Modelling the on-road traffic emissions from Catalonia (Spain) for photochemical air pollution research: weekday/weekend differences R. Parra & J. M. Baldasano Environmental Modelling Laboratory, Universitat Politècnica de Catalunya (UPC), Spain Abstract On-road traffic can vary between weekdays and weekends, both in timing and magnitude. It influences atmospheric emissions behaviour and tropospheric ozone concentrations. Giving priority to the bottom-up approach, we developed the high spatial (1 km2) and temporal (1 hour) resolution EMICAT2000 model, to estimate atmospheric emissions for the year 2000 from Catalonia (Spain). Its on-road traffic component was built with a high-resolution digital road-map and considered the fleet of vehicles, traffic intensity and profiles for weekdays and weekends; for every highway and the most important roads and urban streets (daily average traffic > 5000). Most of the emission factors were taken from the top-down COPERTIII model. We used specific Carbon Bond 4 profiles for NMVOC speciation. During the year 2000, emissions of NOx were 62.4 kt yr-1, VOCs 50.5 kt yr-1, CO 259.0 kt yr-1, SO2 1.3 kt yr-1, PST 15.7 kt yr-1. Emissions are mainly located on the Metropolitan Area of Barcelona and on the axis of highways following the coastline. Gasoline vehicles represent 70% of the park, but they emit ~57% of NOx and ~92% of VOC. Due to the drop of heavy duty vehicle traffic, NOx emissions on weekends are ~22% lower and emission of VOC are only slightly higher. Influences of these changes on ozone pollution could be deeply understood using a chemical transport model. Development of high spatial resolution emission is expensive both in time and resources, but the benefits and potential uses completely justify the investment. Keywords: traffic, emission, speciation, weekend effect, Catalonia. Air Pollution XII, C. A. Brebbia (Editor) © 2004 WIT Press, www.witpress.com, ISBN 1-85312-722-1 4 Air Pollution XII 1 Introduction Ozone is a secondary air pollutant whose behaviour is difficult to understand and control. It is formed by photochemical reactions of nitrogen oxides (NOx) and volatile organic compounds (VOC). Also, ozone could be transported from the stratosphere. Its measures at any given location may have contributions from different sources as background concentration, long-range and regional transport, or local photochemistry. On-road traffic, among other sources, is recognized as a significant source of ozone precursors. In Catalonia (Spain), located in the NE of the Iberian Peninsula, more than 1390 cases of excedances in the legislative hourly concentration threshold for people information (180 µg m-3) were recorded in the period 1991 – 2002, and almost 80% of these happened into the period June – August. Mathematical modelling could be used to explain this source of pollution (Rusell and Dennis [7]) using a chemical transport model (CTM). Before this activity, it is necessary to know the magnitude, superficial and temporal distribution of ozone precursors. Focussing in this objective for Catalonia, we developed the first version of the EMICAT2000 emission model. It includes the biogenic, on-road traffic and most important industrial sources. Previously, and toward on-road traffic emission, Delgado et al. [3] made the most important work, presenting results for the year 1994. Being both on-road traffic and knowledge under continuous changes over time, (e.g. park vehicles evolution, emission level reductions, more reliable emission factors, more developed emission models), on-road traffic is one of the sources demanding continuously improvements and updating. We describe the model basis and results obtained using the on-road traffic emissions component actually built into EMICAT2000. 2 Method EMICAT2000 was built taking the year 2000 as reference. Primary pollutants included are NOx, VOC, carbon monoxide (CO), sulphur dioxide (SO2) and total suspended particles (PST). 