scaledep! - emep · 6 2. d z } } o } p Ç! 2.1 simulations! each mode lling team carried out four...
TRANSCRIPT
1
ScaleDep
Performance of European chemistry-‐transport models as function of horizontal spatial resolution
C. Cuvelier, P. Thunis, D. Karam European Commission, DG Joint Research Centre Institute for Environment and Sustainability I-‐21020 Ispra (Va), Italy
M. Schaap, C. Hendriks, R. Kranenburg TNO Built Environment and Geosciences, Dept. of Air Quality and Climate P.O. Box 80015, NL-‐3508TA Utrecht, The Netherlands
H. Fagerli, A. Nyiri, D. Simpson, P. Wind, M. Schulz Air Pollution Section Research Department, Norwegian Meteorological Institute P.O. Box 43, Blindern, N-‐0313, Oslo, Norway
B. Bessagnet, A. Colette, E. Terrenoire, L. Rouïl Industriel et des Risques Parc Technologique, ALATA, F-‐60550 Verneuil-‐en-‐Halatte, France
R. Stern Freie Universität Berlin Institut für Meteorologie und Troposphärische Umweltforschung Carl-‐Heinrich-‐Becker Weg 6-‐10, D-‐12165 Berlin, Germany
A. Graff Umweltbundesamt, Postfach 1406 D-‐06813 Dessau-‐Roßlau, Germany J.M. Baldasano, M.T. Pay Barcelona Supercomputing Center c/ Jordi Girona 29, E-‐08034 Barcelona, Spain
2
The exercise showed that the model responses to an increase in resolution show a broadly consistent picture among all models.
The analysis showed that the grid size does not play a major role for air quality model calculations which are targeted on the determination of the background (non-‐urban) air quality. Downscaling model resolution does not change concentration estimates and model performance at rural and EMEP sites. It does not neither impact the performance of models regarding the capture of the seasonal variability of pollutants.
However, the grid resolution plays an important role in agglomerations characterized by high emission densities. The urban signal, i.e. the concentration difference between high emission areas and their surroundings, usually increases with decreasing grid size. This grid effect is more pronounced for NO2 than for PM10, because a large part of the urban PM10 mass consists of secondary components. This part of the PM10 mass is less affected by a decreasing grid size in contrast to the locally emitted primary components.
The grid effect differs between urban regions. The strength of the urban signal is a function of local emission conditions (extension of the emission areas, emission density, emission gradient, etc.) and meteorological conditions determining ventilation efficiency. For similar emission conditions, regions with weak wind conditions will show a stronger urban signal than well ventilated regions.
For all models, increasing model resolution improves the model performance at stations near large conglomerations as reflected by lower mean biases for all components and increased spatial correlation coefficients for primary pollutants.
As about 70% of the model response to grid resolution is determined by the difference in emission strengths, improved knowledge on emission spatial variations at high resolution is a key for the improvement of modelled urban increments. For this purpose one relies on the replacement of currently used top-‐down European wide data with national expertise.
It is difficult to define a grid size that is adequate to resolve the urban signal under all conditions occurring in a European-‐wide modelling area. It has been shown before that, ideally, a grid size in the range of a few km down to 1 km should be chosen. Such a small grid size is not feasible for regional model applications because the data demands and operating requirements are far too large. If the main emphasis of a model application is targeted on the determination of background air quality for rural areas and large agglomerations, the grid scale M2 (28 km) or, if the data and operational requirements can be fulfilled, the grid scale M3 (14 km) seems to be a good compromise between a pure background application and an application which reproduces most of the urban signals (M4-‐ 7km or even higher resolution).
It should be noted that the response is more significant for the CHIMERE simulations compared to the other participating models. This seems to be due to a specific treatment of the mixing parameterization over urban areas that yields a bias of four to
3
five times smaller for the highest resolution in urban areas compared to other participating models. This point is still under investigation by INERIS with sensitivity analyses.
The limited impact of horizontal resolution on regional scale model performance shows that a continuing effort is needed to better understand atmospheric processes and interactions with the surface, and to improve our knowledge on emissions (amount, speciation, location and timing).
4
1. Introduction ............................................................................................................................ 5
2. Methodology .......................................................................................................................... 6
2.1 Simulations ........................................................................................................................... 6
2.2 Emissions .............................................................................................................................. 7
2.3 Participating chemistry-‐transport models (CTMs) ................................................................ 9
2.4 Definition of PM10 .............................................................................................................. 13
2.5 Model evaluation procedure .............................................................................................. 14
2.5.1 EMEP data ................................................................................................................... 15
2.5.2 AIRBASE data analysis .................................................................................................. 15
3. Results ................................................................................................................................... 19
3.1 Modelled concentration patterns ...................................................................................... 19
3.2 Scale dependency of model performances ....................................................................... 23
3.2.1 PM10 .......................................................................................................................... 23
3.2.2 NO2 ............................................................................................................................ 26
3.2.3 Ozone ......................................................................................................................... 30
3.3 Scale dependency in terms of seasonal variations .......................................................... 32
3.4 Scale dependency in terms of PM10 components .......................................................... 34
4. Discussion and conclusions ................................................................................................... 37
5. Acknowledgement ................................................................................................................ 39
6. References ............................................................................................................................ 40
7. Appendices ........................................................................................................................... 41
7.1 EMEP model description .................................................................................................... 41
7.2 CHIMERE model description .............................................................................................. 43
7.3 LOTOS-‐EUROS model description ...................................................................................... 46
7.4 REM-‐CALGRID (RCG) model description ............................................................................ 49
7.5 CMAQv5.0 model description ............................................................................................ 56
7.6 The ScaleDep DataBase ...................................................................................................... 59
5
1.
The EMEP MSC-‐W models (http://www.emep.int) have been instrumental to the development of air quality policies in Europe since the late 1970s. In the 1990s the EMEP models became also the reference tools for atmospheric dispersion calculations as input to the Integrated Assessment Modelling, which supports the development of air quality policies in the European Union. Since 1999, the EMEP model has been run on a resolution of 50x50 km2. However, the last years, modification of the EMEP grid has been discussed, an important aspect of which is the grid resolution. An increase in model resolution requires that the input data (most importantly the emissions) are available on the same scale. As an increase in model resolution will increase the computational costs cubically, it is important to determine the at which scale the improvement in resolution does not give improvement in performance any longer and for which the computational effort is not too large. To support EMEP in this decision, an initiative was taken for a model inter-‐comparison exercise aimed at analysing the model performance of different chemical-‐transport models as a function of model horizontal spatial resolution. Five modelling teams participated in the exercise: EMEP MSC-‐W, CHIMERE (INERIS), LOTOS-‐EUROS (TNO), RCGC (Berlin Freie Universität), and CMAQ (BSC). All models were to perform four runs for Europe at different resolution. The specifics of the models and the experimental set-‐up are described in section 2. A (statistical) evaluation of the performance of the models at different resolutions is presented in section 3. Section 4 presents the most important findings from this exercise. Descriptions of the various models that participated in this Scale-‐Dependency exercise are reported in the Appendices (Section 7).
6
2.
2.1 Simulations Each modelling team carried out four model simulations with different horizontal resolutions. The simulations were performed for the year 2009 for the EC4MACS domain encompassing Europe (Figure 1). For details regarding the EC4MACS project we refer to http://www.ec4macs.eu/index.html. Four resolutions were used doubling the resolution between each simulation. The horizontal spatial resolution ranges from a 1.0x0.5 degrees to 0.125 x 0.0625 degrees (see Table 1), which corresponds to 56x56 km and 7x7 km in the Northern part of the domain and to 88x56 km and 11x7 km in the Southern part of the domain, respectively. As the high resolution simulation is very demanding in terms of computing power the EC4MACS domain encompasses southern and central Europe completely, but cuts off the remote area in northern Scandinavia.
Figure 1 : Computational domain (grey area) broken down into 4 resolutions
Table 1 : Definition of the four resolution domains. nx and ny indicate the number of cells in longitude and latitude direction. N and S refer to cells in the Northern and Southern part of the domain, respectively.
Domain nx ny Lon
(degrees) Lat
(degrees) x Lat
(km x km) SW corner grid centre (Lon / Lat)
EC4M1 41 52 1.0 0.5 56 x 56 (N) 88 x 56 (S)
-‐10.000 / 36.125
EC4M2 82 104 0.5 0.25 28 x 28 (N) 44 x 28
-‐10.250 / 36.000
EC4M3 164 208 0.25 0.125 14 x 14 (N) 22 x 14 (S)
-‐10.3750 / 35.9375
EC4M4 328 416 0.125 0.0625 7 x 7 (N) 11 x 7 (S)
-‐10.43750 / 35.90625
7
All other input parameters were not prescribed. This means that the models use different meteorological input data as well as land use data. In case of meteorology most models use data from ECMWF, whereas RCGC used diagnostic meteorology from TRAMPER and CMAQ used meteorological fields from the WRF-‐ARW model. The ECMWF meteorology has a resolution of ca. 16km, so that models running at e.g. 7km are essentially running with fine-‐scale emissions but somewhat smoothed meteorology. Only CMAQ has a meteorological driver that runs at true 7km. This might be one reason for different behaviour of CMAQ to the other models when going down in scale. Boundary conditions were derived from global model climatological data or simulations as well as experimental climatological data. For a more detailed specification of the model and its input data we refer to Table 3 at the end of section 2.3. The output required for the exercise was prescribed and contains hourly as well as daily concentration distributions across Europe (see Table 2). The output species contain the oxidants as well as particulate matter (PM), its components and precursor species. PM and its components include: PM2.5 (< 2.5 µm) and PM10 (< 10 µm), primary particulate matter (PPM), nitrate (NO3), sulphate (SO4), ammonium (NH4), secondary organic aerosol (SOA), Dust and Sea Salt (SS) in fine and coarse fractions (subscript _f and _c, respectively). Besides concentrations also deposition fields (D-‐dry and W-‐wet) were reported to be able to perform a first order assessment of the representation of atmospheric input to eco-‐systems. Although depositions were reported, they were not used in the analysis of this study.
Table 2. Model output variables.
Time resolution
PM and components Gases Deposition Fluxes
Meteorology
Hourly PM2.5, PM10 O3, NO2, NO -‐ U10, T2m, Kz, PBL, u*
Daily PM2.5, PM10 PPM_fine, PPM_coarse NO3_f, NO3_c SO4_f, SO4_c NH4_f, NH4_c SOA_f, SOA_c Dust_f, Dust_c SS_f, SS_c
SO2, NH3, HNO3 D_NOx, D_SOx, D_NHx, W_NOx, W_SOx, W_NHx
Rain amount
2.2 Emissions The anthropogenic emission input was harmonized by using a common EC4MACS emission dataset. The emission dataset was delivered by INERIS for all model resolutions separately. Except for SNAP2, prescribed time profiles and height distributions were used following the EURODELTA protocol (cf. Thunis & Cuvelier et al. (2008)). For SNAP2 daily gridded modulation factors were calculated based on the daily temperatures, while the SNAP2 hourly variations were based on the EURODELTA hour-‐
8
of-‐the-‐day profile. The gridded distribution of anthropogenic emissions used for this exercise were provided by INERIS and they are based on a merging of databases from:
TNO 0.125°×0.0625° emissions for 2007 from MACC (cf. Kuenen et al. 2011)
EMEP 0.5°×0.5° for 2009 (cf. Vestreng et al. 2007)
Emission data from the GAINS database (see http://gains.iiasa.ac.at/gains)
INERIS expertise on re-‐gridding with various proxies (population, landuse, LPS data)
The large point sources (LPS) from the fine scale (0.125°×0.0625°) TNO-‐MACC emissions data for 2007 were added to surface emissions to get only one type of emissions. For the various activity sectors the processing steps were the following:
SNAP2: The country emissions were re-‐gridded with coefficients based on population density and French bottom-‐up data, the methodology was extrapolated to the whole of Europe. For PM2.5, the annual EMEP totals were kept except for the countries CZ, BA,BE, BY, ES, FR, HR, IE, LT, LU, MD, MK, NL, CS, TR. For the rest, PM2.5 emissions from GAINS were used. Additional factors were applied on Polish regions (×4 or ×8) for PM2.5 and PM10 emissions.
SNAP3,7,8,9,10: TNO-‐MACC emissions were used as proxy to regrid EMEP 0.5°x0.5° annual totals
SNAP1,4,5,6
For countries where TNO-‐MACC emissions are not available, EMEP 0.5°×0.5° emissions are used (Iceland, small countries, Asian countries). For SNAP2, we also propose a new temporal profile of the emissions according to the
variable to express is defined as: Dj = max(0, 20-‐TD) where TD is the daily mean 2m ambient temperature. Therefore, a first guess daily modulation factor could be defined as: with the annual 2 emissions are not only
related to the air temperature (e.g. emissions due to production of hot tap water), a second term is introduced in the formula by means of a constant offset C. To better assess the influence of this offset, C can be expressed as a fraction of (degree day average). Considering: C = A . where A = 0.1 (defined by user) (2.2.1) the daily modulation factor (Fj) is therefore defined as: where and (2.2.2)
Note that F is mass conservative over the year and replaces the original monthly and daily modulation factors.
9
As an illustration, Figure 2 shows the 2009 daily modulation factor apply to the SNAP2 emissions at 3 different locations: Katowice (Poland), Paris (France) and Madrid (Spain). We observe that the highest factors for the three locations are logically seen during the winter period and the lowest ones during the summer. This means, for example, that during the cold periods the emission from SNAP2 can be up to three times more intense (e.g. beginning of January for Madrid or end of February for Paris) than during the spring or the autumn periods. Interestingly, we note that in Katowice during the beginning of the year the factors are relatively lower than at the two other locations meaning that over this period the different between the daily mean temperature and the annual mean is lower in Katowice than the one at the two other locations. In other words, Madrid and Paris experienced a cold period in January and end of February, respectively, whose intensity was higher than over Katowice. Inversely, at the end of the year, all the locations experienced a cold outbreak of the same intensity relatively to their local annual mean temperature.
Figure 2: Daily modulation factors (Fj) applied for the SNAP2 emission over the city of Katowice, Paris and Madrid for the year 2009.
2.3 Participating chemistry-‐transport models (CTMs) In this study five Eulerian CTMs are applied to address the sensitivity of the model performance for particulate matter, nitrogen dioxide, and ozone,. The participating modelling systems are:
-‐ EMEP MSC-‐W -‐ CHIMERE (INERIS) -‐ LOTOS-‐EUROS (TNO) -‐ RCGC (Berlin Freie Universität) -‐ CMAQ (BSC)
In Table 3 the main characteristics of these model systems are listed. For detailed model descriptions we refer to the literature on these models. In the Appendices (Section 7), we introduce the models shortly.
