Sensitivity of a spatialized emission inventory of atmospheric pollutants: reconciling environmental concerns with economic aspects
Y. Martinet1, A. Wroblewski1, V. Nollet2 & C. Kergomard3 1Département Chimie et environnement, Ecole Nationale Supérieure des Mines de Douai, France 2Laboratoire de Physico-chimie des processus de combustion et de l'atmosphère, UMR CNRS 8522, Université des Sciences et Technologies de Lille, France 3Laboratoire de géographie des milieux anthropisés, FRE CNRS 2170, Université des Sciences et Technologies de Lille, France
Abstract
Critical realization of a spatialized emission inventory of atmospheric pollutants is an essential step when considering the management of air quality on a meso-scale. It is possible to optimise the performances of such an inventory by reconciling contextual environmental concerns with economic aspects. This could be achieved with various tests of sensitivity concerning several parameters as the resolution of the grid, the level of information of the source files or the diversity of the emissions considered. The results of modeling or simulations using a reactivity or transport model depend on the quality of the information contained in the inventory. The development of a spatialized emission inventory for the Nord - Pas de Calais county lends itself ideally to this kind of work owing to the density and the diversity of the activities it covers. The inventory concerns a large range of pollutants such as SOx, NOx, NMVOC, CO2, CO, CH4, NH3, N2O, HCl, HF, Pb, Zn, Cd, Hg, dioxins and particles. A complementary list exists according to the nature of the activities considered and includes other pollutants among which are Ni, As, benzene or benzo-a-pyrene. Time scale is fixed to one hour considering future atmospheric pollution modeling objectives. The area of study includes all the Nord - Pas de Calais territory. Boundary conditions from Kent, Belgium and the French county Picardie are taken into account. Basic parameters of the inventory are a four square kilometre grid with the highest level of database quality as possible. Various tests of sensitivity are conducted in order to evaluate the impact on modeling results compared to economic investment.
Modeling of air quality is essential to manage air pollution in time and space. It implies a high level of knowledge concerning emissions and dispersion of pollutants in order to deduce level of pollution inside the zone of study. The European directives advocate henceforth the use of modeling as a complementary technique in addition to the measures carried out by air quality associations in slightly polluted areas. In the objective to develop tools of spatialization of the air quality, it is necessary to take part in the comprehension of the phenomena of formation and transport of the secondary photochemical pollutants and with development of applicable strategies of reduction. The comprehension of the phenomena is possible only through numerical modeling of those by means of photochemical computer codes on which results are strongly dependant on the quality of the input data. Critical realization of a spatialized emission inventory represents the basis of any modeling test. It consists of a necessary condition to provide the models calculating atmospheric dispersion and chemical transformations. From a constructive criticism of existing spatialized emission inventories, the methodology can be refined in order to optimise its performances.
2 Evolution of spatialized emission inventories
Spatialized emission inventories appeared during the eighties in the most industrialized countries and include as well anthropogenic and biogenic emissions. They cover important zones on a continental scale and are intended for the comprehension of the phenomena of transport and modeling on large areas. Their limited accuracy constitutes a brake with the development of dispersion simulation of primary and secondary pollutants on finer scales. Then, taking into account the increasing impact of human activities on the environment, the general sudden awareness implies new objectives concerning air pollution. In Europe appear programs intended on dispersion of pollutants and ozone formation on meso-scale. For example, REKLIP is carried out with the beginning of the eighties-ten from a transborder cooperation between France, Switzerland and Germany. It leads to the development of a methodology of spatialized emission inventories who has been applied to the higher Rhine valley for the years 1990 and 1992 (Ponche et al. [2]). It consists in the first complete inventory with high spatial and temporal resolution. Several inventories are following this first experience to end in a basis for the establishment of time inventories for various simulations of estival pollution by ozone. Program PRIMEQUAL-PREDIT (year 2000) is then launched on the agglomeration of Strasbourg in order to carry out investigations concerning the development of a statistical forecast system used for ozone peaks. Recent preoccupations at the French level consist to use very high spatial and temporal resolution inventories in order to evaluate various models of chemistry-transport in troposphere at a meso-scale. This ambitious program called ESCOMPTE is destined to compare their aptitude to describe the reality of the
tropospheric environment of the area of Berre-Marseille. In terms of methodology, several points are especially taking into consideration. The possibility to transfer the method to other areas or to generalize it to more important zones is essential as well as the introduction of reactualisation procedures in order to reduce delays and costs of these phases. Output data has also to be adaptable to the user in terms of precision and the estimation of uncertainties must be the most complete as possible (François et al. [1]).