2.1 Mathematical model Using the bottom-up approach, we built an updated digital map of all the highways and most important roads and streets. Using high spatial (territory was divided with cells of 1 km2) and temporal (1 hour) resolutions; we included three kinds of emissions: (1) hot exhaust emissions, occurring under thermally stabilised engine and exhaust after treatment conditions, (2) cold exhaust emissions, occurring during transient thermal engine operation (cold start), and (3) evaporative emissions, non-exhaust VOC emission relevant for gasoline fleet. Also, we included non-exhaust particles emissions (tire and brake wear, road abrasion). Air Pollution XII, C. A. Brebbia (Editor) © 2004 WIT Press, www.witpress.com, ISBN 1-85312-722-1 Air Pollution XII 5 2.1.1 Hot exhaust emissions They were estimated using equation (1): E ihot r (k, d) = n ∑ Clf.Crd.DAT (k).L (k).F rj r ihot (s r ) j (1) j=1 Parameters: road section (urban, rural or highway type) into the cell k. pollutant (NOx, COV, CO, SO2, PST, CO2, CH4 y N2O). vehicle category total vehicle type number considered (36 for Catalonia). Term: E ihot hot exhaust emission of the pollutant i during one day r (k, d) : (weekday or weekend) in the road section r. It is expressed in g d-1. Data: ihot Fj (s r ) : hot emission factor of the pollutant i, for the vehicle category r: i: j: n: L r (k) : DATrj (k) : j (g km-1). It is a function of the typical speed sr (km h-1). length of the road section r (km). daily average traffic of the category j on the road type r (number o vehicles j per day) Crd : relation between daily traffic for a specific month and DAT. Clf : coefficient for daily traffic (weekday or weekend). Hourly emissions profiles were estimated using equation (2): Crh ihot E ihot (2) .E r (k, d) r (k, h) = 100 Parameter: h: hour (0,1,...........23). Term: E ihot hot emission of the pollutant i during the hour h (g h-1). r (k, h) : Data: C rh : percentage of the traffic during the hour h in relation to the daily traffic. Monthly emissions were obtained adding up the respective daily values. The same procedure was used for annual emission. 2.1.2 Cold exhaust emissions They seem to be most likely for urban driving. They occur for all vehicle categories, but emission factors now can be reasonable estimated for gasoline and diesel passenger cars (Ntziachristos and Samaras [8]). Hourly cold emissions were established as additional emissions using equation 3: n Fjicold − E icold (k, h) = E ihot (at) 1 (3) r r (k, h).β (ltrip, at). F ihot j=1 j ∑ Air Pollution XII, C. A. Brebbia (Editor) © 2004 WIT Press, www.witpress.com, ISBN 1-85312-722-1 6 Air Pollution XII Term: E icold (k, h) r : hourly cold emission of the pollutant i (only for urban roads) (g h-1). Data: β: fraction of the mileage driven with cold engines or catalyst operated below the light-off temperature. It is established as function of the average trip length (ltrip) and the ambient temperature (at). Fjicold Fjihot : cold over hot ratio of pollutant i emissions. 2.1.3 Evaporative VOC emissions There are three primary sources of evaporative emissions. (1) diurnal (daily) emissions, associated with daily variation in ambient temperature that result in vapour expansion inside the gasoline tank; (2) hot soak emissions, caused when a hot engine is turned off, increasing the temperature of the fuel (as into the carburettor) no longer flowing; and (3) running losses, that result of vapour generated in gasoline tanks during vehicle operations. Total daily diurnal emissions were estimated using equation (4): E devap (daily) = N m .(e dm (t min , t max , RVP)) m (4) Parameter: m: number of vehicle categories (gasoline passenger cars with and without control, motorcycles < 50 cc and motorcycles > 50cc). Term: E devap (daily) : m total daily VOC diurnal evaporative emission produced by the vehicles of the category m (g d-1). Data: Nm: number of vehicles of the fleet of Catalonia belonging to the category m e dm : diurnal emission factor (g VOC d-1) for vehicles of the category m. It is a function of the minimum/maximum ambient temperatures (tmax, tmin) and the gasoline volatility (measured by its reid vapour pressure (RVP). Total daily hot soak emissions were estimated using equation (5): swarm E sevap (daily) = N m .