ETC/ATNI Report 18/2021: Benzo(a)pyrene (BaP) annual mapping. Evaluation of its potential regular updating.

The report examines the potential regular production of benzo(a)pyrene (BaP) maps at the European scale in line with the operational production of other air quality maps. Stations measuring BaP are relatively scarce at the European scale, so in order to extend the spatial coverage, so-called pseudo station data have been calculated and used together with the actual BaP measurement data. These pseudo station data are derived from PM2.5 or PM10 measurements in locations with no BaP observations.

21 Jan 2022

Jan Horálek, Markéta Schreiberová, Philipp Schneider

Prepared by:

Jan Horálek, Markéta Schreiberová, Czech Hydrometeorlogical Institute (CHMI, Czechia)

Philipp Schneider, NILU - Norwegian Institute for Air Research (NILU, Norway)

The report explores potential regular production of European-wide benzo(a)pyrene (BaP) maps. It does so based on 2018 and 2019 data. In order to obtain spatial estimates of BaP, datasets of station measurements have been combined with chemical transport modelling and other supplementary data using the Regression – Interpolation – Merging Mapping (RIMM) method.

A mapping approach using only actual BaP measurements (and supplementary data) has been examined. However, due to the poor spatial coverage of BaP measurement stations at the European scale, such maps show large areas with high interpolation uncertainty, mainly in rural regions, but to some extent also in urban areas. Trying to overcome this limitation, so-called pseudo BaP data have been introduced in areas with a lack of BaP stations. In such regions, estimates of BaP were computed based on PM2.5 measurements (or with pseudo PM2.5 data based on PM10 measurements), using exponential regression of the observed BaP concentrations with the PM2.5 data, as well as geographical coordinates and land cover as predictor variables.

Based on both actual BaP measurements and pseudo BaP data, maps of annual average BaP concentration have been prepared for all of Europe. The uncertainty of the final maps was evaluated by leave-one-out cross-validation and by the interpolation relative standard uncertainty. The cross-validation uncertainty has been estimated at a level of about 120% (for 2018) and 100% (for 2019) in rural areas and about 70% (in both years) in urban areas. The interpolation uncertainty has been estimated to be below 60% (in both years) in the majority of the study region. Based on the results, the report recommends the regular production of BaP maps. However, due to the relatively high cross-validation uncertainty (particularly in the rural areas), it is recommended to label them as experimental maps to indicate that they do not yet meet the same accuracy standards as the regularly produced maps of other pollutants.