10
Table 3. Overview of model characteristics. For bibliographical references see the Appendices.
MODEL EMEP MSC-‐W (§7.1)
CHIMERE (§7.2)
LOTOS-‐EUROS (§7.3)
RCGC (§7.4)
CMAQ §7.5)
version rv4beta12 v1.8 v2.1 v5.0
operator met.no INERIS/IPSL-‐CNRS TNO/KNMI/RIVM FU Berlin BSC-‐CNS
contact David Simpson Bertrand Bessagnet Martijn Schaap Rainer Stern José María Baldasano
email [email protected] [email protected]
[email protected] rstern@zedat-‐fu-‐berlin.de [email protected]
VERTICAL MODEL STRUCTURE Vertical layers
20 sigma 9 sigma 4 (3 dynamic layers and a surface layer)
5 fixed terrain following layers
15 sigma
Vertical extent
100 hPa 500 hPa 3500 m 3000 m 50 hPa
Depth first layer
90 m 20 m 25 m 25 m 19 m
NATURAL EMISSIONS BVOC Based upon maps of 115
species from Koeble and Seufert, and hourly temperature and light. See Simpson et al., 2012
MEGAN model Based upon maps of 115 species from Koeble and Seufert, and hourly temperature and light. See Denier van der Gon et al., 2009
Based upon maps of 115 species from Koeble and Seufert, and hourly temperature and light.using emissions factors of Simpson et al. (1999).
MEGAN model v2.0
Forest fires FINNv1, daily Monthyl GFED3 database MACC forest fires none none
Soil-‐NO Simpson et al. (2012) MEGAN model Not used here Simpson et al. (1999) MEGAN model v2.0
Lightning Climatological fields, Köhler et al. (1995)
none none none none
Sea salt Monahan (2006) and Martensson (2006), see Tsyro et al. (2011).
Monahan et al., 2006 Martensson et al., 2006 and Monahan et al., 1986
Gong et al. (1997) and Monahan et al. (1986)
Zhang et al. (2005) and Clarke et al.(2006)
Windblown Dust
Simpson et al., 2012 Vautard et al. (2005), not used here
Denier van der Gon et al. (2009). Not used here
Loosemore & Hunt (2000), Claiborn et al. (1998)
none
Agricultural land management
None none Denier van der Gon et al. (2009), Not used here
none none
Dust traffic suspension
Denier van der Gon et al. (2009).
none Denier van der Gon et al. (2009), Not used here
none none
Saharan dust inflow
Not used here Yes Not used here Not used here BSC-‐DREAM8bv2.0
LAND USE Landuse database
CCE/SEI for Europe, elsewhere GLC2000
GLOBCOVER (24 classes) Corine Land Cover 2000 (13 classes)
Corine Land Cover 2000 (13 classes)
Corine Land Cover 2006 (44 classes)
Resolution Flexible, CCE/SEI ~ 5 km 300 m 1/60 x 1/60 degrees 1/60 x 1/60 degrees 100m x 100m
BOUNDARY CONDITIONS Ozone and oxidants
As available (from global model or specified)
LMDzINCA monthly clim. MOZART 3-‐hourly O3 monthly clim (Logan, 1998); all other species clim. background values
MOZART-‐4/NCEP monthly mean
PM comp. clim. background values GOCART monthly clim. MOZART 3-‐hourly clim. background values clim. background values
11
Table 3. Continued
MODEL EMEP MSC-‐W (§7.1)
CHIMERE (§7.2)
LOTOS-‐EUROS (§7.3)
RCGC (§7.4)
CMAQ (§7.5)
METEOROLOGY Description ECMWF ECMWF IFS + urban mixing ECMWF TRAMPER diagnostic WRF-‐ARW v3.3.1
Resolution 0.22 deg x 0.22 deg 1/2 x 1/4 degrees 1/2 x 1/4 degrees 1/2 x 1/4 degrees The same as CMAQ
PROCESSES Advection Bott (1989a,b) Van Leer (1984) Walcek (2000) Walcek (2000) modified by
Yamartino (2003). Piecewise parabolic method (PPM) (Colella and Woodward, 1984)
Vertical diffusion
Kz approach Kz approach following Troen and Mart, 1986
Kz approach Kz-‐approach ACM2 PBL scheme (Pleim, 2007)
Dry deposition
resistance approach, Simpson et al (2012)
resistance approach Emberson (2000a,b)
DEPAC3.11 / Van Zanten et al.(2010)
resistance approach, DEPAC-‐module
Resistant approach, Venkatram and Pleim (1999)
Landuse class
16 classes 9 classes 9 classes 9 classes USGS 24-‐category system
Compensation points
No, but zero NH3 deposition over growing crops
no only for NH3 (for stomatal, external leaf surface and soil (= 0))
no no
Stomatal resistance
DO3SE-‐EMEP: Emberson et al, 2000, Tuovinen et al., 2004. Simpson et al., 2012
Emberson (2000a,b) Emberson (2000a,b) Wesely (1989) Wesely (1989)
Wet deposotion gases
In-‐cloud and sub-‐cloud scavenging coefficients
In-‐cloud and sub-‐cloud scavenging coefficients
pH dependent scavenging coefficients (Banzhaf et al., 2011)
pH dependent scavenging coefficients
In-‐cloud and sub-‐cloud scavenging which depends
dissociation constants and cloud water pH. Chang et al. (1987)
Wet deposition particles
In-‐cloud and sub-‐cloud scavenging
In-‐cloud and sub-‐cloud scavenging
In-‐cloud and sub-‐cloud scavenging
In-‐cloud and sub-‐cloud scavenging
In-‐cloud and sub-‐cloud scavenging which depends
dissociation constants and cloud water pH. Chang et al. (1987)
Gas phase chemistry
EmChem09 MELCHIOR TNO-‐CBM-‐IV CBM-‐IV CB-‐05 with chlorine chemistry extensions (Yarwood et al., 2005)
Cloud chemistry
Aqueous SO2 chemistry Aqueous SO2 chemistry Banzhaf et al. (2011) Aqueous SO2 chemistry Aqueous SO2 chemistry (Walcek and Taylor, 1986)
Coarse nitrate
yes no yes no Yes
Ammonium nitrate equilibrium
MARS ISORROPIA (Nenes et al., 1999)
ISORROPIA2 ISORROPIA ISORROPIAv1.7
SOA formation
VBS-‐NPAS Simpson et al. (2012)
After Bessagnet et al., 2009
not used SORGAM module (Schell et al., 2001)
SORGAM module (Schell et al., 2001)
VBS Yes, Bergström et al (2012), Simpson et al. (2012)
no not used none none
Aerosol model
Bulk-‐ approach (2 modes) 8 bins (40 nm -‐ 10 µm) Bulk-‐ approach (2 modes) Bulk-‐ approach (2 modes) AERO5 (Carlton et al., 2010) Log-‐normal approach (3 modes)
Aerosol physics
not used coagulation/condensation/nucleation
not used none Coagulation/condensation/nucleation
12
All models have kept there model codes the same for each resolution. The expected change in concentrations due to an increase in resolution is therefore due to the much sharper gradients in the emissions and the sensitivities of process descriptions to concentration differences. Also, the land use data are available as a mosaic at high resolution, meaning that the 7 km cells inside a 56 km cell will have the same land cover total per land use class but that they are also allocated with more detail at the higher resolution. All models interpolate the input meteorological data to the required model resolution. In the case of CMAQ, the WRF-‐AWR model is run at the required resolution for the 4 cases. One exception should be mentioned. Within CHIMERE an adjustment is made to mixing parameters above urban areas. This means that this adjustment is different in the simulations performed here. The full motivation by the CHIMERE team is given in the appendix (Section 7.2). Instead of applying a correcting profile to downscale the model outputs near the ground, the CHIMERE pre-‐processing was modified to diagnose an improved urban meteorology. In short, they argue that the mixing in the urban environment within the canopy layer is overestimated with standard similarity theory. Therefore, the Kz in the first CHIMERE layer above urban areas is modified as follows:
2
1(1 ) zKz ) Kz (z< z (2.1)
where z1 = 20 m, computed with similarity theory. The factor of 2 (derived from the literature) is also applied to lower the wind speeds in the first CHIMERE layer so as to limit the advection and dilution of primary pollutants close to the ground. This refined representation of turbulent mixing as a function of landuse (the fraction of urban grid cells) is analogous to the landuse dependent roughness, or dry deposition. But on the contrary to the later processes that are included in all participating models, only CHIMERE proposed this dedicated urban mixing representation. Impact of the urban correction in CHIMERE
Figure 3 presents the difference in concentrations between the simulation using the urban correction and the simulation which is not using any correction over urban areas, for four main pollutants: NO2, O3, and PM10. For all examined pollutants, we note a rather strong impact over all major European cities. For NO2, we observe an increase in concentrations ranging from a few ppb (suburban areas) to 12-‐15 ppb (e.g. Paris, London, Madrid and Milan). Consequently, to the NO2 concentrations increase over cities, we note a decrease of O3 (NO2 titration essentially), not only over the main cities (5 ppb on average up to 7 ppb in the city centres), but also over medium size cities (1 to 2 ppb). However the strongest impact is observed for PM10 for which an increase from a few µg/m3 to 40 µg/m3 for cities such as Milan or Paris is seen, but up to 70 µg/m3 for the region of Katowice in southern Poland.
13
Figure 3 Difference in concentration for three main pollutants between the simulation using the urban correction and the simulation not using the urban correction for PM10 in µg/m
3 (top left), NO2 in ppb (top right), and O3 in ppb (bottom left). 2.4 Definition of PM10 The participating models differ in the availability of PM components and formation routes. For instance, EMEP, LOTOS-‐EUROS and CMAQ contain coarse mode nitrate formation, whereas the others do not. Also, CHIMERE, CMAQ, EMEP and RCGC report secondary organic aerosol, where the LOTOS-‐EUROS modelling team considers their SOA model formulations too uncertain for use in policy support. The PM10 concentration is calculated as follows in each model:
EMEP PM10 = PPMcoarse+ PPMfine + SO42-‐+NO3
-‐+NH4++ Sea Salt + SOA + Dust
CHIMERE PM10 = PPMcoarse+ PPMfine + SO42-‐+NO3
-‐+NH4++ Sea Salt + SOA + Dust
CMAQ PM10 = PPMcoarse+ PPMfine + SO42-‐+NO3
-‐+NH4++ Sea Salt + SOA + Dust
LOTOS-‐EUROS PM10 = PPMcoarse+ PPMfine + SO42-‐+NO3
-‐+NH4++ Sea Salt
RCGC PM10 = PPMcoarse+ PPMfine + SO42-‐+NO3
-‐+NH4++ Sea Salt + SOA + Dust,
where PPM stands for Primary Particulate Matter. Hence, the modelled components and therefore the explained mass will differ between models hampering the comparability. Also, the components that are not common (dust, SOA, coarse nitrate) are associated with the largest uncertainties. Hence, we also
14
defined a common PM10 measure that includes the PM10-‐components that all models have:
PM10_common = PPM coarse + PPMfine + SO42-‐ + NO3
-‐ + NH4+ + SeaSalt
This quantity allows for a fair comparison between the different models. Note, that due to the neglect of several components an underestimation is expected as observed in many model evaluation studies and previous model intercomparisons (e.g. Stern et al., 2008; Vautard et al., 2009; Solazzo et al., 2012; Pay et al., 2012). However, in this analysis the PM10_common concentrations were not used, so that the PM10 concentrations in the models reflect different compositions. We refer to Table 4 for the PM components that are included in the PM10 and PM2.5 concentration of each model. Table 4. Overview of PM components that are included in the PMfine and PMcoarse fractions of each model. Filled grey = Included in model simulations and reported; x = included in model but not reported, blank = not in the model.
Fine mode Coarse mode
SO4
NO
3
NH
4
PPM
ASO
A
BSO
A
SS
Dus
t
SO4
NO
3
NH
4
PPM
ASO
A
BSO
A
SS
Dus
t
EMEP x x x x CHIMERE LOTOS-‐E RCGC x CMAQ For further information on reported and available results, we refer to Section 7.6 of this report. 2.5 Model evaluation procedure The impact of the increase of the model resolution on its performance needs to be quantified. As an important feature of the increased resolution should be to better describe horizontal gradients and better resolve the gradients between source regions and the regional background, the quantification of the spatial correlations for all resolution is the first analysis performed here. The higher resolution of the emissions and the model may separate rural background monitoring sites from those of urban locations and source areas by the fact that they appear in different model grids. To analyse the improvement in the spatial gradients the spatial correlation, the slope of the best fit and the bias are used. Not only the gradients should be described better, also a small improvement of the temporal representation of the measured times series is expected for primary components. The reason is that plumes of a source region may or may not hit the station in the high resolution, whereas in the coarse resolution the site could be in the same grid cell and it will always be affected. Hence, the second part of the evaluation focuses on the temporal behaviour looking at correlation coefficients and RMSE. The model to measurement comparison was performed at a central location (JRC) to ensure a harmonised evaluation procedure. For this purpose the DeltaTool (Thunis et al., 2011) developed in the frame of the FAIRMODE activity has been used.
15
In this exercise monitoring data from two networks were used (see Table 5).
Table 5. Monitoring data from two different databases
EMEP AIRBASE O3 O3
O3_8HrMax O3_8HrMax NO2 NO2 PM10 PM10 SO4 NO3
The first network is the EMEP network (http://www.emep.int, Törseth et al (2012)). This network was designed to evaluate regional scale models aimed at correctly describing trans-‐boundary air pollution. As these models used to have resolutions of 50 150 km when the network was designed, the locations of most stations are at a considerable distance from source areas in rural or remote regions. The only exception is ammonia, as the major ammonia sources are associated with agricultural rural areas. Hence, the EMEP network may not be the most suitable network to investigate the impact of high resolution modelling. Therefore, we also use the AIRBASE database (http://acm.eionet.europa.eu/databases/airbase/) which contains a host of stations located in rural and urban areas. To focus on the representation of gradients around large agglomerations a specific selection of stations was performed as presented below. Note that the AIRBASE network contains only the major pollutants. Hence, particulate matter speciation is addressed using the EMEP network. 2.5.1 EMEP data The EMEP network is not used to its full extent here. The analysis aimed to the urban agglomerations was prioritised as the first evaluation showed little impact of model resolution at rural monitoring stations. See below. 2.5.2 AIRBASE data analysis To highlight the differences between the model results obtained with the 4 spatial resolutions, we focus the analysis on 30 European urban areas (see Figure 4).