(p.x m .e shot (at, RVP)) m m (at, RVP) + w.x m .e m Term: E sevap (t) : total daily VOC hot soak emission (g d-1). m (5) Data: p: fraction of trips finished with hot engine. It depends on the average monthly ambient temperature. Air Pollution XII, C. A. Brebbia (Editor) © 2004 WIT Press, www.witpress.com, ISBN 1-85312-722-1 Air Pollution XII 7 w: fraction of trips finished with cold or warm engine or with catalyst below its light-off temperature. xm: mean number of trips of the vehicles type m. shot e m (at, RVP) : emission factor for hot soak emissions (g trip-1). e swarm (at, RVP) : emission factor for cold and warm soak emissions (g trip-1). m Last two emission factors depend on the ambient temperature and on gasoline volatility (RVP). Total hourly diurnal and soak emissions were established, dealing the daily value with averaged hourly values toward the average daily ambient temperature. Hourly diurnal emissions were spatially disaggregated according to the traffic volume of each vehicle category and population density. Running losses emission factors were estimated using equation (6). Daily and hourly running losses were estimated using equivalents equations as the hot exhaust emissions. rwarm Fjrevap = (p.x j .e rhot (at, RVP)) j (at, RVP) + w.x j .e j (6) Parameter: j: vehicle category. The same considered for cold emissions. Term: Fjrevap : running losses VOC emission factor (g km-1). Data: e rhot j (at, RVP) : emission factor for hot running losses (g km-1) e rwarm (at, RVP) : emission factor for warm running losses (g km-1). j Last two emission factors depend on the ambient temperature and on gasoline volatility (RVP). Equations (4) to (6) were taken from the top-down model COPERTIII (Ntziachristos and Samaras [8]), developed for be used in Europe. 2.2 Base information Daily average traffic information, vehicle park composition by type kind of road (differencing between weekday and weekend profiles), average speeds, temperature information, timing traffic profiles (monthly, daily and hourly) and fuel properties; were collected from different sources. Exhaust emission factors for NOx, VOC, CO, PST (for diesel vehicles), were obtained from Ntziachristos and Samaras [8]. Exploitation of these and other information required by the equations, and modelling hypothesis are fully described in Parra [5]. 3 Results 3.1 Annual emissions Table 1 shows the contribution on fleet, mileage driven and ozone precursors emission. Total mileage driven of the EMICAT2000 road net adds up 32 202 Air Pollution XII, C. A. Brebbia (Editor) © 2004 WIT Press, www.witpress.com, ISBN 1-85312-722-1 8 Air Pollution XII millions km yr-1. NOx were emitted mainly by duty diesel vehicles (37%), gasoline passenger cars without catalyst (30%) and heavy duty gasoline cars (20%). VOC were emitted mainly by gasoline passenger cars without catalyst (38%), motorcycles (30%) and heavy duty gasoline cars (14%). Figure 1 shows the structure of the annual emission of primary pollutants: 62.4 kt yr-1 of NOx (16%), 50.5 kt yr-1 of VOC (13%), 259 kt yr-1 of CO (66.7%), 1.3 kt yr-1 of SO2 (0.3%) and 15.7 kt yr-1 (4%) of PST. Table 1: Contribution on fleet, mileage driven and ozone precursors emission from on-road traffic in Catalonia for the year 2000. Mileage driven Type of vehicle -1 NOx VOC -1 -1 Fleet (%) (Mkm yr ) (%) (t yr ) (%) (t yr ) (%) Gasoline passenger cars without catalyst Gasoline passenger cars with catalyst 24 6 987 22 18 555 30 19 095 38 33 7 252 23 4 104 7 5 020 10 Diesel passenger car 17 6 358 20 3 722 6 1 034 2 Heavy duty gasoline vehicles 4 1 533 5 12 577 20 7 051 14 37 Duty diesel vehicles 12 8 095 25 23 181 Motorcycles 10 1 978 6 270 Total: 100 32 202 3 110 6 15 201 30 100 62 409 100 50 511 100 Figure 1: On-road traffic annual emission by type of vehicles of primary pollutant in Catalonia during the year 2000. Heavy duty gasoline vehicles and motorcycles contribute with low mileage driven percents (5 and 6%, respectively), but their emission factors are high, providing