2.1 Recent applications of spatialized emission inventories
Applications of these modern databases are nowadays potentially numerous. From these, studies likely to support the future economic and protection of the air policies are possible. The fields concerned are as well physico-chemistry as the public health, or the management of the environment in general and on various scales. Impact studies can be lead and strategies of reduction of the emissions elaborated to reach these objectives. For example, it is possible to apprehend the sensitivity of the quality of the air to the standards of European emissions by 2015 (Vinuesa et al. [5]), or to model the impacts of the automobile reformulated and oxygenated fuels on the primary and secondary pollutants (Ponche et al. [2]).
2.2 Nord - Pas de Calais program preoccupations
Modeling of pollutants dispersion and transformation on a regional scale implies the knowledge of flows, chemical nature and detail of the emissions of each branch of activity in space and time. This requires the establishment of databases whose precision, level of information and flexibility of actualisation are determining to solve encountered problems. The modeling phase consists then in fixing available models and in validating them by comparison between results of calculations and measures realized by the air quality inspection networks. Integration of all the data in a geographic information system of the air quality makes it possible to determine charts of exposure of the populations to medical risk, from which it is possible to imagine a reasoned development of the territory. It is important to take into consideration that these environmental concerns have to be in accordance with some economic aspects. Developing and perpetuating such a tool as a spatialized emission inventory of atmospheric pollutants on a regional scale necessitates to involve human and financial means in time. To optimize quality of the information and possibilities of regular reactualisations, Nord - Pas de Calais program proposes to test the sensibility of a spatialized inventory in regards of accuracy of some parameters as resolution of the grid, level of information of the source files or diversity of the emissions considered. In order to enjoy of a maximum of possibilities, primary data has to be as complete as possible and known with a maximum of accuracy. From this data, it is thus possible to debase the quality of the information and to apprehend the impact of this alteration in analysing the output of the models, considering that the effectiveness on the photochemical codes and the models of dispersion depends on the quality of the information contained in the inventory.
Two means steps need to be distinguished during the elaboration of the spatialized inventory. The first consists in the choice, the collection and the organization of all the data which will constitute the emission inventory. The second phase can also occur and contains a geographic information system methodology and a spatialization of the inventory in a grid which can be applicable to the models.
3.1 General characteristics of the emission inventory
Several parameters which will be potentially debased during the tests of sensibility have to be fixed to begin the study: area considered, spatial and temporal resolution, pollutant direct impact and further reactivities taken into account, and pollutants discriminated.
3.1.1 Area of study The zone of study includes the entirety of the Nord - Pas de Calais county and appears in Figure 1. The boundary conditions are to be taken into account for Belgium, Kent (England) and Picardie (France) counties. Total surface covered is of two hundred kilometres from east to west and one hundred and twenty kilometres from north to south, which equals to twenty-four thousand square kilometers.
3.1.2 Spatial resolution Definition of the grid is a fundamental stage of the realization of the spatialized emission inventory, implied by the objectives to reach. It is indeed necessary to consider the availability and the cost of the data, as well as the final objective of modeling meso-scale air quality. Lastly, uncertainties on the data have to be apprehended with a maximum of details. Nord - Pas de Calais county presents dense industrial fabric and road network. Density of the population is consequent. However, the accuracy of the data can be limited sometimes on a communal scale and plural sources of uncertainties resulting from the use of models themselves as well as the quality of the data does not make relevant the definition of a grid lower than four kilometers square. It is however possible to carry out a finer spatialized emission inventory on the level of great agglomerations such as those of Lille or Dunkerque thanks to a finer knowledge of some data, in particular concerning the urban traffic and the residential sector.
3.1.3 Temporal resolution Modeling on a meso-scale the fields of concentration of the atmospheric pollutants in the lower atmosphere requires a model generating climatological fields, but also a module of chemistry describing the complex mechanisms of the physical-chemistry of the troposphere. Some cycles of formation-consumption such as the one of tropospheric ozone present chemical reactions kinetically limitant and of which characteristic times are of about one hour. The data bases
must thus supply the chemical module on this time basis to lead to a compatible modeling of the phenomena to the field realities.
Figure 1: Nord - Pas de Calais situation into the area of study. Geographical projection used is Universal Transverse Mercator zone 31N (UTM) with Nouvelle Triangulation Française (NTF) coordinates.
