Vincenzo Simeone

Optimal design of artificial neural networks by a multi-objective strategy: groundwater level predictions

Abstract Currently, environmental modelling is frequently conducted with the aid of artificial neural networks (ANNs) in an effort to achieve greater accuracy in simulation and forecasting beyond that typically obtained when using solely linear models. For the design of an ANN, modellers must contend with two key issues: (a) the selection of model input and (b) the determination of the number of hidden neurons. A novel approach is introduced to address the optimal design of ANNs based on a multi-objective strategy that enables the user to find a set of feasible ANNs, determined as optimal trade-off solutions between model simplicity and accuracy. This is achieved in a multi-objective fashion by simultaneously minimizing three different cost functions: the model input dimension, the hidden neuron number and the generalization error computed on a validation set of data. The multi-objective approach is based on the Pareto dominance criterion and an evolutionary strategy has been employed to solve the combinatorial optimization problem. From a theoretical perspective, the choice of a multi-objective approach marks an attempt to account for, and overcome, the “curse of dimensionality” and to circumvent the drawbacks of “overfitting” that are inherent in ANNs. Moreover, it is demonstrated that the strategy renders the choice of the ANN more robust, as is evident by “unseen data” in the testing stage, since structure determination is not merely based on the statistical evaluation of the generalization performance. The methodology is tested and the results are reported in a case study relating groundwater level predictions to total monthly rainfall.

Inferring groundwater system dynamics from hydrological time-series data

Abstract The problem of identifying and reproducing the hydrological behaviour of groundwater systems can often be set in terms of ordinary differential equations relating the inputs and outputs of their physical components under simplifying assumptions. Conceptual linear and nonlinear models described as ordinary differential equations are widely used in hydrology and can be found in several studies. Groundwater systems can be described conceptually as an interlinked reservoir model structured as a series of nonlinear tanks, so that the groundwater table can be schematized as the water level in one of the interconnected tanks. In this work, we propose a methodology for inferring the dynamics of a groundwater system response to rainfall, based on recorded time series data. The use of evolutionary techniques to infer differential equations from data in order to obtain their intrinsic phenomenological dynamics has been investigated recently by a few authors and is referred to as evolutionary modelling. A strategy named Evolutionary Polynomial Regression (EPR) has been applied to a real hydrogeological system, the shallow unconfined aquifer of Brindisi, southern Italy, for which 528 recorded monthly data over a 44-year period are available. The EPR returns a set of non-dominated models, as ordinary differential equations, reproducing the system dynamics. The choice of the representative model can be made both on the basis of its performance against a test data set and based on its incorporation of terms that actually entail physical meaning with respect to the conceptualization of the system. Citation Doglioni, A., Mancarella, D., Simeone, V. & Giustolisi, O. (2010) Inferring groundwater system dynamics from hydrological time-series data. Hydrol. Sci. J. 55(4), 593–608.

Geomorphometric analysis based on discrete wavelet transform

This work presents a geomorphometric approach for outlining anomalies of the topographic surface that may be related to geological structures or to geomorphological phenomena. It is based on 2D discrete wavelet transform of digital elevation models. This transform is used to extract singularities of a series of data. This is specifically applied to a digital elevation model, in order to get its detail coefficients and to have evidence about their variations and values. In particular, this approach can be helpful for the delineation and identification of landforms singularities, like landslides and geological structures. The potential and effectiveness of this approach is shown by an application to a case study about a large deep-seated landslide, located at the central-south front of the Apennine in South Italy.

Analysis of infiltration processes into fractured and swelling soils as triggering factors of landslides

Rainfall infiltration can cause a dramatic decrease of suction in unsaturated soils and, consequently, of shear strength, triggering various instability phenomena, such as the slip of steep surface soil layers. Swelling of cracked soils and capillary barrier effects, induced by fine-grained soils overlying a more permeable material, can also affect water flow through this type of soil systems. In the past, few studies on infiltration and rainfall-induced landslides considered the simultaneous effects of surface cracks, swelling materials, and/or the capillary barrier phenomenon. To this purpose, this paper presents the results obtained by a dual-permeability model, which simulates water flow through a fractured swelling soil overlying a more permeable soil and focusing on the influence of these phenomena on triggering of landslides. Numerical results show that for high-intensity precipitations, flow through fractures quickly reaches significant depths and the capillary barrier is broken, while soil swelling leads to a uniform narrowing of cracks. On the other hand, for low-intensity precipitations, fracture flow and swelling are limited only to the first 30–50xa0cm of the topsoil, while cracks almost completely closed. Evaluations of the slope stability show that prolonged low-intensity rainfalls might be more dangerous than short high-intensity rains in triggering surface landslides.

Analysis of rainfall infiltration effects on the stability of pyroclastic soil veneer affected by vertical drying shrinkage fractures

The paper presents a preliminary, simplified evaluation of the effects of rainfall infiltration on the stability of slopes in layered pyroclastic soils affected by shrinkage vertical fractures. The analysis has been developed with a special reference to a stratigraphic sequence obtained by an in situ survey at Pizzo d’Alvano (Southern Italy). The analysis of rainfall infiltration is performed using an original dual-permeability model. Results show how fractures strongly condition infiltration depending on rainfall intensity. Prolonged low-intensity rainfall may lead to a higher saturation of the surface soil layer than short, intense rainfall when water may flow quickly through fractures into the underlying more permeable soil layers. Calculated distributions of pore pressure are used for the slope stability analysis using the infinite slope approach. Variations of the safety factor as a consequence of infiltration show that prolonged rainfall can induce a more relevant decrease in the safety factor than intense precipitations.