important contributions of primary pollutants (31 and 18%). Gasoline vehicles mean 71% of the fleet, but they emit 57% of NOx and 92% of VOC. Air Pollution XII, C. A. Brebbia (Editor) © 2004 WIT Press, www.witpress.com, ISBN 1-85312-722-1 Air Pollution XII 9 3.2 Weekday – weekend emission profiles At weekend there is a reduction of traffic from heavy duty vehicles (60% in average), and also variations of traffic on roads, both imply differences in emission profiles. Figure 2 depicts the evolution of ozone precursors emissions during weekday and weekend of August. Both NOx and VOC profiles are bimodal. At weekday, NOx peaks occur about 9:00 and 18:00, but at weekend first peak is lower and displaced to the right (about 12:00). Total NOx emission on weekend is 22% lower than weekday. For VOC, similar NOx timing is observed, but first peaks have equivalent magnitudes and the second peak of weekend is higher. Total VOC emissions at weekend are slightly higher (4%) than weekday. Figure 2: Hourly emission of ozone precursors during weekday and weekend of August from on-road traffic in Catalonia. Figure 3 shows the position of Catalonia in Europe, the distribution of NOx emission for weekday and the difference between weekday and weekend. In cells, the relation of emission values NOx/VOC for weekday could rise until 2.7, but for weekend, it is no larger than 1.5. Also, for weekend CO and PST reductions were 13 and 12% respectively. Air Pollution XII, C. A. Brebbia (Editor) © 2004 WIT Press, www.witpress.com, ISBN 1-85312-722-1 10 Air Pollution XII Figure 3: Location of Catalonia and distribution of NOx emission for weekday and weekend of August from on-road traffic in Catalonia. At weekend, NOx emission is 22% lower than weekday, mainly due to drop in traffic of heavy duty vehicles. Emissions are mainly located on the Metropolitan Area of Barcelona and on the axis of highways following the coastline, precisely in front of the sea breeze (Baldasano et al. [1]). 4 VOC speciation Vehicles were classified in 5 groups. Using profiles provided by Ntziachristos and Samaras [8], and according to the categories of the Carbon Bond 4 chemical mechanism, we obtained the profiles of the Table 2. Those are actually built into Air Pollution XII, C. A. Brebbia (Editor) © 2004 WIT Press, www.witpress.com, ISBN 1-85312-722-1 Air Pollution XII 11 EMICAT2000, which also produces the data files using the netCDF interface for the third generation Models-3 Air Quality Model System. Table 2: Profiles used in EMICAT2000 for Carbon Bond 4 species (mol h-1) due to 1 g h-1 NMVOC emitted by on-road traffic. Cat. (1) (2) (3) (4) (5) PAR ETH OLE 0.0191370 0.0291000 0.0633813 0.0454749 0.0523800 0.0013200 0.0010673 0.0000000 0.0015391 0.0008834 0.0021989 0.0017019 0.0009258 0.0016681 0.0016143 Carbon Bond 4 species TOL XYL 0.0027566 0.0020294 0.0001424 0.0001894 0.0000718 0.0023353 0.0020733 0.0000712 0.0002820 0.0003768 FORM ALD2 NR 0.0003152 0.0002486 0.0000000 0.0016836 0.0010586 0.0007971 0.0007793 0.0014243 0.0029799 0.0022116 0.0076106 0.0057482 0.0007121 0.0036278 0.0019215 (1) Exhaust emission of conventional vehicles: passenger cars without catalyst, heavy duty gasoline cars and motorcycles. (2) Exhaust emission of gasoline passenger cars with catalyst. (3) Evaporative emissions of gasoline passenger cars and motorcycles. (4) Exhaust emission of diesel passenger cars. (5) Exhaust emissions of heavy duty diesel vehicles. 5 Discussion We developed the on-road traffic component of the EMICAT2000 emission model prioritizing the bottom-up approach for hot emissions. Nevertheless, being very difficult to get all the information for each cell of the domain, or due to the methodologies themselves, for cold and evaporative emissions we used the topdown approach. Main applications are related with photochemical pollution modelling and dynamic of the atmosphere. Considering the morphology of Catalonia and its complex wind fields, high spatial resolution of emissions become necessary. Successful results are being obtained (Barros et al. [2]). Differences and similarities