Nested SIS algorithm for a stochastic geolithologic characterization of heterogeneous aquifer CLAUDIA CHERUBINI, CONCETTA I. GIASI, FAUSTA MUSCI, NICOLA PASTORE Dipartimento di Ingegneria Civile e Ambientale Politecnico di Bari Via Orabona 4, 70100 Bari ITALY claudia.cherubini@poliba.it , f.musci@poliba.it Abstract: -Generally, in classical studies that concern the hydrogeological modeling of aquifers, soil properties are represented through a single homogeneous or heterogeneous domain with equivalent parameters that are not representative to the hydrogeological reality but that fit the observed data by means of a calibration process. A more accurate modeling of the lithologic, geological and structural characters of an aquifer is of extreme importance when simulating fluid flow and solute transport, in order to improve the reliability of the numerical simulations. On the other hand the information available for the setting up of an hydrogeological model is subjected to uncertainties and ambiguities due to not univocal interpretations or to uncertainties linked to the methodologies of measurement of the variables of interest. This implies the impossibility to represent the hydrogeological reality with an unambiguous solution even if optimal. The present paper is aimed at characterizing hydrogeologically a coastal aquifer. The presented methodology is finalized at identifying not a univocal model but a set of “equifinal” solutions. Key-Words: - geolithological characterization, lithofacies, aquifer, Sequential Indicator Simulation, nested simulation, transition probability 1 Introduction Numerical model is an essential problem solving tool which acts in support to decision making.The results obtained by numerical simulation do not perfectly correspond never to the real behavior of the phenomenon of interest due to inevitable uncertainties linked to the control parameters. The phase of calibration of a numerical model does not allow to improve, beyond a certain limit, the concordance between the simulated results and the real behavior, as the observations on which the calibration procedure is based are always insufficient in respect to the high number of parameters to calibrate. For these reasons it proves to be fundamental to take into account the influence that the variability and the uncertainty of the model parameters exert on the simulation results. By means of stochastic simulation techniques, assigning a probability distribution to the variables affected by uncertainty, it is possible to generate a sufficiently high number of possible scenarios, carrying out a numerical simulation for each of them. In the geostatistical context, stochastic simulations provide the realization of equiprobable models of the spatial distribution of the considered variable. Several simulation techniques have been developed for stochastic characterization of groundwater reservoir, but no method is completely satisfactory in all practical situations [5][6]. These algorithms can be classified in two main classes: object method and pixel method. The first class draws objects in the space using specified rules in order to produce lithofacies architecture with very different structures, such as low entropy structures and curvilinear channels. Object method algorithms are more attractive but they are less flexible, more difficult to constrain and often require more input data [12]. Recent algorithms based on multiple-point statistics such as SNESIM, build categorical images constrained to sample data and soft information through a training image that can conveniently be generated by un- conditional object-based simulations [9]. These algorithms are able to model complex curvilinear heterogeneities but require a large RAM memory demand. The second class builds the volume occupied by each lithofacies pixel after pixel. The most popular methods are Sequential Indicator Simulation [4] and Truncated (Pluri)Gaussian simulation [10]. These approaches present the following main disadvantages: they are based on two-point statistics Proceedings of the 4th IASME / WSEAS Int. Conference on WATER RESOURCES, HYDRAULICS & HYDROLOGY (WHH'09) ISSN: 1790-2769 201 ISBN: 978-960-474-057-4