Analysis of the Rainfall Preceding the Activation of the Large Maierato Landslide in 2010

This work focuses on the impact of the antecedent rainfall as triggering factor of the large landslide occurred in Maierato (south Italy) on February 15, 2010. According to previous studies by Guerricchio et al. (Tecniche per la difesa dall’inquinamento – 31° Corso di aggiornamento, pp 661–706, 2010) the predisposing factor of the landslide is an ancient deep-seated gravitational slope deformation that significantly affected landforms, drainage networks and infiltration processes of the whole slope where Maierato is located. Here after a brief introduction of the landslide according to the aforementioned study, a hydrological analysis of the rainfall preceding the landslide is presented. The analysis aims at evaluating the exceptionality and some peculiar characters of rainfall, which may be considered among the triggering factors of the landslide.

Quantitative Geomorphological Analysis Based on Wavelet Transforms

The identification and quantitative study of important geological discontinuities, like those related to large landslides constitutes a paramount problem, which claims for a careful, detailed and, if possible, quantitatively based geomorphologic analysis. Numerical geomorphic analyses represent an interesting approach to these studies, allowing for a detailed and pretty accurate identification of hidden topographic anomalies that may be related to large landslides or other hidden geological structures. Geomorphic numerical analyses herein presented, are performed on the digital elevation model, based on the 2D discrete wavelet transform. This analysis is applied to a case study related to the middle-south Apennine at the front of the Apennine, whereas a really large deep-seated large landslide has been previously identified on the base of different geomorphic analysis. Finally, the analysis emphasizes some peculiar aspect of the buried front of the Apennine, which can potentially bias the landslide.

Simulating floods in ephemeral streams in Southern Italy by full-2D hydraulic models

The simulation of flood events is essential for risk prevention and land regulation purposes. Traditionally, it is performed by decoupling the prediction of hydrograph(s) at some section(s) of the waterway(s) from the delineation of downstream flooded areas by using synthetic hydrologic models and hydraulic inundation models, respectively. In the case of the Apulian ephemeral streams (Southern Italy), the application of such an approach is prevented by the lack of monitored rainfall–runoff data and the discrepancy of some key underlying hypotheses. Thus, the suitability of integrated (hydrologic–hydraulic) full-2D models is investigated here by assuming the rainfall as the only external forcing term into each element of the bi-dimensional domain, where the shallow water equations are integrated. This permits the reproduction of runoff generation, propagation and, eventually, flooding at any point of the catchment. Several model runs under many combinations of hydrological losses and surface roughness parameters demonstrate that the full-2D approach realistically reproduce catchment hydraulic behaviour and predicted inundated areas of Apulian ephemeral streams, thus being of direct relevance for basin management purposes.

Relationships between rain and displacements of an active earthflow: a data-driven approach by EPRMOGA

Inclinometer and piezometer measurements have been carried out since 2005 in a slow active earthflow in a clay shale formation of the Italian Southern Apennines. Previous studies outlined the main geometrical and kinematic features of the landslide and the pore pressure response to rainfall. Displacement rates seem to depend on the hydrological conditions as suggested by their seasonal variations. The availability of long time series of data, in some periods recorded in continuum, allows the use of a data mining approach to evaluate the relations among displacement rates in different points of the landslide, and between displacement rates and rainfall. To define such relations, the evolutionary modelling technique EPRMOGA, based on a genetic algorithm, has been used in this paper. The results give a deeper insight into the landslide behaviour on the one hand and, on the other hand, show the reliability of the technique, also in building up management scenarios. In particular, the results show that the landslide displacement rates in different points of the slip surface, although characterized by different values, are linearly dependent and thus have the same time trend, supporting the hypothesis of a constant soil discharge mechanism of movement. Piezometric data in single points cannot be used, in the considered case, to forecast displacements. The obtained relations allow to quantify the displacement rate variations due to contemporary rainfall. The influence of past rainfall is shown to decrease exponentially with temporal distance. Furthermore, the EPRMOGA simulations seem to confirm that there are no other dominant causes, besides rainfall, responsible of displacement rate variations in time.

Notes on morphological characters of ephemeral streams in Apulia region, south Italy

This work introduces an investigation on the peculiarity of morphological characteristics of the ephemeral streams of the central part of Apulia, in south-east of Italy. In particular, the analysis is focused on their morphological features, allowing to improve the knowledge about their origin and genesis. Moreover, a comparison between the ephemeral streams of the Murgia area and those of north Salento is proposed. These two groups of ephemeral stream share interesting similarities in term of stream length, catchment area, and climatic condition, but are located in different geological situations. The earlier group crosses the cretaceous limestone bedrock, outcropping in Murgia upland, the latter crosses quaternary deposits of Brindisi plateau. It is envisaged that ephemeral streams of Murgia even if located in quite geologically ancient rocks show morphological characteristics typical of a stream evolution at a really earlier stage differently from those of north Salento located in quaternary deposits. This is apparently unmotivated, since it should be the opposite. Therefore a likely reason for this may be a different genesis of ephemeral streams of Murgia.