Reliability of groundwater vulnerability maps obtained through statistical methods.

Statistical methods are widely used in environmental studies to evaluate natural hazards. Within groundwater vulnerability in particular, statistical methods are used to support decisions about environmental planning and management. The production of vulnerability maps obtained by statistical methods can greatly help decision making. One of the key points in all of these studies is the validation of the model outputs, which is performed through the application of various techniques to analyze the quality and reliability of the final results and to evaluate the model having the best performance. In this study, a groundwater vulnerability assessment to nitrate contamination was performed for the shallow aquifer located in the Province of Milan (Italy). The Weights of Evidence modeling technique was used to generate six model outputs, each one with a different number of input predictive factors. Considering that a vulnerability map is meaningful and useful only if it represents the study area through a limited number of classes with different degrees of vulnerability, the spatial agreement of different reclassified maps has been evaluated through the kappa statistics and a series of validation procedures has been proposed and applied to evaluate the reliability of the reclassified maps. Results show that performance is not directly related to the number of input predictor factors and that is possible to identify, among apparently similar maps, those best representing groundwater vulnerability in the study area. Thus, vulnerability maps generated using statistical modeling techniques have to be carefully handled before they are disseminated. Indeed, the results may appear to be excellent and final maps may perform quite well when, in fact, the depicted spatial distribution of vulnerability is greatly different from the actual one. For this reason, it is necessary to carefully evaluate the obtained results using multiple statistical techniques that are capable of providing quantitative insight into the analysis of the results. This evaluation should be done at least to reduce the questionability of the results and so to limit the number of potential choices.

[1]  Jacob Cohen.  A Coefficient of Agreement for Nominal Scales , 1960 .

[2]  Dennis R. Helsel,et al.  Assessing ground-water vulnerability to contamination: Providing scientifically defensible information for decision makers , 2002 .

[3]  Richard M Vogel,et al.  Predicting ground water nitrate concentration from land use , 2005, Ground water.

[4]  Simone Sterlacchini,et al.  Landslide Representation Strategies in Susceptibility Studies using Weights-of-Evidence Modeling Technique , 2007 .

[5]  Simone Sterlacchini,et al.  Spatial and statistical assessment of factors influencing nitrate contamination in groundwater. , 2008, Journal of environmental management.

[6]  Simone Sterlacchini,et al.  The Use of the Weights-of-Evidence Modeling Technique to Estimate the Vulnerability of Groundwater to Nitrate Contamination , 2007 .

[7]  A. Dassargues,et al.  Current trends and future challenges in groundwater vulnerability assessment using overlay and index methods , 2000 .

[8]  Birgit Terhorst,et al.  Landslide susceptibility assessment using “weights-of-evidence” applied to a study area at the Jurassic escarpment (SW-Germany) , 2007 .

[9]  Guocheng Pan,et al.  A Comparative Analysis of Favorability Mappings by Weights of Evidence, Probabilistic Neural Networks, Discriminant Analysis, and Logistic Regression , 2003 .

[10]  T N Narasimhan,et al.  Basic Tenets for Water Management , 2005, Ground water.

[11]  Jagath J Kaluarachchi,et al.  Sustainability of ground water quality considering land use changes and public health risks. , 2006, Journal of environmental management.

[12]  Chang-Jo Chung,et al.  On Blind Tests and Spatial Prediction Models , 2008 .

[13]  M G Rupert,et al.  Calibration of the DRASTIC Ground Water Vulnerability Mapping Method , 2001, Ground water.

[14]  F. Voss,et al.  Predicting the Probability of Elevated Nitrate Concentrations in the Puget Sound Basin: Implications for Aquifer Susceptibility and Vulnerability , 1997 .

[15]  G. Bonham-Carter.  Geographic Information Systems for Geoscientists: Modelling with GIS , 1995 .

[16]  B. Ruddy,et al.  Probability of nitrate contamination of recently recharged groundwaters in the conterminous United States. , 2002, Environmental science & technology.

[17]  A. Brenning.  Spatial prediction models for landslide hazards: review, comparison and evaluation , 2005 .

[18]  Alberto Carrara,et al.  Multivariate models for landslide hazard evaluation , 1983 .

[19]  R. J. Pike,et al.  GIS technology and models for assessing landslide hazard and risk , 2008 .

[20]  M H Hamza,et al.  A GIS-based DRASTIC vulnerability and net recharge reassessment in an aquifer of a semi-arid region (Metline-Ras Jebel-Raf Raf aquifer, Northern Tunisia). , 2007, Journal of environmental management.

[21]  Chang-Jo Chung,et al.  Using likelihood ratio functions for modeling the conditional probability of occurrence of future landslides for risk assessment , 2006, Comput. Geosci..

[22]  G. Raines.  Evaluation of Weights of Evidence to Predict Epithermal-Gold Deposits in the Great Basin of the Western United States , 1999 .

[23]  Chang-Jo Chung,et al.  Is Prediction of Future Landslides Possible with a GIS? , 2003 .

[24]  J. Vrba,et al.  Guidebook on Mapping Groundwater Vulnerability , 1994 .

[25]  Gary L. Raines,et al.  Development and Implementation of a Bayesian-based Aquifer Vulnerability Assessment in Florida , 2007 .

[26]  Cristiano Ballabio,et al.  Influence of threshold value in the use of statistical methods for groundwater vulnerability assessment. , 2009, The Science of the total environment.

[27]  R. Soeters,et al.  Use of Geomorphological Information in Indirect Landslide Susceptibility Assessment , 2003 .

[28]  C. Chung,et al.  Probabilistic prediction models for landslide hazard mapping , 1999 .