Presenter

Mrs Aura Cecilia Alegría Caicedo - ETRO, VUB and ULB [Email]

Abstract

Landmines and explosive remnants of war (ERW) continue to affect the lives of people living in previously war-torn countries. A major challenge in humanitarian mine action (HMA) is finding new and integrated approaches for an efficient land release process where expensive clearance activities are avoided. This thesis proposes the use of geographic information systems (GIS) along with geo-statistical analysis and prediction in HMA, to support the land release process by efficiently mapping areas at risk.

Starting from landmines and ERW contamination information in terms of locations of known landmine accidents/incidents, and minefields or suspected hazardous areas, this thesis proposes three main methodologies for landmine risk mapping. The first approach allows identifying significant landmine clusters and spatial patterns, where an extended kernel density estimation algorithm is proposed, that considers the spatial configuration of the documented hazard areas, particularly, their shape. The second approach investigates the use of geographically weighted regression (GWR) to explore spatially varying relationships between recorded landmine and ERW hazard data and environmental characteristics linked to specific characteristics of the surveyed area in terms of human-activity related variables (e.g. trails and roads used for transportation) and environmental variables (e.g. topography and land cover) data. The third approach builds on species distribution modeling using recorded landmine hazard data and geographical and topographical data, and the maximum entropy (MaxEnt) modeling technique.

The proposed methods have been evaluated for landmine risk mapping using mine action data from Bosnia and Herzegovina (BIH) and Colombia.
Comparison of the landmine risk area predicted by the proposed approaches and the risk areas designated as confirmed or suspected hazard areas by the Bosnia and Herzegovina Mine Action Center (BHMAC) confirmed that the proposed models are able to successfully replicate the confirmed and suspected mine regions delineated by BHMAC in the BIH study area. For the Colombian study area, we do not have confirmed risk maps by the Colombian mine action authorities, to be used for comparison with our produced risk maps. However, we believe that the proposed methods can be used by the mine action authorities to assess the risk related to landmine accidents.