on ozone precursors emissions were found between weekday and weekend, both in magnitude and timing. Their effects originate the phenomenon knew as the “weekend effect” that nowadays is demanding proof analysis, particularly in U.S. Detailed studies indicate that drop in NOx emissions mean higher ozone concentrations on weekends in a VOC limited regime (as urban zones). In contrast, lower ozone is probably produced in locations with NOx limited regime (Heuss et al. [4]). The first kind of this research is under developing for Catalonia using the emissions data provided by EMICAT2000. To identify the kind of regime (NOx or VOC limited) it is essential to employ high spatial resolution on emissions. Implementation of the emission model was both time-and resourcesconsuming. Despite it is not complete (every emission source nowadays is not included), but with biogenic (Parra et al. [6]), on-road traffic and most important industrial emissions, the principal emitter features of Catalonia are described. The gradual inclusion of minor sources as small pollutant industries, aircraft and ship traffic, station services or secondary roads (with low daily average traffic), will contribute mainly in details. Nowadays, Models-3 is one of the most powerful air quality models, under constant improvement. It integrates different U.S. emission models, as GLOBEIS for biogenic emissions or MOBILE for traffic emissions, using an Air Pollution XII, C. A. Brebbia (Editor) © 2004 WIT Press, www.witpress.com, ISBN 1-85312-722-1 12 Air Pollution XII auxiliary processor called SMOKETOOL. These emission models are not adequate for being used directly in other regions. Being EMICAT2000 developed explicitly for Catalonia and proved with Models-3, it provides total control, freedom and flexibility on the emission topic. It could be easily applied and exploited for other regions with similar characteristics On-road traffic model into EMICAT2000 could be the basis for forecasting the future air quality of Catalonia over different scenarios, as probably or hypothetical changes in fleet vehicles or new quality of fuels. The emission model demands a bigger effort between the elements required for air quality modelling, but the benefits and potential applications are obvious. References [1] Baldasano J.M., L. Cremades and C. Soriano (1994) Circulation of Air Pollutants over the Barcelona Geographical Area in Summer. Proceedings of Sixth European Symposium Physico-Chemical Behaviour of Atmospheric Pollutants. Varese (Italy), 18-22 October, 1993. Report EUR 15609/1 EN: 474-479. [2] Barros N., I. Toll, C. Soriano, P. Jiménez, C. Borrego and J. M. Baldasano (2003) Urban Photochemical Pollution in the Iberian Peninsula: Lisbon and Barcelona Airsheds. Air&Waste Manage. Assoc, 53, 347-359. [3] Delgado R., I. Toll, C. Soriano and J.M. Baldasano (2000) Vehicle Emission Model of Air Pollutants from Road Traffic. Application to Catalonia (Spain). Edts. J.W.S. Longhurst, C. Brebbia and H. Power. Air Pollution VIII. WIT press: 379-388 [4] Heuss, J., Kahlbaum, D. & Wolf, G. (2003). Weekday/weekend ozone differences: what can we learn from them?. Air & Waste Manage. Assoc, 53, 772 – 788. [5] Parra, R. (2004). Development of the EMICAT2000 emission model for photochemical dispersion modelling. Application for Catalonia (Spain). PhD Thesis, Technical University of Catalonia, Barcelona (Spain). [6] Parra, R., Gassó, S. & Baldasano, J.M. (2004). Estimating the biogenic emissions of non-methane volatile organic compounds of the North Western Mediterranean vegetation (Catalonia – Spain). The Science of the Total Environment (In press). [7] Rusell, A. & Dennis, R. (2000). NARSTO Critical Review of Photochemical Models and Modeling. Atmospheric Environment, 34, 2 283– 2 324. [8] Ntziachristos, L. & Samaras, Z. (2000). COPERTIII Computer programme to calculate emissions from road transport. Methodology and emission factors (Version 2.1). European Environment Agency. Technical report No 49. Air Pollution XII, C. A. Brebbia (Editor) © 2004 WIT Press, www.witpress.com, ISBN 1-85312-722-1