Method validation for the identification of asbestos–cement roofing

Abstract

The aim of this multidisciplinary work is to assess the potentiality of remote sensing multispectral infrared visible imaging spectrometer (MIVIS) data classified for mapping asbestos–cement roofing. In order to validate the methodology, measurement were carried out on the ground in order to later verify the results between the processed data and reality. All roofs classified as asbestos–cement were then sampled and analysed by phase contrast optical microscopy and/or scanning electron microscopy. The average classification accuracy obtained corresponds to 89.1%, and the classification accuracy of the test pixels of asbestos–cement is equal to 94.3%. Only 5.7% of pixels were misclassified. Information about the presence of asbestos–cement in the studied area has been also collected. The asbestos–cement surfaces of buildings vary from 100 to 5,000 m2, totalling to 30,800 m2, which is approximately 400,400 kg of asbestos surfaces in an area of 5.2 km2. The integration of these techniques, resulting from both MIVIS data classification and the results provided by laboratory analyses of the roofs samples, in particular from those not detected by processing MIVIS data, allowed the validation and improvement of this method, and the possibility to develop researches specifically aimed at highlighting the state of alteration of asbestos-cement surfaces. Regardless of these encouraging results, further testing in different areas is still needed in order to improve the methodology developed.