Using Decision Tree Analysis and GIS in Modelling (semi)VOC Emissions at the European Scale

Risk assessment of semi Volatile Organic Compounds (semi-VOCs) is a fundamental part in the regulation of production and use in industries and households. Emission inventories are a natural starting point in risk assessments and, given the complex use and emission patterns of the many thousands VOCs, emission estimates are often one of the most uncertain and problematic parts in risk assessments. Some critical issues are quantifying production and use amounts of chemicals and chemical containing products, assigning amounts to industrial activities and household products, identifying use and emission patterns, identifying receiving environmental compartment and quantifying the emission to these, which depend on production volumes, chemical properties, and their mode of use, in a non-trivial way. To ensure reliable risk assessments, emission estimates are sought which need to be realistic and, at the same time, do not require excessive effort in the modeling of emission inventories. The report proposes a method to capitalize on the information in the European Chemicals Bureau risk assessment reports (RARs), available for a limited number of chemicals, to train decision trees that allow estimating emissions of chemicals to different environmental compartments. The report also illustrates how these estimates can be used in conjunction with geographic information system (GIS) processing of spatial data to map emissions. Examples are drawn with reference to the case of the European Union. It is shown how quick, spatially distributed estimates of emissions to specific environmental compartments can be obtained to be used in screening level assessment. The method outlined in the report allows a quick and reliable estimation of the fraction of total chemical production that results in emission to a specific environmental medium, using data mining techniques and GIS. This can result helpful within the new procedures for risk assessments guided by REACH, as a way to exploit data from existing risk assessments for predicting and mapping emissions of chemicals that have not yet been assessed.

JRC.H.5 - Rural, water and ecosystem resources