16
Figure 4. Urban areas selected for the analysis. Available measurements within a radius of 30 km around each city area are selected for urban background stations. For comparison with AIRBASE rural background and EMEP stations, the selected radius around each city is 200 km.
Within a radius of 30 km around each city (regardless of the city size) all AIRBASE urban background measurements are used to evaluate the model results and assess the impact of increased resolution. This 30 km radius is chosen because it leads to a surface around each station approximately equal to the area of a 50x50 km EMEP grid cell. The number of stations available within each city area differs from city to city as does the split in terms of station types (Rural, Urban, EMEP). An overview of the stations used per city for the evaluation is provided in Table 6 and in Section 7.6. Within a radius of 30 km around each city not many rural stations are included in the analysis. To provide some classical evaluation in terms of comparison with rural background stations a larger radius of 200 km has been considered, in which all AIRBASE-‐rural and EMEP measurement stations have been used and compared to model results.
17
Table 6. Overview of the available measurement stations per city, pollutants and station type. For the AIRBASE urban groups a radius of 30 km around each city is assumed whereas for the AIRBASE rural background and EMEP groups, the radius is 200 km.
It is important to remember that the number of stations included in each of the City groups is different. While the total number of rural stations (R) used to produce the average for PM10 is 151 it reaches 98 for urban stations (see Table ). The weight of the various cities into this average is also a key element which needs to be remembered for the interpretation. While for Warsaw 10 urban background stations are considered, other cities like Stockholm only have one.
The analysis is performed for
Daily mean PM10 concentrations;
O3Urban Rural EMEP Urban Rural EMEP Urban Rural EMEP
Amsterdam AMS 2 11 7 12 1 12Athens ATH 2 1 2 1Barcelona BAR 5 4 5 6 4 6Berlin BER 5 10 1 6 9 3 9Bilbao BIL 4 5 1 4 5 1 4 5 1Bruxelles BRU 11 1 16 1 16Bucarest BUC 1 0 1 1 1Budapest BUD 1 2 2 1 2 1 2 2Cologne COL 7 9 9Dublin DUB 2 2 2 2 1 2Hambourg HAM 6 7 1 6 10 2 10Krakow KRA 5 3 3 3 2 3Leeds LEE 1 2 3 2 2 3 3Lisbon LIS 8 4 11 4 10 4London LON 3 3 1 5 5 3 5 5 3Lyon LYO 2 2 3 9 3 9Madrid MAD 5 9 3 7 9 3 7 9 3Marseille MAR 4 1 4 10 1 10Milan MIL 8 7 9 13 9 13Munich MUN 1 5 1 10 1 10Naples NAP 1 2 2Paris PAR 7 6 19 11 12 11Prague PRA 4 20 1 4 14 2 14 1Rome ROM 6 2 1 6 2 6 2Sevilla SEV 2 2 5 3 4 3Sofia SOF 1 1 1Stockholm STO 1 2 1 1 2 1 2 1Valencia VAL 2 3 1 2 5 1 2 5 1Vienna VIE 3 21 1 4 19 3 19Warsaw WAR 10 6 3 3 3 1Total 98 151 14 126 201 10 93 201 16
PM10 NO2
18
Hourly NO2; Daily mean O3, and daily maximum of the running 8 hourly mean O3;
and mostly focuses on urban background station types since the increased resolution is expected to have its maximum gain at those stations. In order to better highlight model differences in terms of urban areas and station types, groups of stations are generated in which statistical performance indicators are averaged. Finally it is important to note that differences between resolutions should be seen as grid resolution increments rather than urban increments. Indeed, in the case of a very large city area characterized by a homogeneous distribution of the emissions, no difference between resolutions should be expected. A list of the various City groups for the three pollutants PM10, NO2, and O3 is given in Section 7.6.
19
3.
3.1 Modelled concentration patterns Figures 5, 6, 7 show the annual mean concentrations of PM10, NO2, and O3 as calculated for the four different horizontal grid resolutions. Figure 5 shows that for PM10 going from the coarse grid EC4M1 (56km) to the fine grid EC4M4 (7km), the concentration pattern increasingly reflects the underlying emission pattern. The calculated concentrations increase in particular in the high emission density areas. However the increase is moderated because a large part of the total PM10 is provided by secondary aerosols which are less affected by the grid size than de primary PM components. Also for NO2 the concentration pattern reflects increasingly the underlying emission pattern going to higher resolution. Overall, the increase in structure is tremendous. Especially going from 56 to 14 km adds a lot of detail, whereas the step from 14 km to 7 km does not show such a large change in the structure. The calculated concentrations increase in particular in the high emission density areas. Compared to NO2 and PM10, the effect of a decreasing grid size is small for ozone. In general, there are only small changes in rural areas. In urban areas, a decrease of the grid size leads also to a decrease of the calculated O3 concentrations as titration by local NO sources is enhanced.
20
Figure 5. Modelled annual-‐mean PM10 fields (µg/m3) at M1 (56km) and M4 (7km) resolution.
21
Figure 6. Modelled annual-‐mean NO2 fields (µg/m3) at M1 (56km) and M4 (7km) resolution.
22
Figure 7. Modelled annual-‐mean O3 fields (µg/m3) at M1 (56km) and M4 (7km) resolution.
23
3.2 Scale dependency of model performances 3.2.1 PM10 In Figure 8 the annual average PM10 concentrations are grouped by station types (i.e. all stations belonging to a station type are grouped together). For PM10, the grid resolution increment is very weak at AIRBASE rural and EMEP stations in all models. Models exhibit a similar behaviour in terms of impact of resolution but the magnitude is different. Urban increments are similar for the CHIMERE, LOTOS-‐EUROS, and RCGC, whereas for EMEP and CMAQ they are significantly lower. For EMEP we explain this by the difference in the surface layer depth, being deeper in EMEP. Note that all models show more or less the same steps for each increase in resolutions.
Figure 8. Annual averaged PM10 concentrations per station type, from left to right: EMEP (E), Rural (R), Urban(U) for the 4 spatial resolution (light blue EC4M4-‐7km, orange EC4M3-‐14km, dark blue EC4M3-‐28km, red EC4M1-‐56km). U groups are based on the 30 km radius; E and R groups are based on a radius of 200 km.
24
A significant part (~70%) of the concentration increment (7-‐56 km) can be explained by the emission density increment as demonstrated by the high values of R2 (Figure 9), with the exception of CMAQ. This is mostly seen for EMEP, CHIMERE, and RCGC whereas some additional variance in the concentration increments is seen for the other models. Most of this additional variance happens in stations belonging to countries like Italy, Greece, Portugal and Spain. CHIMERE and RCGC clearly show the most intense response followed by LOTOS-‐EUROS whereas EMEP and CMAQ exhibits significantly lower increments. The absolute grid effect for PM10 is lower than for NO2 (see Section 3.2.2), as explained by the higher importance of the rural background levels containing secondary material for PM10. On the other hand, the slopes for PM10 expressing the concentration increase per unit emission are steeper than for NO2. Possible explanation for the steeper slopes and slightly different behaviour for PM10 in comparison to NO2 may lie in the fact that NO2 increments may be limited due to the availability of ozone as NO2 is formed through titration of ozone.
Figure 9. Relation between PM10 concentrations (µg/m3) and PPM emission deltas for EC4M1-‐56km and EC4M4-‐7km resolutions. Stations are indicated per country.
25
Figure 10 shows, for the EMEP model (similar LOTOS-‐EUROS, RCGC, CMAQ) and CHIMERE the annual mean PM10 concentrations at the urban stations per agglomeration. In this picture the same reduction in the bias is observed as given in Figure 8. Here it can be seen that the response per agglomeration is variable. For EMEP, the impact of an increase in resolution is small, except for the city of Krakow, where the major impact is seen between EC4M2-‐28km and EC4M3-‐14km. For CMAQ, the impact of resolution is more significant. In particular we notice the difference between EC4M3-‐14km and EC4M4-‐7km for cities like, Krakow, Munchen, Paris, and Rome.
Figure 10. EMEP (similar for LOTOS-‐EUROS, RCGC, CMAQ) and CHIMERE results for annual averaged PM10 concentrations per city area for urban background stations at the 4 spatial resolutions (light blue EC4M4-‐7km, orange EC4M3-‐14km, dark blue EC4M2-‐28km, red EC4M1-‐56km) A summary of the spatial statistical analysis for daily PM10 is given in Figure 11. The main findings are:
-‐ Spatial correlation is hardly affected by resolution change (with the exception of CMAQ) but the slope is significantly improved with higher resolution as a result of a lower bias at urban stations.
-‐ At urban locations and comparing EC4M1-‐56km to EC4M4-‐7km, the bias for PM10 reduced by almost 6.5 µg/m3 for CHIMERE, by 5.5 µg/m3 for RCGC and LOTOS-‐EUROS, whereas the EMEP bias reduces about 3.3 µg/m3 and only 1.2 µg/m3 for CMAQ.
-‐ Performance at rural sites is stable with resolution in all models, except for CMAQ
where an increase is seen for Slopes and Correlations.
26
Figure 11. Summary of statistical analysis for daily PM10. Note the different scales between the station groups
3.2.2 NO2 As seen in Figure 12 where the annual average NO2 concentrations are grouped by station types, the grid resolution increment is as expected very weak at rural and EMEP stations. It is interesting to note the small decrease of NO2 modelled at the EMEP stations. All models agree with a negligible impact at rural and EMEP stations and on the trends at urban stations. Although the magnitude differs between models, the gain resulting from an increased resolution is significant for urban stations (from 10 to 20 µg/m3 between EC4M1-‐56km and EC4M4-‐7km). The largest increments are seen for CHIMERE and RCGC and slightly lower responses for the other models. For EMEP, LOTO and CMAQ, the largest gains seem to be between M1(56km) and M2(28km) and from M2(28km) to M3(14km) with a kind of saturation to the higher resolutions. CHIMERE and RCGC react more regularly to the resolution change than the others.
27
Figure 12. Annual averaged NO2 concentrations per station type, from left to right: EMEP (E), Rural (R), Urban (U), for the 4 spatial resolution (light blue EC4M4-‐7km, orange EC4M3-‐14km, dark blue EC4M3-‐28km, red EC4M1-‐56km). U groups are based on a radius of 30 km, E and R on 200 km radius.
One of the major causes of these model resolution increments is the emissions, and in particular the differences between emission densities at fine and coarse resolutions. Figure 13 illustrates the relation between concentration deltas (Concentration [7km] Concentration [56km]) and emission deltas (Emission [7km] Emission [56km]). As seen from this figure most of the concentration increment (delta) can be explained by the emission delta. The spread around the trend line provides some information on the importance of other factors such as the local meteorological or the role of chemistry. Most of this additional spread happens in stations belonging to countries like Italy, Greece, Portugal and Spain. About 70% of the model response to grid resolution is determined by the difference in emission strength with the exception of CMAQ for which this number reduces to about 50%.
28
Figure 13. Relation between NO2 concentrations (µg/m3) deltas and NOx emission deltas between EC4M1-‐56km and EC4M4-‐7km resolutions. Stations are indicated per country.
Figure 14 shows, for the EMEP model (similar LOTOS-‐EUROS, CMAQ) and CHIMERE (similar RCGC), the annual mean NO2 concentrations at the urban stations per agglomeration. In this picture the same reduction in the bias is observed as given in Figure 12. Here it can be seen that the response per agglomeration is variable. For EMEP, the differences for Hamburg and Prague are small, whereas the increase of resolution from EC4M1-‐56km to EC4M4-‐7km for Barcelona, Brussels, London, Madrid, and Paris are similar (no change from EC4M3-‐14km to EC4M4-‐7 km). Athens is a special case with regular large differences (13 µg/m3) between the resolutions. For CHIMERE we notice small impact in the City areas of Bilbao, Marseilles and Prague, while the increase in resolution from EC4M1-‐56km to EC4M4-‐7km is significant in contrast to the EMEP group results.
29
Figure 14. EMEP (similar for LOTOS-‐EUROS, CMAQ) and CHIMERE (similar RCGC) results for annual averaged NO2 concentrations per city area for urban background stations at the 4 spatial resolutions (light blue EC4M4-‐7km, orange EC4M3-‐14km, dark blue EC4M2-‐28km, red EC4M1-‐56km)
The spatial variation is further looked at in Figure 15. These show that the explained spatial variability improves as function of resolution. Moreover, the slope of the fits increase significantly indicating that the models explain better the magnitude of the variability between urban regions.
30
Figure 15. Comparison of observed and modelled annual average concentrations of NO2 for the urban agglomerations. For the EC4M1-‐56km (upper panel) and EC4M4-‐7km (lower panel) the fits parameters are shown.
A summary of the spatial statistical analysis for hourly NO2 is given in Figure 16. The main findings are:
-‐ As expected performance at rural sites is stable with resolution, although a significant increase in the slope is seen for CMAQ, and to a lesser extent for CHIMERE.
-‐ For urban stations there is a strong dependence on the resolutions, with an increase in slopes, and increase in correlations, and a reduction of the bias for all the models. The largest gaina appear to occur between EC4M1-‐56km and EC4M3-‐14km, with a saturation between EC4M3-‐14km and EC4M4-‐7km.
Figure 16. Summary of statistical analysis for hourly NO2. Note the different scales between the station groups
3.2.3 Ozone For ozone we focus our analysis on the model performance for the daily maximum of the running 8 hour mean. The reason is that the analysis is then focussed on the high ozone regime during daytime, and is less sensitive to the impact of differences between models on night time mixing and titration. First, the annual average O3Max8Hr levels (Figure 17) show a different behaviour compared to PM and NO2. Due to the secondary nature of ozone combined with the titration impact of NOx emissions near sources the levels are lower inside a city than outside. The average pattern as function of station
31
type and the response to a resolution change between all models is very similar. Average levels for urban stations decrease towards the observed values for all models.
Figure 17. Annual averaged ozone concentrations per station type, from left to right: Urban (U), Rural (R), EMEP (E), for the 4 spatial resolution (light blue EC4M4-‐7km, orange EC4M3-‐14km, dark blue EC4M2-‐28km, red EC4M1-‐56km). U City groups are based on the 30 km radius ; E and R groups are based on a radius of 200 km
The model performance evaluation shows that the models do not respond to the resolution change in the rural areas (see Figures 17 and 18). Statistical parameters are more or less constant. At rural stations the impact of resolution is rather small, while an increase in resolution has a significant effect for the urban locations. At these stations the slopes decrease for EMEP, LOTOS-‐EUROS and RCGC, while CHIMERE and CMAQ show a minimum at EC4M2-‐28km. Spatial correlations decrease slightly with increasing resolution. Also the urban bias improve, except for LOTOS-‐EUROS. However, increasing resolution and adding more local variability decreases the representation of the spatial contrasts, although for NO2 the spatial patterns become better between cities. This may mean that it is not the variability in NOx emission source strengths between the
32
urban regions that is the most important uncertainty for ozone gradients during the day at these scales. Instead, differences in mixing regimes, chemical regimes and uncertainties in NMVOC speciation could be more important.
