assimilation scheme. Nudging is shown to be successful in improving the hydrological simulation results, and it introduces little computational cost, in terms of CPU and other numerical aspects of the model’s behavior, in some cases even improving numerical performance compared to model runs without nudging. We also examine the sensitivity of the model to nudging term parameters including the spatio-temporal influence coefficients in the weighting functions. Overall the nudging algorithm is quite flexible, for instance in dealing with concurrent observation datasets, gridded or scattered data, and different state variables, and the implementation presented here can be readily extended to any of these features not already incorporated. Moreover the nudging code and tests can serve as a basis for implementation of more sophisticated data assimilation techniques in a Richards equation-based hydrological model.

Catchment-scale hydrological modeling and data assimilation

This special issue of Advances in Water Resources presents recent progress in the application of DA (data assimilation) for distributed hydrological modeling and in the use of in situ and remote sensing datasets for hydrological analysis and parameter estimation. The papers were presented at the De Wageningse Berg conference center, september 2001

A modelling study of seawater intrusion in the Korba coastal plain, Tunisia

Abstract A numerical model that treats density-dependent variably saturated flow and miscible salt transport is used to investigate the occurrence of seawater intrusion in the Korba coastal plain of northeastern Tunisia. We examine the effects of and interplay between pumping, artificial recharge, soil/aquifer properties, and the unsaturated zone. The data processing steps undertaken in this study are briefly described, and a critical assessment is given of the data availability and of the requirements for successful monitoring and modeling of seawater intrusion risks in heavily exploited coastal aquifers such as those found in the semi-arid regions of the Mediterranean basin. An idea of the extent of over-exploitation of the Korba aquifer is obtained by examining the pumping and rainfall/infiltration data, and the simulation results support groundwater pumping as the mechanism for and seawater intrusion as the origin of the salt contamination observed in the soils and subsurface waters of the Korba plain.

The influence of a confining layer on saltwater intrusion under surface recharge and groundwater extraction conditions

In many coastal areas intensive irrigation and groundwater withdrawal can exert counterbalancing effects on the dynamics of seawater intrusion in the aquifer, particularly when non-aquifer irrigation water (artificial recharge) is also used. When, as is often encountered, a shallow phreatic aquifer is separated from a deeper aquifer by a possibly discontinuous and less permeable confining layer, the effect of this semi-permeable layer on the dynamic balance between freshwater and saltwater can be very important. It can act as a barrier to seawater encroachment but also hinder attempts to replenish an over-pumped confined aquifer or to flush salts out of the phreatic aquifer. The presence of localized areas of severe upconing or salt buildup will depend not only on the distribution of pumping wells but also on the heterogeneity of the confining layer. In this study we want to quantify the impact of the confining layer by running simulations based on different realizations of its hydraulic conductivity distribution. One of the questions we seek to address is how important it is to accurately characterize (and therefore how extensively one needs to measure) the properties of such a confining layer. The study site we will use is a 273 km 2 region that forms part of the plain of Oristano on the western coast of central Sardinia (Italy) characterized by Quaternary alluvial deposits, the presence of several lagoons, and about 25000 active and inactive pumping wells. The simulations will be performed with the CODESA-3D model, a three-dimensional finite element model of coupled density-dependent and variably saturated flow and transport.

An optimal switching mechanism for a combined Picard- Newton method for the solution of Richards´equation

Richards’ equation, describing flow in partially saturated porous media, contains nstrong nonlinearities arising from pressure head dependencies in soil moisture and nhydraulic conductivity. Additionally, the time- dependent nature of boundary nconditions can alter the nonlinear characteristics of equation during a transient nsimulation.nVarious iterative methods are used for solving this nonlinear equation, most ncommonly the quadratically convergent Newton – Raphson technique and the simpler nbut only linearly convergent Picard method (successive approximation). The initial nsolution estimate can have a large influence on the behavior of these iterative nschemes, and we have observed through many applications of our numerical nsubsurface flow models that the Newton scheme is more sensitive to the initial nsolution than the Picard scheme is used to calculate improved initial guess nfor the Newton iteration. This scheme should achieve quadratic convergence while nimproving the global behavior of the iteration at less cost and complexity than nalternative globalization techniques such as line search and trust region nmethods. In this work the combined Picard – Newton method is investigated via a ntheoretical analysis, based on a Taylor – Frechet expansion of the nonlinear