Mapping asbestos-cement roofing with hyperspectral remote sensing over a large mountain region of the Italian Western Alps

Our plain-English summary of Mapping asbestos-cement roofing with hyperspectral remote sensing over a large mountain region of the Italian Western Alps, by F. Frassy, G. Candiani, M. Rusmini, et al (Sensors, 2014).

The study

• Authors: F. Frassy, G. Candiani, M. Rusmini, P. Maianti, A. Marchesi, F. Rota Nodari, G. Dalla Via, C. Albonico, M. Gianinetto
• Journal: Sensors
• Published: 2014;14(9):15900-13
• Cited by: 13 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

The World Health Organization estimates that 100 thousand people in the world die every year from asbestos-related cancers and more than 300 thousand European citizens are expected to die from asbestos-related mesothelioma by 2030. Both the European and the Italian legislations have banned the manufacture, importation, processing and distribution in commerce of asbestos-containing products and have recommended action plans for the safe removal of asbestos from public and private buildings. This paper describes the quantitative mapping of asbestos-cement covers over a large mountainous region of Italian Western Alps using the Multispectral Infrared and Visible Imaging Spectrometer sensor. A very large data set made up of 61 airborne transect strips covering 3263 km2 were processed to support the identification of buildings with asbestos-cement roofing, promoted by the Valle d'Aosta Autonomous Region with the support of the Regional Environmental Protection Agency. Results showed an overall mapping accuracy of 80%, in terms of asbestos-cement surface detected. The influence of topography on the classification's accuracy suggested that even in high relief landscapes, the spatial resolution of data is the major source of errors and the smaller asbestos-cement covers were not detected or misclassified.

In plain English

This was a remote sensing study, not a health study. The authors used an airborne hyperspectral scanner (MIVIS), flown over the Valle d'Aosta Autonomous Region in the Italian Western Alps in fall 1999, to map asbestos-cement roofs. Sixty-one flight lines were processed with a classification method based on ten roofs found in an earlier field survey. Staff of the regional environmental protection agency then checked buildings on the ground from January 2010 to September 2011. The results table counted 1,942 roof outcomes: 833 correctly classified (43%), 928 missed (48%) and 181 wrongly flagged (9%). For roofs larger than 144 m2, correct detection rose to 75%. Measured by roof surface area, 80% was correctly detected and 15% missed. Most smaller roofs were missed, so the method tends to underestimate. Altitude, slope and aspect showed weak correlations with accuracy, which the authors linked to where large roofs were located. The authors concluded that mountain topography did not significantly affect accuracy and that data resolution was the major source of errors. They called the time gap between flight, processing and field surveys a weakness.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Airborne hyperspectral remote sensing mapping study with field validation. Asbestos-cement roofs were classified with the Spectral Angle Mapper algorithm, using ten roofs from a field survey as reference samples.
• Who was studied: Buildings with asbestos-cement roofs in the Valle d'Aosta Autonomous Region (3263 km2), North-West Italy. Ground validation was carried out from January 2010 to September 2011 by technical staff of ARPA Valle d'Aosta, who checked all buildings in the region.
• Exposure: Asbestos-cement roofing (covers). The article names no specific fiber type. Roofs were mapped from a MIVIS airborne survey flown at the beginning of fall 1999; 61 of 70 flight strips were used.
• Main results:
  • Counting roof units over the whole data set, Table 2 lists 1942 outcomes in total: 833 correctly classified, 181 commission errors (false positives) and 928 omission errors (false negatives). The authors described this as almost half of ground-surveyed roofs missed and fewer than 10% misclassified.
  • For roofs larger than 144 m2, compared with all roofs counted by unit: correct classification rose from 43% to 75%, commission error fell from 9% to 6%, and omission error fell from 48% to 19%.
  • Measured by roof surface area, correct classification was 80% and omission error was 15%.
  • Topography showed weak correlations with accuracy: R2 of 0.55 for altitude and 0.52 for slope. The share of roofs larger than 144 m2 correlated more strongly with accuracy: R2 of 0.80 by altitude class, 0.82 by slope class and 0.94 by aspect class.
• The authors' conclusion: The authors concluded that in a mountainous area topographic features do not significantly affect mapping accuracy, and that the spatial resolution of the data is the major source of errors.
• Limitations the authors state:
  • The authors called the time gap between the flight, the data processing and the field surveys a weakness that complicated the analysis.
  • Smaller roofs were mostly missed, so the method tends to underestimate asbestos-containing roofs.
  • Some roofs detected in 1999 had disappeared by the time of validation, either because they had been removed or because the buildings were affected by the 2000 flood.
  • Flight lines over the large area had differing spatial resolution and differing atmospheric and illumination conditions, which made processing the whole data set difficult.

Topics

Medical terms in this study, explained in our Lexicon: Asbestos-Cement, Mesothelioma.

Related studies

• Remote sensing for risk mapping of Aedes aegypti infestations: Is this a practical task? — C. Lorenz, F. Chiaravalloti-Neto, et al. (2020)
• Characterization of the dust/smoke aerosol that settled east of the World Trade Center (WTC) in lower Manhattan after the collapse of the WTC 11 September 2001 — P.J. Lioy, C.P. Weisel, et al. (2002)
• Estimation of the amount of asbestos-cement roofing in Poland — E. Wilk, M. Krówczyńska, et al. (2017)
• Health and environmental consequences of the world trade center disaster — P.J. Landrigan, P.J. Lioy, et al. (2004)
• Potential toxicity of nonregulated asbestiform minerals: balangeroite from the western Alps. Part 1: Identification and characterization — C. Groppo, M. Tomatis, et al. (2005)
• Simulation tests to assess occupational exposure to airborne asbestos from artificially weathered asphalt-based roofing products — P. Sheehan, F. Mowat, et al. (2010)

Browse more research in Sensors, or return to medical research.