Figure 18. Summary of statistical analysis for O3Max8hr. Note the different scales between the station groups
3.3 Scale dependency in terms of seasonal variations
Comparison between the 5 model results at two horizontal resolutions (EC4M1-‐56km and EC4M4-‐7km) against observations was performed in order to evaluate the impact of an increased resolution on the representation of the variation between summer and winter of the PM10, NO2, and O3 concentrations. The dynamical evaluation approach was used in order to better apprehend the seasonal variability of these three pollutants. Three types of stations were studied: EMEP, rural and urban. Results are shown in Figure 19. The dynamic evaluation of the model performance shows that the models do not respond much to the resolution change in all three types of stations. For O3, slight improvement of the representation of the seasonal variability in urban and rural areas can be seen in Figure 19a and Figure 19b especially for CHIMERE and RCGC for which the underestimation of the variation summer-‐winter by the models is slightly reduced. For NO2, LOTOS better reproduces the summer-‐winter difference at 7 km in urban areas, but no remarkable differences for the other models at the three types of stations (Fig. 19c and 19d) are seen. For PM10, increasing the resolution from 56 km to 7 km improved the results of the 5 models in urban areas. No significant changes were seen for O3 (Fig. 19e and 19f).
33
Overall, the horizontal resolution does not seem to play a major role in the performance of the models in reproducing the seasonal variability between summer and winter for O3, NO2 and PM10 concentrations at different type of stations.
Figure 19: Differences between the O3 concentrations in summer and in winter observed values on the x-‐axis and modelled results on the y-‐axis. (a) shows the results for the 5 models at 56 km of horizontal resolution and (b) at 7km of horizontal resolution. (c) idem as (a) but for NO2. (d) idem as (b) but for NO2. (e) idem as (a) but for PM10 and (f) idem as (b) but for PM10. Stations are grouped as Urban (U), Rural (R), and EMEP (E).
34
3.4 Scale dependency in terms of PM10 components A comparison between the model results at EC4M4-‐7km and EC4M1-‐56km in terms of concentrations of the main PM10 components (i.e. PPM10, NO3
-‐, SO42-‐, NH4
+, Dust and Sea Salt) at three different station groups (urban, rural and EMEP) is shown in Figure 20. For PPM10, the
-‐ -‐
µg/m3
Figure 20. Scale dependency in terms of PM10 components for EC4M1-‐56km resolution compared to EC4M4-‐7km resolution indicated by
35
Figures 21 and 22 show the impact of resolution (EC4M1-‐56km vs EC4M4-‐7km) for the Sulphate and Nitrate components of PM10. Figure 21 shows the annual mean value of SO4-‐10 at three groups of EMEP stations (DE, ES, IT), where sufficient observational data were available. The German and the Italian group are composed of only one station, while the Spanish group contains 5 stations (see Table 7)
Group EMEP Station Codes Station Names DE DE0044R Melpitz ES ES0008R
ES0010R ES0014R ES0016R ES0036R
Niembro Cabo_de_Creus Els_Torms O_Saviñao Montseny
IT IT0001R Montelibretti
Table 7. Groups of EMEP stations used for the Sulphate and Nitrate analysis. As expected, the impact of resolution on SO4-‐10 is consistent with figure 20 (SO4, EMEP) where a larger set of EMEP stations was considered (see Section 7.6). A comparison with observations shows that the model results range, roughly speaking, from 50% of the observational values to 150% with the remark that the CHIMERE model is overestimating in all cases, while EMEP, LOTO and CMAQ are underestimating. RCGC performs well at the DE stations, but underestimates at ES and IT.
Figure 21. Impact of resolution (M1 vs M4) on the annual mean value of SO4-‐10 and comparison with observations at some EMEP stations in DE (Germany), ES (Spain), and IT (Italy).
Similar remarks hold for NO3-‐10 in Figure 22. Small impact of resolution on model results, and in general an underestimation of the observations, except for CMAQ in DE and ES (M1).
36
Figure 22. Impact of resolution (M1 vs M4) on the annual mean value of NO3-‐10 and comparison with observations at some EMEP stations in DE (Germany), ES (Spain), and IT (Italy).
It should be mentioned here that the MOD-‐OBS intercomparison was not the main objective of this study which was more focused on the scale dependency of model results. These two MOD-‐OBS intercomparisons are just two examples of a (future) larger study on model validation of PM components, for which high quality observational data is needed. We refer to ongoing work within the EuroDelta project.
37
4.
The ScaleDep exercise showed that the model responses to an increase in resolution show a broadly consistent picture among all models. The analysis showed that the grid size does not play a major role for air quality model calculations which are targeted on the determination of the background (non-‐urban) air quality. Downscaling model resolution does not change concentration estimations and model performance at rural and EMEP sites. The grid resolution plays an important role in agglomerations characterized by high emission densities. The urban signal, i.e. the concentration difference between high emission areas and their surroundings, usually increases with decreasing grid size. This grid effect is more pronounced for NO2 than for PM10, because a large part of the urban PM10 mass consists of secondary components. This part of the PM10 mass is less affected by a decreasing grid size in contrast to the locally emitted primary components. The grid effect differs between urban regions. The strength of the urban signal is a function of local emissions conditions (extension of the emission areas, emission density, emission gradient, etc.) and meteorological conditions determining ventilation efficiency. For similar emission conditions, regions with weak wind conditions will show a stronger urban signal than well ventilated regions. For all models, increasing model resolution improves the model performance at stations near large conglomerations as reflected by lower biases for PM10, NO2, and O3 and increased spatial correlation for NO2. As about 70% of the model response to grid resolution is determined by the difference in emission strength, improved knowledge on spatial variation in emission at high resolution is key for the improvement of modelled urban increments. For this purpose one relies on the replacement of currently used top-‐down European wide data with national expertise. It should be mentioned that there are many benefits to having emissions data collected at finer resolution than investigated here. One major reason is that there is a frequent need to re-‐project emissions data in different coordinate systems. In the near future for example there will be a need to run models in either the traditional EMEP polar-‐stereographic projection, or longitude-‐latitude. Indeed, even the long-‐lat projection of 0.1° resolution suggested for EMEP does not fit perfectly with the 0.125° system used here. Further, there is an increasing need to use meteorology from global and/or regional climate models, and these can use so-‐called rotated longitude-‐latitude projections. With any re-‐mapping, the finer the resolution of the base emissions system the more accurate any re-‐mapped emission inventory can be. It is difficult to define a grid size that is adequate to resolve the urban signal under all conditions occurring in a European-‐wide modelling area. Ideally, a grid size in the range
38
of a few km down to 1 km should be chosen. Such a small grid size is not feasible for regional model applications because the data demands and operating requirements are far too large. If the main emphasis of a model application is targeted on the determination of background air quality for rural areas and large agglomerations, the grid scale EC4M2-‐28km (0.5° Longitude and 0.25° Latitude) or, if the data and operational requirements can be fulfilled, the grid scale EC4M3-‐14km (0.25° Longitude and 0.125° Latitude) seems to be a good compromise between a pure background application and an application which reproduces most of the urban signals (EC4M4-‐7km resolution or even higher resolution). It should be noted that the response is more significant for CHIMERE simulations compared to the other participating models. It seems to be due to a specific treatment of the mixing parameterization over urban areas. This point is still investigated by INERIS with sensitivity analyses. The limited impact on regional scale model performance shows that a continuing effort is needed to better understand atmospheric processes and interactions with the surface, and to improve our knowledge on emissions (amount, speciation, location and timing).
39
5.
The modelling teams that provided input to this exercise were funded through their own channels. The EMEP MSC-‐W team received funding through EMEP under UN-‐ECE, as well as through the EU ECLAIRE project. INERIS was funded by the French Ministry in charge of Ecology. The LOTOS-‐EUROS modelling team was funded by the Dutch Ministry for Infrastructure and Environment. The RCGC modelling team was funded by Umweltbundesamt Germany, project number 21981. The CMAQ model -‐
-‐0050) from the Consolider-‐Ingenio 2010 program of the Spanish Ministry of Economy and Competitivity.
40
6.
Kuenen, J., H. Denier van der Gon, A. Visschedijk, H. van der Burgh, R. van Gijlswijk: MACC European emission inventory 2003-‐2007, TNO report, TNO-‐060-‐UT-‐2011-‐00588 Pay M.T., P. Jiménez-‐Guerrero, O. Jorba, S. Basart, M. Pandolfi, X. Querol, and J.M. Baldasano: Spatio-‐temporal variability of levels and speciation of particulate matter across Spain in the CALIOPE modelling system. Atmos Environ, 46, 376-‐396, 2012. Thunis P., C. Cuvelier, P. Roberts, L. White, L. Post, L. Tarrason, S. Tsyro, R. Stern, A. Keschbaumer, L.Rouil, B. Bessagnet, R. Bergstrom, M. Schaap, G. Boersen, P. Builtjes: EuroDelta-‐II, Evaluation of a Sectorial Approach to Integrated Assessmnet Modelling including the Mediterranean Sea. JRC Scientific and Technical Reports EUR 23444 EN 2008. Thunis P., E. Georgieva, A. Pederzoli, 2011: The DELTA tool and Benchmarking Report template:
http://fairmode.ew.eea.europa.eu/ Törseth, K.; Aas, W.; Breivik, K Fjæraa, A. M.; Fiebig, M.; Hjellbrekke, A. G.; Lund Myhre, C.; Solberg, S. & Yttri, K. E. Introduction to the European Monitoring and Evaluation Programme (EMEP) and observed atmospheric composition change during 1972-‐-‐2009 Atmos. Chem. Physics, 2012, 12, 5447-‐5481 Vautard, R., B. Bessagnet, M. Chin, and L. Menut: On the contribution of natural Aeolian sources to particulate matter concentrations in Europe: testing hypotheses with a modelling approach, Atmospheric Environment, 39, 3291 3303, 2005. Vestreng, V. , G. Myhre, H. Fagerli, S. Reis, and L. Tarrason: Twenty-‐five years of continuous sulphur dioxide emission reduction in Europe, Atmos. Chem. Phys., 7, 3663 3681, 2007
41
7.
7.1 EMEP model description The EMEP MSC-‐W (Meteorological Synthesizing Centre -‐West) model is a development of the 3-‐D chemical transport model of Berge and Jakobsen (1998), extended with photo-‐oxidant and inorganic aerosol chemistry (Andersson-‐Sköld and Simpson, 1999; Simpson et al., 2003, 2012), and organic aerosol modules (Bergström et al., 2012). In this exercise we use model version rv4beta12, which is identical to the rv4 version documented in Simpson et al. (2012) except for some minor updates.
