Normalized Landslide Index Method for susceptibility map development in El SalvadorNat Hazards


Ari J. Posner, Konstantine P. Georgakakos
Earth and Planetary Sciences (miscellaneous) / Atmospheric Science / Water Science and Technology



Normalized Landslide Index Method for susceptibility map development in El Salvador

Ari J. Posner1 • Konstantine P. Georgakakos1,2

Received: 18 June 2014 / Accepted: 7 August 2015  Springer Science+Business Media Dordrecht 2015

Abstract El Salvador and Central America in general are highly prone to landsliding. In

November 1998, Hurricane Mitch killed 240 people, displaced about 85,000 people, caused more than $600 million in economic losses, and damaged about 60 % of the nation’s roads (Rose et al. in Natural hazards in El Salvador. Geologic Society of America of special paper 375, Boulder, 2004). An understanding of susceptibility of locations to landsliding is critical for development and mitigation planning. This work presents the development of the Normalized Landslide Index Method which is a derivative of the bivariate statistical methods commonly used in landslide susceptibility assessment. The resultant map was amended through a tangential analysis, also commonly used in landslide susceptibility mapping, the Analytical Hierarchy Process (AHP), which reduces multicriteria analysis to pair-wise comparisons. The assimilation of results from the AHP analysis into the statistically derived susceptibility map skewed the original results by emphasizing the extremes already found. It was determined that addition of AHP results did not increase the value of the derived susceptibility map. Finally, a physically based a priori approach to landslide susceptibility mapping, developed by El Salvador National

Service of Territorial Studies, was compared to the statistically derived map developed herein. It was found that the a priori approach was not sufficiently discriminant to be useful for planners and regulators, as very large areas were designated high susceptibility that included areas with low slope angles. The development of the normalized landslide index is a significant improvement to the class of bivariate statistical strategies to assess regional landslide susceptibility.

Keywords Landslide susceptibility  Analytical Hierarchy Process  Landslide index method  Bivariate statistics  El Salvador & Ari J. Posner 1 Hydrologic Research Center, 12555 High Bluff Drive, Suite 255, San Diego, CA 92130, USA 2 Scripps Institution of Oceanography, University of California at San Diego, La Jolla, CA 92093,

USA 123

Nat Hazards

DOI 10.1007/s11069-015-1930-4 1 Introduction

Shallow landslides are triggered by rainfall events of high intensity and short duration.

Although landslides may occur in small portions of a hilly or mountainous landscape, their impact to the safety and well-being of inhabitants and the local economy is disproportionately felt (Meisina et al. 2013). The necessity to identify landslide-prone areas at the regional scale is critical to plan mitigation measures and to alert populations during extreme rainfall events.

Landslide susceptibility and/or hazard mapping has evolved dramatically with the widespread availability of geographic information system technology and the availability of spatial datasets including land classification and other information from remote sensing.

Methodologies for the development of landslide susceptibility maps can be divided into statistical methods that use the locations of existing landslides along with land classification databases for statistical analysis and physically based methods that predict landslide susceptibility independent of the locations of previous landslide events and rely solely on land, climate, and in some cases seismic activity geo-spatial classifications (Van Den

Eeckhaut et al. 2006). There are advantages and disadvantages to each strategy. Statistical models are primarily limited due to the fact that the conceptual model on which they are founded is that ‘‘the past (and present) landslide locations are the key to the future’’ (Carrara et al. 1995; Zeˆzere 2002). The accuracy and precision of physically based methods are highly dependent upon the detail of available climatologic, hydrologic, and geomorphologic temporally and spatially distributed datasets.

Approaches to landslide hazard/susceptibility mapping have evolved and now include bivariate, multivariate logistic regression, fuzzy logic, and artificial neural network analysis (van Westen 1997; Dai et al. 2001; Lee and Min 2001; Ercanoglu and Gokceoglu 2004; Lee et al. 2004a, b; Komac 2006). A promising strategy comes from the Wharton School of

Business in the form of a decision-making tool known as the Analytical Hierarchy Process (Saaty 1980). Yalcin (2008) compared the use of Analytical Hierarchy Process (AHP) with two other commonly used methods for landslide susceptibility mapping: the statistical index and waiting factor methods, and found that the Analytical Hierarchy Process did significantly better than the statistical methods employed at predicting the locations of known landslides based on topographic and geomorphologic features.

This paper describes the development of a statistical-based landslide susceptibility map, which is subsequently refined using subjective criteria developed through use of the AHP.

The susceptibility map weights are compared both before and after implementation of

AHP, in order to assess the impact of its assimilation. Then, the statistically based developed susceptibility map is compared to a physically based susceptibility map that was developed by the El Salvador National Service of Territorial Studies (SNET). 2 Study area

The region of Central America is very prone to natural hazards (Rose et al. 2004). El

Salvador is located on the western side of Central America, and it is bounded by the Pacific

Ocean to the west, the Rı´o Sumpul to the east on its border with Honduras, the Rı´o Paz on the northern border with Guatemala, and the Rı´o Goascora´n on the southern border with

Nicaragua. Despite being Central America’s smallest country in area (21,040 km2), El

Salvador has the highest population density (*290 persons/km2).