A ROC analysis-based classification method for landslide susceptibility maps

A landslide susceptibility map is a crucial tool for land-use spatial planning and management in mountainous areas. An essential issue in such maps is the determination of susceptibility thresholds. To this end, the map is zoned into a limited number of classes. Adopting one classification system or another will not only affect the map’s readability and final appearance, but most importantly, it may affect the decision-making tasks required for effective land management. The present study compares and evaluates the reliability of some of the most commonly used classification methods, applied to a susceptibility map produced for the area of La Marina (Alicante, Spain). A new classification method based on ROC analysis is proposed, which extracts all the useful information from the initial dataset (terrain characteristics and landslide inventory) and includes, for the first time, the concept of misclassification costs. This process yields a more objective differentiation of susceptibility levels that relies less on the intrinsic structure of the terrain characteristics. The results reveal a considerable difference between the classification methods used to define the most susceptible zones (in over 20% of the surface) and highlight the need to establish a standard method for producing classified susceptibility maps. The method proposed in the study is particularly notable for its consistency, stability and homogeneity, and may mark the starting point for consensus on a generalisable classification method.

[1]  G. Jenks.  The Data Model Concept in Statistical Mapping , 1967 .

[2]  P. C. Smits,et al.  QUALITY ASSESSMENT OF IMAGE CLASSIFICATION ALGORITHMS FOR LAND-COVER MAPPING , 1999 .

[3]  P. Atkinson,et al.  Generalised linear modelling of susceptibility to landsliding in the Central Apennines, Italy , 1998 .

[4]  Veronica Tofani,et al.  Identification of landslide hazard and risk ‘hotspots’ in Europe , 2013, Bulletin of Engineering Geology and the Environment.

[5]  C. Baeza,et al.  Statistical and spatial analysis of landslide susceptibility maps with different classification systems , 2016, Environmental Earth Sciences.

[6]  I. C. Martí.  Elaboración y validación de un modelo jerárquico derivado de SIOSE , 2013 .

[7]  J A Swets,et al.  Measuring the accuracy of diagnostic systems. , 1988, Science.

[8]  L. Ayalew,et al.  The application of GIS-based logistic regression for landslide susceptibility mapping in the Kakuda-Yahiko Mountains, Central Japan , 2005 .

[9]  Ryan E. Leigh.  Using genetic algorithms to create believable agents , 2006 .

[10]  Fausto Guzzetti,et al.  Tier-based approaches for landslide susceptibility assessment in Europe , 2013, Landslides.

[11]  Robert C. Holte,et al.  Cost curves: An improved method for visualizing classifier performance , 2006, Machine Learning.

[12]  P. Reichenbach,et al.  Estimating the quality of landslide susceptibility models , 2006 .

[13]  Carmen Cadarso-Suárez,et al.  OptimalCutpoints: An R Package for Selecting Optimal Cutpoints in Diagnostic Tests , 2014 .

[14]  Manoj K. Arora,et al.  Approaches for comparative evaluation of raster GIS-based landslide susceptibility zonation maps , 2008, Int. J. Appl. Earth Obs. Geoinformation.

[15]  J. Chacón,et al.  Engineering geology maps: landslides and geographical information systems , 2006 .

[16]  I. Evans.  The selection of class intervals , 1977 .

[17]  Giovanni B. Crosta,et al.  Techniques for evaluating the performance of landslide susceptibility models , 2010 .

[18]  David A. Bennett,et al.  Using Genetic Algorithms to Create Multicriteria Class Intervals for Choropleth Maps , 2003 .

[19]  K. Clayton,et al.  Transactions of the Institute of British Geographers , 1959 .

[20]  M. Greiner,et al.  Principles and practical application of the receiver-operating characteristic analysis for diagnostic tests. , 2000, Preventive veterinary medicine.

[21]  Seymour Geisser,et al.  Comparing two tests used for diagnostic or screening purposes , 1998 .

[22]  G. Foody.  Thematic map comparison: Evaluating the statistical significance of differences in classification accuracy , 2004 .

[23]  Stephen V. Stehman,et al.  Design and Analysis for Thematic Map Accuracy Assessment: Fundamental Principles , 1998 .

[24]  Farrokh Nadim,et al.  Statistical modelling of Europe-wide landslide susceptibility using limited landslide inventory data , 2012, Landslides.

[25]  J. R. Landis,et al.  The measurement of observer agreement for categorical data. , 1977, Biometrics.

[26]  P. Frattini,et al.  Comparing models of debris-flow susceptibility in the alpine environment , 2008 .

[27]  Jean-Philippe Malet,et al.  Climate-physiographically differentiated Pan-European landslide susceptibility assessment using spatial multi-criteria evaluation and transnational landslide information , 2014 .

[28]  H. A. Nefeslioglu,et al.  Application of logistic regression for landslide susceptibility zoning of Cekmece Area, Istanbul, Turkey , 2006 .

[29]  D. Varnes.  Landslide hazard zonation: A review of principles and practice , 1984 .

[30]  Lalit Kumar,et al.  Comparative assessment of the measures of thematic classification accuracy , 2007 .

[31]  B. Jiang.  Head/Tail Breaks: A New Classification Scheme for Data with a Heavy-Tailed Distribution , 2012, 1209.2801.

[32]  John C. Davis,et al.  Using multiple logistic regression and GIS technology to predict landslide hazard in northeast Kansas, USA , 2003 .

[33]  M. Charlton,et al.  Quantitative geography : perspectives on spatial data analysis by , 2001 .

[34]  Xuan Zhu,et al.  GIS for Environmental Applications: A practical approach , 2016 .

[35]  Arna Fariza,et al.  A comparison between natural and Head/tail breaks in LSI (Landslide Susceptibility Index) classification for landslide susceptibility mapping: A case study in Ponorogo, East Java, Indonesia , 2015, 2015 International Conference on Science in Information Technology (ICSITech).

[36]  T. Saaty,et al.  The Analytic Hierarchy Process , 1985 .

[37]  Melody Y. Kiang,et al.  A comparative assessment of classification methods , 2003, Decis. Support Syst..

[38]  Andrea G. Fabbri,et al.  Validation of Spatial Prediction Models for Landslide Hazard Mapping , 2003 .

[39]  C. Metz.  Basic principles of ROC analysis. , 1978, Seminars in nuclear medicine.

[40]  P. Peduzzi,et al.  Global landslide and avalanche hotspots , 2006 .

[41]  Jean-Philippe Malet,et al.  Integrating Spatial Multi-criteria Evaluation and Expert Knowledge for Country-Scale Landslide Susceptibility Analysis: Application to France , 2013 .

[42]  Saro Lee.  Comparison of landslide susceptibility maps generated through multiple logistic regression for three test areas in Korea , 2007 .

[43]  Steven McGee,et al.  Simplifying likelihood ratios , 2002, Journal of General Internal Medicine.

[44]  M. Zweig,et al.  Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine. , 1993, Clinical chemistry.

[45]  Hengxing Lan,et al.  A modified frequency ratio method for landslide susceptibility assessment , 2017, Landslides.

[46]  J. Fleiss,et al.  Statistical methods for rates and proportions , 1973 .

[47]  D. Roberts,et al.  Sources of error in accuracy assessment of thematic land-cover maps in the Brazilian Amazon , 2004 .