The model includes 20 vertical layers, using terrain-‐following coordinates; the lowest layer has a thickness of about 90m. The meteorological fields for 2009 are derived from the European Centre for Medium Range Weather Forecasting Integrated Forecasting System (ECMWF-‐IFS) model (http://www.ecmwf.int/research/ifsdocs ) / The model uses essentially two modes for particles, fine and coarse aerosol, although assigned sizes for some coarse aerosol vary with compound. The parameterization of the wet deposition in the model is based on Berge and Jakobsen (1998) and includes in-‐cloud and sub-‐cloud scavenging of gases and particles. Further details, including scavenging ratios and collection efficiencies, are given in Simpson et al. (2012). Boundary concentrations of most long-‐lived model components are set using simple functions of latitude and month (see Simpson et al., 2012 for details). For ozone more accurate boundary concentrations are needed and these are based on climatological ozone-‐sonde data-‐sets, modified monthly against clean air surface observations at Mace Head on the west coast of Ireland (Simpson et al., 2012). REFERENCES
Andersson-‐Sköld, Y. and Simpson, D.: Comparison of the chemical schemes of the EMEP MSC-‐W and the 985 IVL photochemical trajectory models, Atmos. Environ., 33, 1111 1129, 1999. Berge, E. and Jakobsen, H. A.: A regional scale multi-‐layer model for the calculation of long-‐term transport and deposition of air pollution in Europe, Tellus, 50, 205 223, 1998. Bergström, R., Denier van der Gon, H.A.C.; Prévôt, A.S.H., Yttri, K.E., and Simpson, D.: Modelling of organic aerosols over Europe (2002-‐2007) using a volatility basis set (VBS) framework: application of different assumptions regrading the formation of secondary organic aerosols. Atmos. Chem. Phys., 12, 5425-‐5485, 2012. Bott, A. A positive definite advection scheme obtained by non-‐linear re-‐normalization of the advection fluxes Mon. Weather Rev., 1989, 117, 1006-‐1015Köhler, I.; Sausen, R. & Klenner, G. NOx production from lightning 1995, 343-‐345
42
Koeble, R. and Seufert, G., 2002, Novel maps for forest trees in Europe. Proceedings of the 8th European Symposium on the Physico-‐Chemical Behaviour of Atmospheric Pollutants. JRC Ispra. Köhler, I., Sausen, R. , Klenner, G. NOx production from lightning 1995, Report DLR Oberpfaffenhofen, Germany, p.343-‐345 Simpson, D., Fagerli, H., Jonson, J., Tsyro, S., Wind, P., and Tuovinen, J.-‐P.: The EMEP Unified Eulerian Model. Model Description, EMEP MSC-‐W Report 1/2003, The Norwegian Meteorological Institute, Oslo, Norway, 2003. Simpson, D., Benedictow, A., Berge, H., Bergström, R., Emberson, L. D., Fagerli, H., Flechard, C: Hayman, G. D., Gauss, M., Jonson, J. E., Jenkin, M. E., Nyíri, Á., Richter, C., Semeena, V. S., Tsyro, S., Tuovinen, J.-‐P., Valdebenito, A., and Wind, P.: The EMEP MSC-‐W chemical transport model technical description, Atmos. Chem. Phys. 12, 7825-‐7865, doi:10.5194/acp-‐12-‐7825-‐2012, http://www.atmos-‐chem-‐phys.net/12/7825/2012/, 2012. Tsyro, S.; Aas, W.; Soares, J.; Sofiev, M.; Berge, H. & Spindler, G. Modelling of sea salt concentrations over Europe: key uncertainties and comparison with observations Atmos. Chem. Physics, 2011, 11, 10367-‐10388 Tuovinen, J.-‐P.; Ashmore, M.; Emberson, L. & Simpson, D. Testing and improving the EMEP ozone deposition module Atmos. Environ., 2004, 38, 2373-‐2385
43
7.2 CHIMERE model description CHIMERE is designed to calculate the concentrations of usual chemical species that are involved in the physic-‐chemistry of the low troposphere. CHIMERE has been described in detail several times: Schmidt et al. (2001) for the dynamics and the gas phase module; Bessagnet et al. (2004, 2008, 2009) for the aerosol module. The aerosol model species are sulphates, nitrates, ammonium, secondary organic aerosol (SOA), sea-‐salt and dust. The particles size distribution ranges from 40 nm to 10 µm and are distributed into 8 bins (0.039, 0.078, 0.156, 0.312, 0.625, 1.25, 2.5, 5, 10 µm). For more detail on the latest development one can refer to the online documentation : (http://www.lmd.polytechnique.fr/chimere). A coarse domain encompassing the four nested domain has been defined with a 50 km resolution. Boundary conditions are monthly mean climatologies taken from the LMDz-‐INCA (Schulz et al. 2006) model for gaseous species and from the GOCART model (Ginoux et al., 2001) for aerosols (desert dust, carbonaceous species and sulphate). Within EC4MACS, the vertical resolution has been improved close to the ground level. Nine vertical levels are selected with a first layer height raising at 20-‐25 m. Six biogenic species (isoprene, -‐pinene, -‐pinene, limonene, ocimene, and NO) were calculated using the MEGAN model (Guenther et al., 2006). We also accounted for fire emissions from GFED. The fire emissions GFED (Global Fire Emissions Database) are derived from satellite observations (Giglio et al., 2010). These emission data are monthly climatologies available on the 1996-‐2009 period at 0.5 ° resolution over Europe. Only fire emissions of CO, NOx, SO2 and PM components have been considered in CHIMERE. Correction of meteorology on urban areas in CHIMERE: In dispersion models at the urban scale, the lower part of the boundary layer is often represented using parameterizations derived from the theory of similarity of the surface layer. The urban effects are then considered by changes in surface roughness and heat flux. Nevertheless, these formulations should only be used in the inertial sublayer (ISL) which is well above the tops of the building. Indeed, in the sub-‐rough layer (RSL or urban sublayer), i.e. in the immediate vicinity of the urban canopy elements, the flow has a rather complex structure (Raupach, 1980) and the similarity theory cannot be applied. Rotach (1993ab, 1995) analyzed mean flow and turbulence measurements inside and above an urban street canyon. He found that one of the characteristic features of the RSL is an increase in the absolute value of Reynolds stress, from essentially zero at the average zero plane displacement height up to a maximum value, which he observed at about two times the average building height. The RSL extends from the ground to a level where horizontal homogeneity of the flow is well established, so about 2 to 5 times the average height of the elements of the canopy (Raupach et al., 1991) or up to several tens of meters in major cities. In a first approach, the Schmidt number defined by the ratio of diffusivity of momentum and mass diffusion coefficient is taken equal to 1. Considering this, Kz is given by :
44
zuwuKz '' (7.2.1)
Considering this, we can assume at urban ground level a negligible value of Kz. Then Kz should increase with the Reynolds stress to a maximum value at 2 to 5 times the average height of the buildings. This first level of CHIMERE is currently 20 m. This is clearly two times under the average of the buildings height in big cities. So that we can assume that the corresponding Reynolds stress at this level has not yet reach his maximum value and the corresponding value of Kz is overestimate even if we can also assume that Kz is underestimate at the second level of CHIMERE. At the first level of CHIMERE, Kz should be given by :
dz
zuwuKz
levelstz
z
1
0
'' (7.2.2)
In a first approach, if we assume that the theory of similarity is applicable above the first level, we can take for urban areas :
2
theroy) similaritythewithcomputedlevelKz (first st level) Kz (z< fir (7.2.3)
This coefficient is applied to correct the wind speeds in the first CHIMERE layer so as to limit the advection of primary pollutants close to the ground. This coefficient is a compromise between several values reported in the literature. REFERENCES Bessagnet, B., A. Hodzic, R. Vautard, M. Beekmann, S. Cheinet, C. Honoré, C. Liousse, L. Rouil, Aerosol modelling with CHIMERE preliminary evaluation at the continental scale, Atmospheric Environment, Volume 38, Issue 18, June 2004, Pages 2803-‐2817, ISSN 1352-‐2310, 10.1016/j.atmosenv.2004.02.034, 2004. Bessagnet B., L.Menut, G.Curci, A.Hodzic, B.Guillaume, C.Liousse, S.Moukhtar, B.Pun, C.Seigneur, M.Schulz Regional modelling of carbonaceous aerosols over Europe -‐ Focus on Secondary Organic Aerosols, Journal of of Atmospheric Chemistry, 61, 175-‐202, 2009. Bessagnet B., L.Menut, G.Aymoz, H.Chepfer and R.Vautard, Modelling dust emissions and transport within Europe: the Ukraine March 2007 event, Journal of Geophysical Research -‐ Atmospheres, 113, D15202, doi:10.1029/2007JD009541, 2008. Emberson, L.D., Ashmore, M.R., Simpson, D., Tuovinen, J.-‐P., Cambridge, H.M., 2000a. Towards a model of ozone deposition and stomatal uptake over Europe. EMEP/MSC-‐W 6/2000, Norwegian Meteorological Institute, Oslo, Norway, 57 pp. Emberson, L.D., Ashmore, M.R., Simpson, D., Tuovinen, J.-‐P., Cambridge, H.M., 2000b. Modelling stomatal ozone flux across Europe. Water, Air and Soil Pollution 109, 403-‐413.
45
Giglio, L., Randerson, J. T., van der Werf, G. R., Kasibhatla, P. S., Collatz, G. J., Morton, D. C., and DeFries, R. S.: Assessing variability and long-‐term trends in burned area by merging multiple satellite fire products, Biogeosciences, 7, 1171-‐1186, doi:10.5194/bg-‐7-‐1171-‐2010, 2010. Ginoux, P., Chin, M., Tegen, I., Prospero, J.M., Holben, B., Dubovik, O., Lin, S.-‐J., Sources and distributions of dust aerosols simulated with the GOCART model. Journal of Geophysical Research 106, 20,255 20,273, 2001. Guenther, A., Karl, T., Harley, P., Wiedinmyer, C., Palmer, P., Geron, C., 2006. Estimates of global terrestrial isoprene emissions using MEGAN (Model of Emissions of Gases and Aerosols from Nature). Atmos. Chem. Phys. 6, 31813210, 2006. Monahan, E.C., Spiel, D.E., Davidson, K.L., 1986. A model of marine aerosol generation via whitecaps and wave disruption. Oceanic Whitecaps and their role in air/sea exchange, edited by Monahan, E.C, and Mac Niocaill, G., pp. 167-‐174 Nenes, A., Pilinis, C., and Pandis, S. N., 1999, Continued development and testing of a new thermodynamic aerosol module for urban and regional air quality models, Atmos. Environ., 33, 1553 1560, 1999. Schmidt, H., Derognat, C., Vautard, R., Beekmann, M. A comparison of simulated and observed ozone mixing ratios for the summer of 1998 in western Europe. Atmospheric Environment 35, 6277 6297, 2001. Schulz, M., Textor, C., Kinne, S., Balkanski, Y., Bauer, S., Berntsen, T., Berglen, T., Boucher, O., Dentener, F., Guibert, S., Isaksen, I.S.A., Iversen, T., Koch, D., Kirkevå g, A., Liu, X., Montanaro, V., Myhre, G., Penner, J.E., Pitari, G., Reddy, S., Seland, O., Stier, P., Takemura, T., Radiative forcing by aerosols as derived from the AeroCom present-‐day and pre-‐industrial simulations. Atmos. Chem. Phys. 6, 5225 5246, 2006.
-‐Tunnel Study of Turbulent Flow Close to Regularly Arrayed Roughness elements, Boundary-‐Layer Meteorol. 18, 373 397, 1980.
-‐Wall Turbulent Boundary Layers, Appl. Mech. Rev. 44, 1 25, 1991.
Reynolds Stress, Boundary-‐Layer Meteorol. 65, 1 28, 1993a.
Boundary-‐Layer Meteorol. 66, 75 92, 1993b. Troen, I. and L. Mahrt, 1986: A simple model of the atmospheric boundary layer: Sensitivity to surface evaporation. Bound.-‐Layer Meteorol., 37, 129-‐148. Vautard, R., B. Bessagnet, M. Chin, and L. Menut: On the contribution of natural Aeolian sources to particulate matter concentrations in Europe: testing hypotheses with a modelling approach, Atmospheric Environment, 39, 3291 3303, 2005. B. van Leer, Multidimensional explicit difference schemes for hyperbolic conservation laws, in Computing Methods in Applied Sciences and Engineering VI , edited by R. Growinski and J. L. Lions (Elsevier, Amsterdam, 1984)
46
7.3 LOTOS-‐EUROS model description In this study we used LOTOS-‐EUROS v1.8, a 3-‐D regional CTM that simulates air pollution in the lower troposphere. Previous versions of the model have been used for the assessment of (particulate) air pollution (e.g. Schaap et al., 2004a; 2004b; 2009; Barbu et al., 2009; Manders et al., 2009; 2010). For a detailed description of the model we refer to Schaap et al. (2008), Wichink Kruit et al. (2012) and abovementioned studies. Here, we describe the most relevant model characteristics and the model simulation performed in this study. The model uses a normal longitude latitude projection and allows to specify the model resolution and domain within its master domain encompassing Europe and its periphery. The model top is placed at 3.5 km above sea level and consists of three dynamical layers: a mixing layer and two reservoir layers on top. The height of the mixing layer at each time and position is extracted from ECMWF meteorological data used to drive the model. The height of the reservoir layers is set to the difference between ceiling (3.5 km) and mixing layer height. Both layers are equally thick with a minimum of 50 m. If the mixing layer is near or above 3500 m high, the top of the model exceeds 3500 m. A surface layer with a fixed depth of 25 m is included in the model to monitor ground-‐level concentrations. Advection in all directions is handled with the monotonic advection scheme developed by Walcek (2000). Gas phase chemistry is described using the TNO CBM-‐IV scheme (Schaap et al., 2009), which is a condensed version of the original scheme by Whitten et al. (1980). Hydrolysis of N2O5 is described following Schaap et al. (2004a). Aerosol chemistry is represented with ISORROPIA2 (Fountoukis and Nenes, 2007). The pH dependent cloud chemistry scheme follows Banzhaf et al. (2011). Formation of course-‐mode nitrate is included in a dynamical approach (Wichink Kruit et al., 2012). Dry deposition for gases is modelled using the DEPAC3.11 module, which includes canopy compensation points for ammonia deposition (Van Zanten et al., 2010). Deposition of particles is represented following Zhang et al. (2001). Stomatal resistance is described by the parameterization of Emberson et al. (2000a,b) and the aerodynamic resistance is calculated for all land use types separately. Wet deposition of trace gases and aerosols are treated using simple scavenging coefficients for gases (Schaap et al., 2004b) and particles (Simpson et al., 2003). The model set-‐up used here does not contain secondary organic aerosol formation or a volatility basis set approach as we feel that the understanding of the processes as well as the source characterization are too limited for the current application. REFERENCES Banzhaf, S., Schaap, M., Kerschbaumer, A., Reimer, E., Stern, R., van der Swaluw, E., Builtjes, P., 2011. Implementation and evaluation of pH-‐dependent cloud chemistry and wet deposition in the chemical transport model REM-‐Calgrid, Atmospheric Environment 49, 378-‐390 Barbu, A.L., Segers, A.J., Schaap, M., Heemink, A.W., Builtjes, P.J.H., 2009. A multi-‐component data assimilation experiment directed to sulphur dioxide and sulphate over Europe. Atmospheric Environment 43, 1622-‐1631
47
Emberson, L.D., Ashmore, M.R., Simpson, D., Tuovinen, J.-‐P., Cambridge, H.M., 2000a. Towards a model of ozone deposition and stomatal uptake over Europe. EMEP/MSC-‐W 6/2000, Norwegian Meteorological Institute, Oslo, Norway, 57 pp. Emberson, L.D., Ashmore, M.R., Simpson, D., Tuovinen, J.-‐P., Cambridge, H.M., 2000b. Modelling stomatal ozone flux across Europe. Water, Air and Soil Pollution 109, 403-‐413. Fountoukis, C., Nenes, A., 2007. ISORROPIA II: A computationally efficient thermodynamic equilibrium model for K+-‐Ca2+-‐Mg2+-‐NH4
+-‐Na+-‐SO42-‐-‐NO3
-‐-‐Cl-‐-‐H2O aerosols. Atmospheric Chemistry and Physics 7 (17), 4639-‐4659. Manders, A.M.M., Schaap, M., Hoogerbrugge, R., 2009. Testing the capability of the chemistry transport model LOTOS-‐EUROS to forecast PM10 levels in the Netherlands. Atmospheric Environment 43, 4050-‐4059, doi:10.1016/j.atmosenv.2009.05.006. Manders, A.M.M., Schaap, M., Querol, X., Albert, M.F.M.A., Vercauteren, J., Kuhlbusch, T.A.J. & Hoogerbrugge, R., 2010. Sea salt concentrations across the European continent. Atmospheric Environment 44, 2434-‐2442 Schaap, M., Denier Van Der Gon, H.A.C., Dentener, F.J., Visschedijk, A.J.H., Van Loon, M., Ten Brink, H.M., Putaud, J.-‐P., Guillaume, B., Liousse C., Builtjes, P.J.H., 2004a. Anthropogenic Black Carbon and Fine Aerosol Distribution over Europe. Journal of Geophysical Research 109, D18201, doi: 10.1029/2003JD004330 Schaap, M., van Loon, M., ten Brink, H.M., Dentener, F.D., Builtjes, P.J.H., 2004b. Secondary inorganic aerosol simulations for Europe with special attention to nitrate. Atmospheric Chemistry and Physics 4, 857-‐874. Schaap, M., Sauter, F., Timmermans, R.M.A., Roemer, M., Velders, G., Beck, J., Builtjes, P.J.H., 2008. The LOTOS-‐EUROS model: description, validation and latest developments, International Journal of Environment and Pollution 32 (2), 270 290 Schaap, M., Manders, A.A.M., Hendriks, E.C.J., Cnossen, J.M., Segers, A.J.S., Denier van der Gon, H.A.C., Jozwicka, M., Sauter, F.J., Velders, G.J.M., Matthijsen, J., Builtjes, P.J.H., 2009. Regional Modelling of Particulate Matter for the Netherlands. PBL report 500099008, Bilthoven, The Netherlands. (http://www.rivm.nl/bibliotheek/rapporten/500099008.pdf) Simpson, D., Fagerli, H., Jonson, J.E., Tsyro, S., Wind, P., Tuovinen, J-‐P., 2003. Transboundary Acidification, Eutrophication and Ground Level Ozone in Europe, Part 1: Unified EMEP Model Description. EMEP Report 1/2003, Norwegian Meteorological Institute, Oslo, Norway Van Zanten, M.C., Sauter, F.J., Wichink Kruit, R.J., Van Jaarsveld, J.A., Van Pul, W.A.J., 2010. Description of the DEPAC module: Dry deposition modelling with DEPAC_GCN2010. RIVM report 680180001/2010, Bilthoven, the Netherlands Walcek, C.J., 2000. Minor flux adjustment near mixing ratio extremes for simplified yet highly accurate monotonic calculation of tracer advection. Journal of Geophysical Research D: Atmosphere 105 (D7), 9335-‐9348 Whitten, G., Hogo, H., Killus, J., 1980. The Carbon Bond Mechanism for photochemical smog, Environmental Science and Technology 14, 14690-‐14700.
