engleski

The relevance, strength and likelihood of occurrence of the minefield indicators and signatures used in the airborne and spaceborne remote sensing of mine contaminated areas

The contamination by land mines and unexploded ordnance (UXO) is extremely difficult problem that requires engagement of all national resources of the afflicted country, and wide international support. Indeed the main problem is large contaminated area that is not in normal use, this produces very strong impact in the social, economic, political, humanitarian and other domains. In Croatia in 1996 was 12000 km2 suspected to be contaminated by landmines (assessment of the UN Mine Action Center Croatia), and in 2003 this area is 1600 km2. Experience in Croatia from 1996 to 2003 shows that only 10 % of the suspected area is really contaminated. Similar ratio of the contaminated and the suspected area was assessed in several other countries contaminated by landmines. Therefore follows conclusion that strategic problem for the countries that are contaminated by landmines is to reduce the suspected area urgently and proceed by demining in the sustainable manner. The remote sensing technology has potentials for this application, several international projects were realised or are under way, with aim to reduce suspected area (several European countries - Pilot project Masambique ; European Commission – PARADIS, ARC, SMART ; European Commission and USA - Satellite based GIS for the mine action ; private consortium – MineSeeker). The actual available airborne sensors cannot detect landmines that lay in the ground and are covered by soil and by wild vegetation for longer time (e.g. in Croatia from the year 1991). Several former scientific projects were aimed to detect landmines and UXO (Pilot project Masambique, MineSeeker), but this ill posed problem was redefined and following projects had more realistic and affordable goals. In this paper, we consider the use of the physical entities (working term - minefield indicators) that indicate presence of the contamination by landmines or absence of the contamination, the related features and the electromagnetic signatures. The identification of the minefield indicators started by field work in 50 km2 of the minefields and the suspected areas, in which different experts collected the relevant information, data and knowledge. This process continued in several iterations from 2001 to 2003, while the evolutionary analysis, approval and improvement were applied. The data, information and knowledge that were collected in real contaminated areas were compared with the data contained in imagery and data collected by the airborne passive and active sensors and with the contextual information, data and knowledge. The active sensor was four bands polarimetric Experimental SAR (DLR), while passive electro-optical sensors were multi-spectral scanner Daedalus (DLR), digital matrix camera in visible and near infrared bands (CROMAC), thermal infrared camera (CROMAC), hyper spectral line scanner for visible and near infrared bands (CROMAC). Several steps of the analysis of sensor images and data were applied: the enhancement of the features, extraction of the features, derivation of the signatures, ranking in accordance to the probability of the detection and their reliability (the confusion matrix and distance metrics). The analysis of the relevance, strength and likelihood of occurrence of the minefield indicators is the main result reported in the paper.

suspected area; remote sensing technology; minedfiled indicators; multispectral scanner DAEDALUS; hyperspectral line scanner; GIS

nije evidentirano