In this study, the performance of the source apportionment model applications were evaluated by comparing the model results provided by 44 participants adopting a methodology based on performance indicators: z-scores and RMSEu, with pre-established acceptability criteria. Involving models based on completely different and independent input data, such as receptor models (RMs) and chemical transport models (CTMs), provided a unique opportunity to cross-validate them. In addition, comparing the modelled source chemical profiles, with those measured directly at the source contributed to corroborate the chemical profile of the tested model results. The most used RM was EPA- PMF5. RMs showed very good performance for the overall dataset (91% of z-scores accepted) and more difficulties are observed with SCE time series (72% of RMSEu accepted). Industry resulted the most problematic source for RMs due to the high variability among participants. Also the results obtained with CTMs were quite comparable to their ensemble reference using all models for the overall average (>92% of successful z-scores) while the comparability of the time series is more problematic (between 58% and 77% of the candidates’ RMSEu are accepted). In the CTM models a gap was observed between the sum of source contributions and the gravimetric PM10 mass likely due to PM underestimation in the base case. Interestingly, when only the tagged species CTM results were used in the reference, the differences between the two CTM approaches (brute force and tagged species) were evident. In this case the percentage of candidates passing the z-score and RMSEu tests were only 50% and 86%, respectively. CTMs showed good comparability with RMs for the overall dataset (83% of the z-scores accepted), more differences were observed when dealing with the time series of the single source categories. In this case the share of successful RMSEu was in the range 25% - 34%.
C. A. Belis, D.P. (2018). Results of the first European Source Apportionment intercomparison for Receptor and Chemical Transport Models, EUR 29254 EN. Luxembourg : Publications Office of the European Union [10.2760/41815].
Results of the first European Source Apportionment intercomparison for Receptor and Chemical Transport Models, EUR 29254 EN
E. Venturini;
2018
Abstract
In this study, the performance of the source apportionment model applications were evaluated by comparing the model results provided by 44 participants adopting a methodology based on performance indicators: z-scores and RMSEu, with pre-established acceptability criteria. Involving models based on completely different and independent input data, such as receptor models (RMs) and chemical transport models (CTMs), provided a unique opportunity to cross-validate them. In addition, comparing the modelled source chemical profiles, with those measured directly at the source contributed to corroborate the chemical profile of the tested model results. The most used RM was EPA- PMF5. RMs showed very good performance for the overall dataset (91% of z-scores accepted) and more difficulties are observed with SCE time series (72% of RMSEu accepted). Industry resulted the most problematic source for RMs due to the high variability among participants. Also the results obtained with CTMs were quite comparable to their ensemble reference using all models for the overall average (>92% of successful z-scores) while the comparability of the time series is more problematic (between 58% and 77% of the candidates’ RMSEu are accepted). In the CTM models a gap was observed between the sum of source contributions and the gravimetric PM10 mass likely due to PM underestimation in the base case. Interestingly, when only the tagged species CTM results were used in the reference, the differences between the two CTM approaches (brute force and tagged species) were evident. In this case the percentage of candidates passing the z-score and RMSEu tests were only 50% and 86%, respectively. CTMs showed good comparability with RMs for the overall dataset (83% of the z-scores accepted), more differences were observed when dealing with the time series of the single source categories. In this case the share of successful RMSEu was in the range 25% - 34%.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