48
Wichink Kruit, R., Schaap, M., Sauter, F., Van der Swaluw, E., Weijers, E., 2012. Improving the understanding of the secondary inorganic aerosol distribution over the Netherlands. TNO report TNO-‐060-‐UT-‐2012-‐00334, Utrecht Zhang, L., Gong, S., Padro, J., Barrie, L., 2001. A size-‐segregated particle dry deposition scheme for an atmospheric aerosol module. Atmospheric Environment 35, 549-‐560
49
7.4 REM-‐CALGRID (RCG) model description Overview: REM-‐CALGRID is an urban/regional scale chemical Eulerian grid model development. Rather than creating a completely new model, the urban-‐scale photochemical model CALGRID (Yamartino et al., 1992) and the regional scale model REM3 (Stern, 1994; Hass et al., 1997; Römer et al., 2003) were used as the starting point for the new ur-‐ban/regional scale model, REM-‐CALGRID (RCG). The premise was to design an Eulerian grid model of medium complexity that fulfills the requirements of the European
on air quality modelling. The directive demands that a model must be capable of hourly predictions for periods of a year or more. To comply with these preconditions, RCG can be used on the regional and the urban scale for short-‐term and long-‐term simulations of oxidant and aerosol formation. The model includes the following features:
A generalized horizontal coordinate system, including latitude-‐longitude coordi-‐nates;
A vertical transport and diffusion scheme that correctly accounts for atmos-‐pheric density variations in space and time, and for all vertical flux components when employing either dynamic or fixed layers;
A new methodology to eliminate errors from operator-‐split transport and to ensure correct transport fluxes, mass conservation, and that a constant mixing ratio field remains constant;
Inclusion of the recently improved and highly-‐accurate, monotonic advection scheme developed by Walcek (2000). This fast and accurate scheme has been further modified to exhibit even lower numerical diffusion for short wavelength distributions;
The latest release of the CBM-‐IV photochemical reaction scheme; The ISORROPIA equilibrium aerosol modules, that treats the thermodynamics of
inorganic aerosols; An simplified version of the SORGAM equilibrium aerosol module, that treats
the thermodynamics of organic aerosols; Simple modules to treat the emissions of sea salt aerosols and wind-‐blown dust
particles; A simple wet scavenging module based on precipitation rates; An emissions data interface for long term applications that enables on-‐the-‐fly
calculations of hourly anthropogenic and biogenic emissions; One-‐way-‐nesting capabilities.
Horizontal and vertical grid system: Pollutant concentrations in RCG are calculated at the center of each grid cell volume, representing the average concentration over the entire cell. Meteorological fields provided externally by a meteorological driver are internally assigned to a staggered grid arrangement. State variables such as temperature, pressure or water vapor are as the pollutants located at cell center, and represent grid cell average conditions. Wind components and diffusion coefficients are defined at cell interfaces to describe the
50
transfer of mass in and out of each cell face, thus allowing to solve the transport equations in flux form. The RCG model uses a 3-‐dimensional system of grid cells. Two types of horizontal grids are possible
a rectangular grid (e.g., Lambert Conformal or UTM), where grid cells are assumed square and always have the same area,
a geographical grid, where longitude represents east-‐west location, latitude represents north-‐south location.
In the vertical the coordinates are terrain following with the top of the modelling domain a fixed height above the local terrain. There are two choices for the vertical grid system:
An arbitrary, in space and time fixed grid with layer heights, A dynamic grid with an arbitrary amount of layers varying in both space and
time according to the horizontal and vertical variations of the mixing height. In RCG, the 1-‐dimensional, highly-‐accurate, monotonic advection scheme developed by Walcek (2000) is used for the horizontal advection of pollutants. This algorithm uses a low-‐spatial-‐order accuracy definition of within-‐cell concentrations, coupled with a
creation of small numerical diffusion, good transport fidelity in terms of phase speed, group velocity, and low shape
distortion, prohibition of negative concentrations, ability to cope with space-‐time varying vertical level structures, mass conservation, limitations on the creation of new maximum or minimum concentration
values, and, complete monotonicity of mixing ratios.
Vertical diffusion parameters are derived using the Monin-‐Obukhov similarity theory for the description of the structure of the diabatic surface layer (K-‐theory). Chemistry: An updated version of the lumped gas phase chemistry scheme CBM-‐4 (Gery et al.,
-‐Product Isoprene scheme (Carter, 1996), is used for the simulations. Homogeneous and heterogeneous conversion of NO2 to HNO3 is added. In addition to gaseous phase, also simple aqueous phase conversion of SO2 to H2SO4, through oxidation by H2O2 and ozone, has been incorporated. Equilibrium concentrations for SO2, H2O2 and ozone are calculated using Henry constants and assuming progressive cloud cover for relative humidity above 80%. Effective rate constants for the aqueous phase reactions SO2+H2O2 and SO2+O3 have been calculated for an average pH of 5 using acid/base equilibrium and kinetic data from Seinfeld and Pandis (1998).
51
For numerical solution of the chemistry differential equations system, an Eulerian backward iterative method with a variable time step control is used. Photolysis rates are derived for each grid cell assuming clear sky conditions as a function of solar zenith angle, altitude and total ozone column. The Tropospheric Ultraviolet-‐Visible Model (TUV) radiative transfer and photolysis model, developed at the National Center of Atmospheric Research (Madronich and Flocke, 1998) is used to provide the RCG model with a multi-‐dimensional lookup table of photolytic rates. Aerosol treatment: A bulk approach is used for the aerosol formation, i.e. aerosol growth is not considered and the major PM constituents are treated as a single model species with a given log-‐normal size distribution. RCG treates two modes: a fine mode (PM<2.5 µm) and a coarse mode (2.5 µm <PM<10 µm). The equilibrium between gaseous nitric acid, ammonia and particulate ammonium nitrate and ammonium sulphate and aerosol water is calculated with the ISORROPIA thermodynamic module (Nenes et al., 1999) as a function of temperature and humidity. Production of secondary organic aerosol (SOA) is treated with a simplified version of the SORGAM module (Schell et al., 2001) which calculates the partitioning of semi-‐volatile organic compounds produced during VOC oxidation between the gas and the aerosol phase. Overall, in RCG the following different chemical fractions are considered to contribute to PM10, i.e. particulate matter with a dynamical diameter up to 10 µm:
PM10=PMcoarse+PM2.5prim+EC+OCprim+SOA+SO42-‐+NO3
-‐+NH4++Na++Cl-‐
PMcoarse = mineral coarse particles, emitted by anthropogenic and natural sources PM2.5prim = mineral fine particles, emitted by anthropogenic and natural sources EC = elemental Carbon, fine particles, emitted by anthropogenic sources OCprim = organic Carbon, fine particles, emitted by anthropogenic sources SO4
2-‐ = sulphate aerosol, fine particles, formed in the atmosphere NO3
-‐ = nitrate aerosol, fine particles, formed in the atmosphere NH4
+ = ammonium aerosol, fine particles, formed in the atmosphere
SOA = sum of the secondary organic aerosols, formed in the atmosphere Na+ = sea salt sodium, coarse particles, emitted by sea water Cl-‐ = sea salt chloride, coarse particles, emitted by sea water Dry and wet deposition: Dry deposition for gaseous species and particles is calculated using the resistance analogy. Turbulent and laminar resistance are calculated from surface roughness, Monin-‐Obukhov length, friction velocity, molecular diffusivity. Surface resistance is computed following Erisman and Pul (1994) for different surfaces taking into account species dependent (Henry constant, oxidation power), micro-‐meteorological, and land-‐use information. For particles, surface resistance is zero. The atmospheric resistances are large for particles in the accumulation mode (0.1 m<Ø<1 m), because neither Brownian motion, nor sedimentation are effective pathways; these resistances are
52
calculated for the different species using the fixed size distributions given above. Wet deposition of gases due to in and below cloud scavenging is parameterised as a function of the species dependent Henry constant and the precipitation rate. Wet deposition of particles is treated in RCG using a simple scavenging coefficient approach with identical coefficients for all particles. Meteorological Input: RCG requires meteorological data at hourly intervals. They consist of the following three-‐dimensional fields, which must cover the whole three-‐dimensional model domain:
U-‐ and V-‐wind components, temperature, water vapour, density. Two-‐dimensional fields are:
Monin-‐Obukov-‐length, friction velocity, precipitation, cloud cover, mixing layer height, surface temperature, surface wind speed, snow cover.
For standard applications, all this meteorological data is produced employing a diagnostic meteorological analysis system based on an optimum interpolation procedure on isentropic surfaces developed at Freie Universität Berlin. The system utilizes all available observed synoptic surface and upper air data as well as topographical and land use information (Reimer and Scherer, 1992). Other data sources can be used, if available. Emissions input: RCG model requires annual emissions of VOC, NOx, CO, SO2, CH4, NH3, PM10, and PM2.5, split into point and gridded area sources. Mass-‐based, source group dependent NMVOC profiles are used to break down the total VOC into the different species classes of the chemical mechanisms. Hourly emissions are derived during the model run using sector-‐dependent, month, day-‐of-‐week and hourly emissions factors. PM10 emissions are split into a PM2.5 and a coarse PM (PM10 PM2.5) part, the PM2.5 part is further split into mineral dust, EC and primary OC. EC fractions in PM2.5 emissions for different SNAP sectors are taken from Builtjes et al. (2003). For primary OC, the following crude method is applied to estimate these emissions in a preliminary way: In the 1996 NEI emission data base (http://www.epa.gov/ttn/chief/), an average OCprim/EC emission ratio for all sectors of about two can be derived for the US. This ratio is then applied to Europe regardless of the SNAP sector, i.e. the OC fractions are set as the double of the EC fractions, except if the sum of the two factors would be larger than unity. In this case (fEC>0.33), fOC is set as: fOC fEC. Biogenic VOC-‐emissions are derived using the emissions factors for isoprene and OVOC (Other VOCs) as described in Simpson et al. (1999). Terpene emission factors are taken from the CORINAIR emission hand-‐book. The biogenic calculations of trees are based on the land-‐use data for 115 forest trees (Koeble and Seufert, 2002). Light intensity and temperature dependencies are also considered. Soil NO emissions are calculated as a function of fertilizer input and temperature following Simpson et al. (1999).
53
Resuspension of mineral aerosol from natural soils is included as a function of friction velocity and the nature of soil; both the direct entrainment of small particles (Loosemore and Hunt, 2000) and saltation, i.e. the indirect entrainment due to large particles which fall back to the soil and entrain smaller particles (Claiborn et al., 1998) is taken into account. The sea-‐(1997) and Monahan et al. (1986) as a function of size and wind speed. REFERENCES
Beekmann M., A. Kerschbaumer, R. Stern, E. Reimer, D. Möller., 2007, PM measurement campaign HOVERT in the Greater Berlin area: model evaluation with chemically specified particulate matter observations for a one year period, Atmos. Chem. Phys.,7, 55-‐68, 2007 Builtjes, P. J. H.,2003, Aerosols over Europe, Focus on Black carbon, TNO report R2003/146, February 2003. Carter, W.,1996, Condensed atmospheric photooxidation mechanisms for isoprene, Atmos. Environ., 30, 4275 4290, 1996. Claiborn, C., Lamb, B., Miller, A., Beseda, J., Clode, B., Vaughan, J., Kang, L., and Nevine, C., 1996, Regional measurements and modelling of windblown agricultural dust: The Columbia Plateau PM10 Program, J. Geophys. Res., 103(D16), 19 753 19 767, 1998. Erisman, J. and Van Pul, A.,1994, Parameterization of surface resistance for the quantification of atmospheric deposition of acidifying pollutants and ozone, Atmos. Environ., 28, 2595 2607, 1994. Gery, M. W., Witten, G. Z., Killus, J. P., Dodges, M. C.,1989, A photochemical kinetics mechanism for urban scale and regional scale computering modelling, J. Geophys. Res., 94(D10), 12925 12956, 1989. Gong, S. L., Barrie, L. A., and Blanchet, J.-‐P., 1997, Modelling sea-‐salt aerosols in the atmosphere. 1. Model development, J. Geophys. Res., 102, 3805 3818, 1997. Hass, H., Bultjes, P. J. H., Simpson, D., and Stern, R.,1997, Comparison of model results obtained with several European regional air quality models, Atmos. Environ., 31, 3259 3279, 1997, Koeble, R. and Seufert, G., 2002, Novel maps for forest trees in Europe. Proceedings of the 8th European Symposium on the Physico-‐Chemical Behaviour of Atmospheric Pollutants. JRC Ispra. Loosemore, G. A. and Hunt, J. R., 2000, Dust resuspension without saltation, J. Geophys. Res., 105(D16), 20 663 20 5 671, 2000. Madronich, S. and S. Flocke (1998). The role of solar radiation in atmospheric chemistry, in Handbook of Environmental Chemistry (P. Boule, ed.), Springer_Verlag, Heidelberg, pp. 1-‐26, 1998. Monahan, E.C, Speil, D., Speil, K. (1986), Oceanic whitecaps. Reidel 1986
54
Nenes, A., Pilinis, C., and Pandis, S. N.,1999, Continued development and testing of a new thermodynamic aerosol module for urban and regional air quality models, Atmos. Environ., 33, 1553 1560, 1999. Odum, J. R., Hoffmann, T., Bowman, F., Collins, D., Flagan, R. C., and Seinfeld, J. H.,1996, Gas/Particle partitioning and secondary organic aerosol yields, Environ. Sci. Technol., 30, 25802585, 1996. Reimer, E. und Scherer, B. (1992), An operational meteorological diagnostic system for regional air pollution analysis and long term modelling, in Air Pollution Modelling and its Application IX, eds. H. v. Dop und G. Kallos, NATO Challenges of Modern Society, Kluwer Academic/Plenum Publisher, New York. Römer, M., Beekmann, M., Bergström, R., Boersen, G., Feldmann, H., Flatøy, F., Honore, C., Langner, J., Jonson, J. E., Matthijsen, J., Memmesheimer, M., Simpson, D., Smeets, P., Solberg, S., Stern, R., Stevenson, D., Zandveld, P., and Zlatev, Z.,2003, Ozone trends according to ten dispersion models, EUROTRAC-‐2 special report, GSF, Munich, Germany, 2003. Schell, B., Ackermann, I. J., Hass, H., Binkowski, F., and Ebel, A.,2001, Modelling the formation of secondary organic aerosol within a comprehensive air quality model system, J. Geophys. Res., 106(D22), 28 275 75 28 293, 2001. Seinfeld, J. and Pandis, S. N.: Atmospheric Chemistry and Physics, John Wiley and Sons, 1998. Simpson, D., Winiwarter, W., Börjesson, G., Cinderby, S., Ferreiro, A., Guenther, A., Hewitt, C. N., Janson, R., Khalil, M. A. K., Owen, S., Pierce, T. E., Puxbaum, H., Shearer, M., Skiba, U., Steinbrecher, R., Tarrason, L., and Oiquist, M. G.,1999: Inventorying emissions from nature in Europe, J. Geophys. Res., 104(D7), 8113 8152, 1999. Stern, R., 2003, Entwicklung und Anwendung des chemischen Transportmodells REM-‐CALGRID. Abschlussbericht zum FuE-‐Vorhaben 298 41 252 des Umweltbundes
Stern, R., Yamartino, R., Graff, A. ( 2006). Analyzing the response of a chemical transport model to emissions reductions utilizing various grid resolutions. 28th ITM on Air Pollution Modelling and its Application. May 15-‐19, 2006, Leipzig, Germany Stern, R., Builtjes, P., Schaap, M., Timmermans, R., Vautard, R., Hodzic, A., Memmesheimer, M., Feldmann, H., Renner, E., Wolke, R., Kerschbaumer, A. (2008). A model inter-‐comparison study focussing on episodes with elevated PM10 concentrations. Atmospheric Environment 42 4567-‐4588. 2008. Thunis, P., L. Rouil, C. Cuvelier, R.Stern, A. Kerschbaumer, B. Bessagnet, M. Schaap, P. Builtjes, L. Tarrason, J. Douros, N. Moussiopoulos, G. Pirovano, M. Bedogni (2007). Analysis of model responses to emission-‐reduction scenarios within the CityDelta project, Atmospheric Environment 41 (2007) 208-‐220 Thunis, P., C. P. Roberts, L. White, L. Post, L. Tarrasón, S. Tsyro, R. Stern, A. Kerschbaumer, L. Rou l, B. Bessagnet, R. Bergström, M. Schaap, G.A.C. Boersen, P.J.H. Builtjes (2008). Evaluation of a Sectoral Approach to Integrated Assessment Modelling including the Mediterranean Sea. European Commission Joint Research Centre. Institute for Environment and Sustainability. EUR 23444 EN 2008.
