An experience on modelling-based assessment of the air quality within the Air Quality Directive framework Ana Isabel Miranda, Isabel Ribeiro, Patrícia Fernandes, Alexandra Monteiro, Cristina Monteiro, Carlos Borrego
2008 Air Quality Directive| air quality assessment Assessment strategy depends on upper and lower assessment thresholds Those fixed measurements may be supplemented by modelling techniques and/or indicative measurements to provide adequate information on the spatial distribution of the ambient air quality. Fixed measurements shall be used Upper assessment threshold Combination of fixed measurements and modelling techniques and or indicative measurements may be used Concentration Lower assessment threshold Modelling techniques or objective-estimation shall be sufficient SO2, NO2, NOx, PM10, PM2,5, Pb, C6H6, CO
Upper and lower thresholds exceedances for 2006-2010 (5 years period). Assessment for 2010 and 2011 Upper and lower thresholds exceedances for 2006-2010 (5 years period). Exceedances of upper and lower assessment thresholds shall be determined on the basis of concentration during the previous 5 years where sufficient data are available. An assessment threshold shall be deemed to have been exceeded if it has beed exceed during at least 3 separate years out of those previous five years.
AQ assessment based on a combination of Modelling and Measuring values The approach Modelling Monitoring NO2, O3, PM10, PM2.5, SO2, CO, C6H6. 1. Monitoring stations selection and data treatment, for the period 2006-2010 2. Comparison with the upper and lower thresholds, for every pollutant 3. 2010 data treatment for the model evaluation 1. Model application to Portugal (5 km x 5 km), 2010 and 2011 2. Bias correction based on the multiplicative ratio adjustment technique 3. Evaluation (using the DELTA tool when possible) AQ assessment based on a combination of Modelling and Measuring values
Meteorological conditions MM5-EURAD Emissions Meteorological conditions
Simulation domains 125x125 km2 25x25 km2 5x5 km2
Emissions EMEP 2008 for the larger domains Portuguese inventory 2008 for the portuguese domain Industrial processes Transports
Monitoring stations
Bias-correction techniques We started to compare… SUBST an additive correction of the mean bias RAT a multiplicative ratio correction First we start to compare two simple bias-correction: one using na additive correction and a multiplicative ratio, both tested using different period of previous days (3-4 and 7 days) to correct the forecasted bias
Bias-correction techniques RAT & SUBST (Borrego et al., 2011, Atmospheric Environment) Bias-correction techniques RAT & SUBST PM10 O3 What we found is that ratio adustment using previous days was the most successful technique, comparing to the others… In this sense, RAT will be focus of our work after BIAS correction, model results have a decrease > 70% on the average systematic error the multiplicative ratio: better correction technique synoptic conditions are characterized by a 3-4 day period.
RDE = 15%; R = 0.63; bias = -1.01 µg.m-3; MSE = 4.31 µg.m-3 Validation SO2 Hourly values – Calendário 2010 measured modelled RDE = 15%; R = 0.63; bias = -1.01 µg.m-3; MSE = 4.31 µg.m-3
SO2 4th maximum of the daily averages (protection of human health) Upper Lower LV A excedência dos limiares de avaliação superior e inferior deve ser determinada a partir das concentrações dos cinco anos anteriores, caso se encontrem disponíveis dados suficientes (eficiência mínima de 85%). Considera -se que um limiar de avaliação foi ultrapassado se tiver sido excedido em, pelo menos, três desses cinco anos. Avaliação para os anos 2010 e 2011 Análise excedências LAI e LAS com base nos anos 2006 a 2010 Zones and agglomerations Threshold values were not exceeded 14
SO2 4th maximum of the daily averages (protection of human health) < LAT LAT-UAT > UAT It was not possible to have the needed data everywhere 2010 2011
SO2 Annual average winter period (ecosystem protection) 2010 2011 16 < LAT LAT-UAT >UAT 2010 2011 16
SO2 GIS Annual average winter period (ecosystems protection) Modelling results CORINE Land Cover 2006 GIS 17
SO2 25th maximum hourly value LV 2010 2011 18
Validation NO2 Hourly data
Validation NO2 Hourly data
Zones and agglomerations NO2 Annual mean Upper Lower LV Zones and agglomerations 21
NO2 Annual mean LV < LAT LAT-UAT > UAT 2010 2011 22
Population data at sub-municipality level NO2 Annual mean Modelling results Population data at sub-municipality level GIS National roads network
Zones and agglomerations NO2 19th maximum of the hourly averages (protection of human health) Upper Lower LV Zones and agglomerations 24
NO2 19th maximum of the hourly averages 2011 2010 25 LV < LAT LAT-UAT > UAT 2010 2011 25
NO2 19th maximum of the hourly averages
NO2 Number of hours exceeding the LV 2010 2011 27
Validation PM10 Daily average
Validation PM10 Daily average
PM10 Annual average LV < LAT LAT-UAT > UAT 2010 2011 30
PM10 Daily average LV < LAT LAT-UAT > UAT 2010 2011 31
Final comments This work was requested by the Portuguese Agency for the Environment. It was presented to the national agency and to the different regional entities in charge of the air quality assessment in Portugal. Several comments and feedbacks were received. They were very interested and willing to do a cost-benefit analysis of using modelling tools and reducing monitoring stations instead of keeping the maintenance costs they’re facing nowadays. We’re doing 2012 and we’re going to increase spatial resolution. Notwithstanding this modelling work no modelling-based report to the Comission was delivered.
But … There is a strong difficulty to trust models and people is afraid of using them, because: they were always working with AQ monitoring networks and that’s what they know they think models are a kind of “monster” and they do not provide a really added value February 2009
Attitude towards models changed and people are much more receptive to their use for air quality assessment. Thanks to FAIRMODE!!!!
Thank you very much!!! www.ua.pt/gemac miranda@ua.pt