It is only in this condition, and combined at an adequate data base weather, that the elaborate tool can allow a better apprehension of the episodes of pollution. It is then a question of evaluating the causes and of elaborating adapted strategies of emission control.
3.1.4 Pollutant activities covered Structure of the emission inventory is based on Selected Nomenclature for sources of Air Pollution nomenclature (SNAP), developed as part of the CORe INventory of AIR emissions project (CORINAIR) for distinguishing emission source sectors, sub-sectors and activities. An informatic form is established for each one of the SNAP sub-sectors and several individual sheets are added for some specific activities generating characteristic emissions. Supporting regional economic activities data bases with this nomenclature induces a total coverage of pollutant activities of the area.
3.1.5 Listing of pollutants A mean list of sixteen essential pollutants is established and contains SOx, NOx, CO2, CO, CH4, NH3, N2O, HCl, HF, Pb, Zn, Cd, Hg, PCDD/PCDF, total non methane volatic organic compounds (NMVOC) and total suspended particles (TSP). Indeed, the complexity and the incomplete comprehension of the
troposheric physical-chemistry implies not to be focused on a finished list of pollutants. Actualisation modules are also planed as well as the use of characteristic speciation profiles for NMVOC to supply the models.
3.2 Structure of the emission inventory
The inventory is extracted from database SIRENE, published by the French statistic institute INSEE. It contains all existing economic activity in the Nord - Pas de Calais county, which represents more than 80 000 establishments. Each of them is located with high level of accuracy and affected by an activity code in accordance with French activities nomenclature called NAF. In order to create a link between this nomenclature and the SNAP nomenclature, a specific table has been developed. Each NAF code corresponds to one or several SNAP codes in order to link polluting activities and economic activity. A new database is also created and contains more than 200 000 polluting activities. This database is distributed in 45 EXCEL emission sheets corresponding to one or several SNAP codes according to emission calculation methodology or kind of polluting activity. It is possible to group some emission sheets to distinguish some sectors as industry, residential and tertiary, transport, and biogenic and others. Each sheet contains several chapters among which a listing of sources, emission factors, energy consumptions and some specific calculating procedures.
3.2.1 Industry Industrial emissions have to be distinguished on two kinds of sources: specific and zonal sources. Mathematical expression of these emissions is: k, space unit (km²); t, time unit (h); j, industrial activity ; p, pollutant (t) ; E, emissions (t) ; P, activity (MWh) ; F, emission factor (t/MWh) ; α, specific emissions ; β, zonal emissions (with α=1 and β=0 OR α=0 and β=1). This sector is divided into five parts which contain 20 emission sheets : combustion in energy and transformation industries, combustion in manufacturing industry, production processes, extraction and distribution of fossils fuels and geothermal energy, solvents and other products use. Specific sources consist in 205 installations depending to a particular environmental legislation. These are included in the spatialized emission inventory in 3D. Measures from emission control organisms and industrial declarations of emissions are available for different pollutants among which SOx, NOx, NMVOC, N2O and HCl. Emissions of SOX from these sources are illustrated in Figure 2. Other industrial sources are taken into account as zonal sources at town scale and classified according to the SNAP. Emissions are calculated according to energy consumptions associated to industrial processes and emissions factors extracted from different sources are applied. Each SNAP code corresponds to an
emission file which contains listing of sources, emission factors, energy consumptions, and all elements which can be used to calculate emissions. Output data is a table describing emissions by pollutant at town scale.
Figure 2: Emissions of SOx (ton/yr) from specific sources.
These annual emissions have to be translated by hours thanks to activity profiles from different kind of industries.
3.2.2 Residential and tertiary Residential and tertiary emissions are considered with an exclusive zonal approach. These can be mathematically expressed by this form :
k, space unit (km²) ; t, time unit (h) ; c, kind of fuel ; p, pollutant (t) ; E, emissions (t) ; C, energy consumption (MWh) ; F, emission factor (t/MWh). This sector is composed by a single part entitled non-industrial combustion plants and contain 9 sheets. Residential emissions include heating as well as bathroom hot water consumptions in a single sheet. These are extracted from an INSEE database describing housing of all Nord - Pas de Calais cities. Several parameters are taking into account as surface, kind of fuel used and completion date of the structures. Consumption ratios are applied in function of these parameters to calculate energy consumptions and deduced emissions by the use of classic emission factors. Figure 3 shows zonal NOX emissions of the sector.