55
Van Loon, M., Roemer, M. G. M., Builtjes, P. J. H., Bessagnet, B., Rouil, L., Christensen, J., Brandt, J., Fagerli, H., Tarrason, L., Rodgers, I., Teasdale, I., Stern, R., Bergström, R., Langner, J., and Foltescu, V., 2004, Model intercomparison in the framework of the review of the Unified EMEP model. TNO-‐report R2004/282, 53 pp. Available at http://www.mep.tno.nl, 2004 Van Loon, L., Vautard,R., Schaap, M., Bergström, R., Bessagnet, B., Brandt, J., Builtjes, P., Christensen, J., Cuvelier, K., Jonson, J., Langner, J., Roberts, L., Rouil, L., Stern, R., Tarrasón, L. Thunis, P., Vignati, P., White, L.,Wind, P. (2007). Evaluation of long-‐term ozone simulations from seven regional air quality models and their ensemble average. Atmospheric Environment 41, 2083-‐2097 Vautard, R., Builtjes, P. H. J., Thunis, P., Cuvelier, K., Bedogni, M. Bessagnet, B., Honoré, C., Moussiopoulos, N., Pirovano, G., Schaap, M. Stern, R. Tarrason, L., Wind, P., 2007, Evaluation and intercomparison of Ozone and PM10 simulations by several chemistry-‐transport models over 4 european cities within the City-‐Delta project. Atmospheric Environment 41 (2007) 173-‐188 Wesely, M. 1989. Parameterization of surface resistance to gaseous dry deposition in regional scale numerical models. Atmos. Environ., 23, 1293-‐1304. Walcek, C. J., 2000, Minor flux adjustment near mixing ratio extremes for simplified yet highly accurate monotonic calculation of tracer advection, J. Geophys. Res., 105(D7), 9335 9348, 2000. Yamartino, R.J., J. Scire, G.R. Carmichael, and Y.S. Chang (1992). The CALGRID mesoscale photochemical grid model-‐I. Model formulation, Atmos. Environ., 26A (1992), 1493-‐1512. Yamartino, R.J., 2003, Refined 3-‐d Transport and Horizontal Diffusion for the REM/CALGRID Air Quality Model. Bericht zum Forschungs-‐ und Entwicklungsvorhaben 298 41 252 auf dem Gebiet
rung der Meteorologie.
Yamartino, R, Flemming, J. Stern, R., 2004, ADAPTATION OF ANA LYTIC DIFFUSIVITY FORMULATIONS TO EULERIAN GRID MODEL LAYERS OF FINITE THICKNESS. 27th ITM on Air Pollution Modelling and its Application. October 25-‐29, 2004, Banff, Canada.
56
7.5 CMAQv5.0 model description CALIOPE is an air quality modelling system developed at the Barcelona Supercomputing Center which currently forecasts air quality for Europe and Spain (Baldasano et al. 2011). The system integrates a meteorological model (WRF-‐ARW), an emission model (HERMES), a chemical transport model (CMAQ) and a mineral dust model (BSC-‐DREAM8b). CMAQ has been developed under the leadership of the Atmospheric Modelling Division of the Environmental Protection Agency (EPA) National Exposure Research Laboratory in Research Triangle Park, North Carolina USA. The modelling system and its source codes are freely available for use by air quality regulators, policy makers, industry, and scientists to address multiscale, multi-‐pollutant air quality concerns. It includes a chemistry transport model that currently allows for the simulation of concentrations and deposition of the major air pollutants. Because of its generalized coordinate system and its advanced nesting features CMAQ can be used to study the behaviour of air pollutant form local to regional scales. A detailed description of the model system is given by Byun and Schere (2006). There are several comprehensive evaluations of CMAQ which establish model credibility for a wide range of application in USA (e.g. Mebust et al., 2003; Eder and Yu, 2006; Appel et al., 2007, 2008; Simon et al, 2012) and Europe (e.g. Matthias, 2008; Pay et al., 2010; Basart et al., 2012). The last version of CMAQ modelling system (version 5.0) used in this exercise integrates the last scientific upgrades. The gas-‐phase oxidations in the atmosphere are described in the CB-‐05 chemical mechanism (Yarwood et al., 2005). Aerosols are represented by the modal aerosol module AERO5. Three log-‐normal modes spanning three size categories Aitken, accumulation and coarse. CMAQv5.0 allows semi-‐volatile aerosol components to condense and evaporate from the coarse mode and non-‐volatile sulphate to condense on the coarse mode. Dynamic mass transfer is simulated for the coarse mode, whereas the fine modes are equilibrated instantaneously with the gas phase. CMAQv5.0 simulates oxidative aging reaction for primary organic aerosol as a second order reaction between reduced primary organic carbon and OH radicals (Simon and Bhave, 2012). Additional biogenic precursors such as isoprene and sesquiterpenes were incorporated in version 4.7. New secondary organic aerosol treatment, explained in Carlton et al. (2010), allows three biogenic and four anthropogenic VOC classes to form a variety of semivolatile and nonvolatile products after reacting with gas-‐phase oxidants. In addition, non-‐volatile SOA can be formed via aqueous-‐phase oxidation of glyoxal and methylglyoxal or oligomerization of semivolatile SOA. In total, 19 separate SOA types are formed and tracked. The production of sea salt emission is implemented as a function of wind speed and relative humidity (Gong, 2003; Zhang et al., 2005) form open ocean and surf zone (Clarke et al., 2006). Thermodynamic equilibrium between gas and inorganic fine aerosols is determined by the ISORROPIAv2, is included in CMAQ5.0.
57
Meteorological input data for the CMAQ model are processed using the WRF-‐ARW model version v3.3.1. Concerning natural emissions, MEGAN v2.0 is used to simulate biogenic emission. Furthermore, the long-‐range transport of mineral dust from Sahara desert is modelled by BSC-‐DREAM8b. The CMAQ horizontal grid resolution corresponds to that of WRF-‐ARW. Its vertical structures was obtained by a collapse from the 38 WRF layers to a total of 15 layers steadily increasing form the surface up to 50 hPa with a stronger density within the PBL. The mean altitude of the lowest layer of CMAQ in CALIOPE system is of 19.5 ± 0.5 m above ground level. REFERENCES Appel, K. W., Bhave, P. V., Gilliland, A. B., Sarwar, G., and Roselle, S. J.: Evaluation of the Community Multiscale Air Quality (CMAQ) model version 4.5: Sensitivities impacting model performance; Part II particulate matter, Atmos. Environ., 42, 6057 6066, 2008. Appel, K. W., Gilliand, A. B., Sarwar, G., and Gilliam, R. C.: Evaluation of the Community Multiscale Air Quality (CMAQ) model version 4.5: Sensitivities impacting model performance, Atmos. Environ., 41, 9603 9615, 2007. Baldasano JM, Pay MT, Jorba O, Gassó S, Jiménez-‐Guerrero P, 2011. An annual assessment of air quality with the CALIOPE modelling system over Spain. Sci Total Environ, 409, 2163-‐2178. doi:10.1016/j.scitotenv.2011.01.041. Basart S, Pay MT, Jorba O, Pérez C, Jiménez-‐Guerrero P, Schulz M, Baldasano JM, 2012. Aerosol in the CALIOPE air quality modelling system: validation and analysis of PM levels, optical depths and chemical composition over Europe. Atmos Chem Phys, 12, 3363-‐3392. doi: 10.5194/acp-‐12-‐3363-‐2012. Byun, D. and Schere, K. L.: Review of the Governing Equations, Computational Algorithms, and Other Components of the Models-‐3 Community Multiscale Air Quality (CMAQ) Modelling System, Appl. Mech. Rev., 59, 51 77, 2006. Carlton, A. G., Bhave, P. V., Napelenok, S. L., Edney, E. O., Sarwar, G., Pinder, R. W., Pouliot, G. A., and Houyoux, M.: Model representation of secondary organic aerosol in CMAQv4.7, Environ. Sci. Technol., 44, 8553 8560, 2010. Chang, J.S., brost, R.A., isaksen, I.S.A., Madronich, S., Middleton, P., Stockwell, W.R., Walcek, C.J., 1987. A three-‐dimensional Eulerian acid deposition model: physical concepts and formulation. J. Geophy. Res., 92, 14681-‐14700. Clarke, A.D., S.R. Owens, and J. Zhou, 2006. An ultrafine sea-‐salt flux from breaking waves: Implications for cloud condensation nuclei in the remote marine atmosphere. J. Geophys. Res. 111 (D06202). Colella, P., Woodward, 1984. The piecewise parabolic method (PPM) for gas-‐dynamical simulation. J. Comput. Phys., 54, 174-‐201.
58
Eder, B. and Yu, S.: A performance evaluation of the 2004 release of Models-‐3 CMAQ, Atmos. Environ., 40, 4811 4824, 2006. Gong, S.L., 2003. A parameterization of sea-‐salt aerosol source function for sub-‐ and super-‐micron particles. J. Geophys. Res. 17, 1097. doi:10.1029/2003GB002079. Matthias, V., 2008. The aerosol distribution in Europe derived with the community multiscale air quality (CMAQ) model: comparison to near surface in situ and sunphotometer measurements. Atmos. Chem. Phys. 8, 5077-‐5097 Mebust, M. R., Eder, B. K., Binkowski, F. S., and Roselle, S. J.: Models-‐3 Community Multiscale Air Quality (CMAQ) model aerosol component, J. Geophys. Res., 108(D6), 4184, doi:10.1029/2001JD001410, 2003. Pay, M.T., Piot, M., Jorba, O., Gassó, S., Gonçalves, M., Basart, S., Dabdub, D., Jiménez-‐Guerrero, P., Baldasano, J.M., 2010a. A full year evaluation of the CALIOPE-‐EU air quality modelling system over Europe for 2004. Atmos. Environ. 44, 3322-‐3342. Pleim, J., 2007. A combined local and nonlocal closure model for the atmospheric boundary layer. Part I: model description and testing. J. Appl. Met. Climatology, 46, 1383-‐1395. Schell, B., Ackermann, I.J., Hass, H., Binkowski, F.S., Ebel, A., 2001. Modelling the formation of secondary organic within a comprehensive air quality model system. J. Geophys. Res., 106, 28275-‐28293. Simon H., K. R. Baker and S. Phillips, 2012. Compilation and interpretation of photochemical model performance statistics published between 2006 and 2012. Atmos. Environ. 61, 124-‐139. Venkatram, A., Pleim, J., 1999. The electrical analogy does not apply to modelling dry deposition of particles. Atmos. Environ., 33, 3075-‐3076. Walcek, C.J. and Taylor, G.R, 1986. A theoretical method for computing vertical distribution of acidity and sulphate production within cumulus clouds. J. Atmos. Sci. 43, 339-‐355. Wesely, M. 1989. Parameterization of surface resistance to gaseous dry deposition in regional scale numerical models. Atmos. Environ., 23, 1293-‐1304. Yarwood, G., Rao, S., Yocke, M., and Whitten, G., 2005. Updates to the Carbon Bond chemical mechanism: CB05. Final report to the US EPA, RT-‐0400675. [Available at http://www.camx.com/publ/pdfs/cb05_final_report_120805.aspx, access 25 January 2013]. Zhang, Y., Liu, P., Pun, B., Seigneur, C., 2006. A comprehensive performance evaluation of MM5-‐CMAQ for the summer 1999 southern oxidants study episode, part iii: diagnostic and mechanistic evaluations. Atmos. Environ. 40, 4856-‐4873. doi:10.1016/j.atmosenv.2005.12.046.
