A comparison of statistical methods and multi-criteria decision making to map flood hazard susceptibility in Northern Iran

Alireza Arabameri; Khalil Rezaei; Artemi Cerdà; Christian Conoscenti; Zahra Kalantari

doi:10.1016/j.scitotenv.2019.01.021

A comparison of statistical methods and multi-criteria decision making to map flood hazard susceptibility in Northern Iran

Sci Total Environ. 2019 Apr 10:660:443-458. doi: 10.1016/j.scitotenv.2019.01.021. Epub 2019 Jan 5.

Authors

Alireza Arabameri¹, Khalil Rezaei², Artemi Cerdà³, Christian Conoscenti⁴, Zahra Kalantari⁵

Affiliations

¹ Department of Geomorphology, Tarbiat Modares University, Tehran 36581-17994, Iran. Electronic address: Alireza.ameri91@yahoo.com.
² Faculty of Earth Sciences, Kharazmi University, Tehran 14911-15719, Iran.
³ Soil Erosion and Degradation Research Group, Departament de Geografia, Universitat de València, Blasco Ibàñez, 28, 46010, Valencia, Spain. Electronic address: Artemio.cerda@uv.es.
⁴ Department of Earth and Marine Sciences (DISTEM), University of Palermo, Palermo, Italy. Electronic address: christian.conoscenti@unipa.it.
⁵ Stockholm University, Department of Physical Geography and Bolin Centre for Climate Research, SE-106 91 Stockholm, Sweden. Electronic address: zahra.kalantari@natgeo.su.se.

PMID: 30640112
DOI: 10.1016/j.scitotenv.2019.01.021

Abstract

In north of Iran, flood is one of the most important natural hazards that annually inflict great economic damages on humankind infrastructures and natural ecosystems. The Kiasar watershed is known as one of the critical areas in north of Iran, due to numerous floods and waste of water and soil resources, as well as related economic and ecological losses. However, a comprehensive and systematic research to identify flood-prone areas, which may help to establish management and conservation measures, has not been carried out yet. Therefore, this study tested four methods: evidential belief function (EBF), frequency ratio (FR), Technique for Order Preference by Similarity To ideal Solution (TOPSIS) and Vlse Kriterijumsk Optimizacija Kompromisno Resenje (VIKOR) for flood hazard susceptibility mapping (FHSM) in this area. These were combined in two methodological frameworks involving statistical and multi-criteria decision making approaches. The efficiency of statistical and multi-criteria methods in FHSM were compared by using area under receiver operating characteristic (AUROC) curve, seed cell area index and frequency ratio. A database containing flood inventory maps and flood-related conditioning factors was established for this watershed. The flood inventory maps produced included 132 flood conditions, which were randomly classified into two groups, for training (70%) and validation (30%). Analytical hierarchy process (AHP) indicated that slope, distance to stream and land use/land cover are of key importance in flood occurrence in the study catchment. In validation results, the EBF model had a better prediction rate (0.987) and success rate (0.946) than FR, TOPSIS and VIKOR (prediction rate 0.917, 0.888, and 0.810; success rate 0.939, 0.904, and 0.735, respectively). Based on their frequency ratio and seed cell area index values, all models except VIKOR showed acceptable accuracy of classification.

Keywords: Environmental management; Kiasar watershed; Modelling; Natural hazard; Soil erosion.