Hyperspectral Discrimination of Asbestos‐Cement Roofing

Asbestos is characterised by a unique set of physical and chemical properties, which has led to its many applications [28], such as roofing coatings: flat and corrugated sheets, floors, water supply pipes, textiles, rope, cord and yarn, paper, friction and composition materials, household products, plastic fillers, clutches, brake linings, gaskets and pads for automobiles [30]. Asbestos‐cement roofing, and building construction materials represent the majority of currently used asbestos‐containing products in the world [5]. Exposure to asbestos and asbestos dust causes a wide range of diseases, such as mesothelioma, asbestosis, and lung cancer [16]. Since asbestos-containing products were proved to be harmful to human health, their usage and production was banned in 55 countries, including from January 1, 2005 all EU countries [6, 15, 20]. In Poland, in particular, asbestos production and trade was forbidden in 1997 [1, 15]. Nevertheless, there are still about 14.5 mln tons of asbestos containing products used across the country [25]; based on estimations performed on asbestos mineral imports data into Poland. Asbestos‐cement roofing discrimination forms the part of the classification issue of built‐up areas, including a detailed investigation of different types of roofing. Built-up areas are of interest of many researchers, both in relation to urban and rural areas. Ben-Dor et al. [3] acquired spectral library from 400 nm to 1100 nm of urban surfaces, and have discussed the significance of spectral regions in mapping urban areas, concluding that urban objects bear spectral fingerprints in the spectral region enabling reasonable spectral recognition. Marino et al. [21] have tested the MIVIS (Multispectral Infrared and Visible Imaging Spectrometer) airborne hyperspectral scanner to map asbestos concrete sheeting. Segl et al. [26] have proposed the incorporation of shape characteristics of urban surface materials in order to increase the reliability of buildings identification and to improve the results of spatial