59
7.6 The SCALEDEP DataBase All data used in the ScaleDep analysis are available on request. The ScaleDep DataBase consists of:
Directory: SD_DATABASE containing all the reported model output for the 4 resolutions M1, M2, M3, M4 Meteorological variables: Kz (vertical diffusion coefficient [m2/sec]), PBL (Planetary boundary layer [m], RAIN (Precipitation [mm/hr]), T2M (Temperature at 2m [K]), USTAR (ustar [m/sec]), U10 (Wind speed at 10 m height [m/sec]). Gas variables: O3, NO2, NO, APINEN, H2O2, HCHO, HNO3, NH3, SO2, PAN, VOC, ISOP in units [µg/m3]. Particulate Matter: PM, PPM, NO3, SO4, NH4, ASOA, BSOA, DUST, SS at 2.5µ fraction and at 10µ fraction. Units [µg/m3]. Deposition: Dry NHx, NOx, SOx, and Wet NHx, NOx, SOx. Units [mg/m2] The reported model results have the following structure (netCDF file):
SD_MODEL_EC4Mx_Variable_Freq_20090101.nc (7.6.1)
with MODEL = [EMEP, CHIM, LOTO, RCGC,CMAQ] Mx = [M1, M2, M3,M4] Variable = see above Freq = [HL, DL], ie hourly values (8760 hours), or daily values (365 days)
The netCDF variables inside (7.6.1) have 3 dimensions (nlon,nlat, freq): nlon number of grid cells in longitude direction, nlat number of grid cells in latitude direction, Frequency. Longitude and Latitude arrays are included as 1 dimensional (nx or ny) or 2 dimensional (nx,ny).
Directory: TIME_FILES_SD containing all processed model results for input to the DeltaTool. The processing has been performed using the DeltaTool PreProcessing utility with variables and station metadata from the startup.ini file and the observational data from the directory monitoring. For details on the use of the DeltaTool and its PreProcessor we refer to the literature in Section 6. The netCDF variables correspond to the stations in the startup.ini file and have 2 dimensions: The time frequency (always 8760) and the number of variables which corresponds to the string format of the global att
Directory: monitoring containing csv files with observational data at the stations. The next Table 7 gives for each of the pollutants PM10, NO2, and O3 the composition of the City groups Urban30, Rural200, Emep200.
60
Table 7. Overview of City groups per pollutant (PM10, NO2, O3), per station type (Urban, Rural, EMEP), and per radius (30 km, 200 km)
Pollutant URE group Station Codes (see startup.ini) PM10 Urban30
(116 stations) AT30601*AT31901*AT9STAD*CZ0ARIE*CZ0ASTO*CZ0SKLM*CZ0SKLS* DEBB021*DEBE010*DEBE018*DEBE034*DEBE068*DEBY039*DEHH008* DEHH015*DEHH033*DEHH059*DEHH074*DEHH076*ES0126A*ES0567A* ES1231A*ES1351A*ES1532A*S1568A*ES1578A*ES1598A*ES1619A* ES1638A*ES1640A*ES1713A*ES1752A*ES1856A*ES1872A*ES1885A* ES1890A*S1900A*FR02031*FR03014*FR03029*FR03047*FR04001* FR04002*FR04004*FR04024*FR04055*FR04098*FR04099*FR20061* FR20062*GB0566A*GB0584A*GB0620A*GB0950A*HU0042A*IE0028A* IE0135A*IT0524A*IT0706A*IT0770A*IT0953A*IT0956A*IT1010A*IT1035A*IT1176A*IT1650A*IT1692A*IT1743A*IT1835A*IT1836A*IT1906A* NL00014*NL00701*PL0010A*PL0038A*PL0039A*PL0120A*PL0125A* PL0132A*PL0133A*PL0134A*PL0141A*PL0143A*PL0216A*PL0273A* PL0274A*PL0399A*PL0404A*PT03071*PT03083*PT03084*PT03085* PT03087*PT03089*PT03093*PT03101*RO0065A*SE0022A
Rural200 (98 stations)
AT0ILL1*AT0PIL1*AT0ZIL1*AT30202*AT30302*AT30403*AT30407* AT30502*AT30603*AT30701*AT31703*AT31902*AT31903*AT31904* AT31905*AT60156*AT72538*BETN012*BETN016*BETN054*BETN063* BETN066*BETN073*BETN085*BETN093*BETN100*BETN113*BETR833* BG0070A*CH0033A*CH0051A*CZ0BKUC*CZ0BLOC*CZ0BMIS*CZ0JDUK* CZ0JKOS*CZ0LFRU*CZ0LJIZ*CZ0LRAD*CZ0LSOU*CZ0PKUJ*CZ0SROZ* CZ0TBKR*CZ0TNUJ*CZ0UDOK*CZ0UKRU*CZ0ULOM*CZ0UMED*CZ0URVH*CZ0USJT*CZ0USMO*CZ0USNZ*CZ0UTUS*CZ0UVAL*DEBB053*DEBB065* DEBB066*DEBB075*DEBE032*DEBE056*DEBY013*DEBY049*DEBY109* DEMV017*DENI059*DENI060*DENI063*DENW064*DENW065*DENW068*DERP015*DERP016*DESH008*DESN074*DEUB005*DEUB030*ES1222A* ES1488A*ES1489A*ES1491A*ES1554A*ES1599A*ES1654A*ES1670A* ES1671A*ES1802A*ES1805A*ES1806A*ES1808A*ES1810A*ES1811A* ES1851A*ES1878A*ES1887A*FR02005*FR04066*FR04158*FR04322* FR04328*FR18019*FR20045*FR25049*GB0617A*GB0953A*HU0040A* IE0090A*IE0126A*IT0989A*IT0992A*IT1418A*IT1464A*IT1646A*IT1904A*IT1911A*IT1948A*NL00131*NL00230*NL00235*NL00246*NL00318* NL00437*NL00444*NL00538*NL00631*NL00633*NL00738*NL00818* NL00918*PL0028A*PL0182A*PL0243A*PL0470A*PT03096*PT03099* PT03102*PT04002*SE0066A
EMEP200 (14 stations)
AT0002R*CZ0003R*DE0002R*DE0007R*ES0001R*ES0008R*ES0009R* ES0012R*ES0013R*GB0036R*HU0002R*IT0001R*SE0012R*SK0007R
NO2 Urban30 (126 stations)
AT30601*AT31901*AT9STAD*AT9STEF*BETB006*CZ0ARIE*CZ0ASTO* CZ0SKLM*CZ0SKLS*DEBB021*DEBE010*DEBE018*DEBE034*DEBE066* DEBE068*DEBY039*DEHH008*DEHH015*DEHH033*DEHH059*DEHH074* DEHH076*ES0124A*ES0126A*ES1231A*ES1351A*ES1422A*ES1425A* ES1532A*ES1551A*ES1568A*ES1578A*ES1598A*ES1619A*ES1638A* ES1640A*ES1644A*ES1653A*ES1679A*ES1713A*ES1752A*ES1816A* ES1856A*ES1885A*ES1890A*FR02031*FR03014*FR03029*FR03047* FR04001*FR04002*FR04004*FR04008*FR04014*FR04017*FR04018* FR04024*FR04037*FR04055*FR04059*FR04060*FR04098*FR04099* FR04100*FR04101*FR04105*FR04146*FR04160*FR20004*FR20061* FR20062*GB0566A*GB0584A*GB0620A*GB0638A*GB0644A*GB0681A* GB0743A*GR0028A*GR0031A*HU0042A*IE0028A*IE0135A*IT0524A* IT0706A*IT0770A*IT0912A*IT0953A*IT0956A*IT1010A*IT1035A*IT1176A*
61
IT1650A*IT1692A*IT1743A*IT1835A*IT1836A*IT1906A*NL00003* NL00014*NL00019*NL00021*NL00022*NL00520*NL00701*PL0010A* PL0033A*PL0034A*PL0038A*PL0039A*PL0134A*PL0141A*PL0143A* PL0273A*PT03010*PT03070*PT03071*PT03082*PT03083*PT03084* PT03085*PT03087*PT03089*PT03093*PT03101*SE0022A
Rural200 (156 stations)
AT0ILL1*AT0PIL1*AT0ZIL1*AT30202*AT30302*AT30403*AT30407* AT30502*AT31502*AT31701*AT31703*AT31902*AT31903*AT31904* AT31905*AT31906*AT60156*AT72123*AT72538*BETN012*BETN016* BETN027*BETN040*BETN046*BETN054*BETN063*BETN066*BETN073* BETN085*BETN093*BETN100*BETN113*BG0070A*CH0033A*CH0051A* CZ0BMIS*CZ0JKOS*CZ0LFRU*CZ0LSOU*CZ0SONR*CZ0TBKR*CZ0UKRU* CZ0ULOM*CZ0UMED*CZ0URVH*CZ0USNZ*CZ0UTUS*CZ0UVAL*DEBB053*DEBB065*DEBB066*DEBB075*DEBE032*DEBE056*DEBE062*DEBY013* DEBY049*DEBY081*DEBY109*DEMV017*DENI059*DENI060*DENI063* DENW064*DENW065*DENW068*DERP015*DERP016*DERP028*DESH008*DESN052*DESN074*DEST069*DEUB005*DEUB030*ES1488A*ES1489A* ES1491A*ES1599A*ES1654A*ES1670A*ES1671A*ES1689A*ES1690A* ES1778A*ES1793A*ES1802A*ES1805A*ES1806A*ES1808A*ES1810A* ES1811A*ES1851A*ES1878A*ES1917A*FR02005*FR02012*FR04038* FR04328*FR11026*FR15001*FR18019*FR20045*FR20049*FR27005* GB0617A*GB0754A*GR0110R*HU0040A*IE0090A*IT0741A*IT0842A* IT0952A*IT0989A*IT0992A*IT1174A*IT1288A*IT1418A*IT1464A*IT1646A*IT1736A*IT1806A*IT1812A*IT1904A*IT1911A*IT1924A*IT1948A* NL00107*NL00131*NL00227*NL00230*NL00235*NL00301*NL00318* NL00437*NL00444*NL00538*NL00620*NL00631*NL00633*NL00738* NL00818*NL00918*PL0014A*PL0028A*PL0058A*PL0128A*PL0182A* PL0243A*PL0314A*PT03096*PT03099*PT03102*PT04002*RO0072A* SE0066A
EMEP200 (10 stations)
ES0001R*ES0008R*ES0009R*ES0012R*ES0013R*GB0034R*GB0036R* GB0037R*GB0038R*GB0045R
O3 Urban30 (93 stations)
AT30601*AT31901*AT9STEF*BETB006*CZ0ASTO*CZ0SKLM*DEBB021* DEBE010*DEBE034*DEBY039*DEHH008*DEHH033*ES0124A*ES0126A* ES1231A*ES1351A*ES1422A*ES1532A*ES1568A*ES1578A*ES1598A* ES1619A*ES1638A*ES1640A*ES1644A*ES1653A*ES1679A*ES1713A* ES1752A*ES1816A*ES1856A*ES1885A*ES1890A*FR03029*FR04002* FR04004*FR04008*FR04017*FR04018*FR04037*FR04049*FR04055* FR04098*FR04100*FR04101*FR04160*FR20004*FR20061*FR20062* GB0566A*GB0584A*GB0620A*GB0638A*GB0644A*GB0681A*GB0743A* GR0028A*GR0031A*HU0042A*IE0028A*IT0524A*IT0706A*IT0770A* IT0912A*IT0953A*IT0956A*IT1010A*IT1035A*IT1176A*IT1650A*IT1692A*IT1743A*IT1835A*IT1836A*IT1906A*NL00520*PL0010A*PL0038A* PL0044A*PL0141A*PL0487A*PT03070*PT03071*PT03082*PT03083* PT03084*PT03085*PT03087*PT03089*PT03093*PT03101*RO0065A* SE0022A
Rural200 (185 stations)
AT0ILL1*AT0PIL1*AT0ZIL1*AT30202*AT30302*AT30403*AT30502* AT30603*AT30701*AT30801*AT31102*AT31502*AT31701*AT31904* AT32101*AT56071*AT60150*AT60156*AT72123*AT72218*AT72538* AT72705*BETN012*BETN016*BETN027*BETN040*BETN046*BETN051* BETN054*BETN063*BETN066*BETN073*BETN085*BETN093*BETN100* BETN113*BG0070A*CH0033A*CH0051A*CZ0BKUC*CZ0BMIS*CZ0CKOC* CZ0JKOS*CZ0LSOU*CZ0SONR*CZ0TBKR*CZ0ULOM*CZ0URVH*CZ0USNZ* CZ0UTUS*CZ0UVAL*DEBB053*DEBB065*DEBB066*DEBB075*DEBE032*
62
DEBE056*DEBE062*DEBY013*DEBY049*DEBY081*DEBY109*DEMV017* DENI059*DENI060*DENI063*DENW064*DENW065*DENW068*DERP015* DERP016*DERP028*DESH001*DESH008*DESH013*DESH014*DESN052* DESN074*DEST069*DEUB005*DEUB030*ES1222A*ES1311A*ES1347A* ES1488A*ES1489A*ES1491A*ES1588A*ES1599A*ES1654A*ES1670A* ES1671A*ES1689A*ES1690A*ES1778A*ES1793A*ES1802A*ES1805A* ES1806A*ES1808A*ES1810A*ES1811A*ES1851A*ES1878A*FR02004* FR02012*FR02023*FR02024*FR03027*FR03031*FR03048*FR03067* FR03083*FR04038*FR04066*FR04142*FR04158*FR04322*FR07029* FR08617*FR11026*FR15001*FR18010*FR18019*FR18036*FR18045* FR20045*FR20049*FR20204*FR25049*FR27005*FR27006*FR32008* FR34043*GB0617A*GB0754A*GR0110R*HU0040A*IE0090A*IE0111A* IT0741A*IT0842A*IT0952A*IT0989A*IT0992A*IT1174A*IT1288A*IT1418A*IT1464A*IT1736A*IT1806A*IT1812A*IT1904A*IT1911A*IT1924A*IT1948A*NL00107*NL00131*NL00227*NL00230*NL00235*NL00301*NL00318* NL00437*NL00444*NL00538*NL00620*NL00631*NL00633*NL00738* NL00818*NL00918*PL0014A*PL0028A*PL0058A*PL0121A*PL0128A* PL0182A*PL0243A*PT03096*PT03099*PT03102*PT04002*RO0072A* SE0066A
EMEP200 (16 stations)
ES0001R*ES0008R*ES0009R*ES0012R*ES0013R*GB0034R*GB0035R* GB0036R*GB0037R*GB0038R*GB0045R*HU0002R*PL0002R*PL0003R* SE0012R*SK0007R