Tertiary sector is divided into 8 parts. This separation is laid down by the heterogeneity of databases which implies different methodologies according to the kind of activity considered : offices, business, education, health, transport, sport and culture, café-hotel-restaurant and other habitat forms.
Figure 3: Zonal NOx emissions of residential sector (t/yr).
These annual emissions need to be refined considering meteorological parameters as temperature along the year. Final emission data have to be describe on an hourly basis.
3.2.3 Transport Transport emissions are treated with two different approaches. Due to the density of the road network in major cities and their suburbs, zonal emissions are considered for these. Mean road network is integrated as linear sources. The sector is divided into two single sheet parts : road transport and other mobile sources and machinery. General mathematical expression of these emissions is given by : k, space unit (km²) ; t, time unit (h) ; r, road section ; j, type of vehicle ; p, pollutant (g) ; E, emissions (g/km/h) ; N, flow of vehicles ; L, section length (km) ; F, emission factor (g/km) ; V, average speed (km/h). Several means of transport are included in the spatialized inventory on top of road transport among which railway, air, maritime and river transports.
Methodologies used are extracted from COPERT III and MEET projects. Specific VOC's speciations are used to further apply photochemical models on the inventory.
3.2.4 Biogenic and others This heading contains heterogeneous sources which are generated as well by agriculture as waste treatment. All these emissions are integrated to the inventory as zonal emissions. This sector is divided into three parts for 12 emission sheets :
waste treatment and disposal, agriculture, other sources and sinks. Mathematical expression of these emissions is : k, space unit (km²) ; t, time unit (h) ; b, biogenic or other activity ; p, pollutant (t) ; E, emissions (t) ; B, activity (MWh, B∈N) ; F, emission factor (t/MWh, t). Figure 4 presents CH4 emissions from enteric fermenting of the livestock. Figure 4: Zonal biogenic emissions of CH4 (t/yr) from enteric fermenting of
the livestock.
Meteorological parameters as temperature, rainfall and UV radiations are very implied in this heading, introducing some variability about emissions along the year.
4 Sensitivity tests and economic approach perspectives
Some tests are realized in order to estimate sensitivity and uncertainties of the emission inventory. Two approaches can be distinguished : working about general characteristics and about input data. Testing general characteristics consists in working about parameters as spatial resolution, diversity of the activities covered or emission factor sources. For example, it is possible to use different sources of emission factors and to proceed to comparisons between two methodologies using measures and calculations techniques. Working about input data is used to optimize data treatments and buying in order to increase performances of the spatialized inventory of emissions with the most interesting cost. Validation of the Nord - Pas de Calais spatialized inventory of emissions has to be finished in order to realize advanced sensitivity tests about input data and general characteristics during the summer. Some modeling simulations are planed implying parameters such as urban speed, level of activity, ambient temperature, heavy vehicles and diesel vehicle parts. Evaluating uncertainties from different steps of the inventory building is an essential preoccupation too. The use of emission factors especially generates variability which is difficult to apprehend. Comparing several methods of emission calculation is an interesting way as well as the use of control measures or results of existing inventories.
References
[1] François S., Boissard C., Mirabel Ph., Noppe J., Ponche J.L., 2000. Méthodologie des cadastres d'émissions atmosphériques: application au programme ESCOMPTE. Publications de l'Association Internationale de Climatologie 13, 493-500.
[2] Ponche J.L., Vinuesa J.F., Mirabel Ph., 2000. Modélisation des impacts de carburants automobiles reformulés et oxygénés sur les polluants primaires et secondaires à l'échelle régionale. Publications de l'Association Internationale de Climatologie 13, 518-527.
[3] Ponche J.L., Schneider, Ch., Mirabel Ph., 1995. Inventaire des émissions atmosphériques du fossé rhénan méridional pour l'année 1990. Rapport final REKLIP LPCA - CGS.
[4] PRIMEQUAL, 2000. Investigations concernant le développement d'un système de prévisions statistiques des pointes de pollution par l'ozone sur la zone de Strasbourg-Kehl. Modélisation d'un épisode de pollution à l'ozone sur l'agglomération strasbourgeoise. Rapport final PRIMEQUAL-PREDIT ASPA.
[5] Vinuesa, J.F., Hyacinthe C., Fayet S., Perron G., Mirabel Ph., Ponche J.L., 2000. Sensibilité de la qualité de l'air dans la vallée du Rhin supérieur aux normes d'émissions européennes à l'horizon 2015. Publications de l'Association Internationale de Climatologie 13, 547-556.
