The first workshop of SATURN, the "urban subproject" of EUROTRAC-2, took place on 28 and 29 August 1997 in Aghia Triada, a suburb of Thessaloniki, Greece. About 30 (out of the 36) Principal Investigators (PIs) of the subproject were either personally present or represented in the workshop. In addition, about two dozens of interested researchers, including the Executive Secretary of EUROTRAC-2 Dr. Peter Borrell, attended the workshop and contributed to its success.
In accordance with the above, SATURN is split into four Main Groups of activities (MGs): MOD, VAL, EXP and INT. Each of the MGs is further subdivided into four tasks (see below sections). Led by an experienced scientist, each task corresponds to an ensemble of activities by a limited number of research groups, thus representing a well-concerted and focused joint venture. More information on SATURN may be found on web page http://aix.meng.auth.gr/lhtee/saturn.html.
Dr. Moussiopoulos then introduced the SATURN Status Report 1997, the first of a yearly series intended primarily to promote the collaboration within the subproject and to facilitate the use of scientific results for practical applications. Differently than the more scientifically oriented Annual Report, the Status Report will summarise the key achievements of each Principal Investigator and provide information on his plans for the next year, QA/QC measures foreseen and the funding situation. The SATURN Status Report is accessible on the web via the above mentioned SATURN home page.
According to the SATURN Status Report 1997, funds are secured for more than 75% of the activities planned for Phase A of the subproject (until the summer of 1999). This very positive figure reflects that most funding agencies are aware of the significance that problems related to urban air pollution are adequately addressed. As an additional promising element, the Status Report 1997 allows concluding that most PIs are adhering to the Operational Plan contained in the subproject's work programme. It is, however, substantial that denser contacts are established and maintained both among the PIs and between SATURN and other bodies, including other subprojects of EUROTRAC-2. Moreover, more efforts will be needed to ensure Quality Assurance in the subproject. Promoting synergy within SATURN and agreeing on necessary QA/QC measures were identified as the main targets of the workshop.
The next sessions of the workshop dealt with the four MGs. Each of these sessions started with an introductory talk by the respective MG leader (who, at the same time, is also one of SATURN's Deputy Coordinators). Subsequently, each PI had the chance to briefly present his plans, with emphasis on actions foreseen in the first phase of the subproject. Sufficient time in each of these sessions was devoted to discussion which proved to be very fruitful. Pre-nominated rapporteurs summarised the proceedings in each of these sessions.
This MG aims at establishing a hierarchy of model tools for describing urban air pollution levels. The modelling activities may be subdivided into four tasks: MOD1, street scale processes; MOD2, urban scale processes; MOD3, integration of models on various scales; and MOD4, processes regarding heterogeneous processes and chemical transformations. The model hierarchy has to be defined and important gaps in current modelling capacity to be identified before the end of 1997.
Street scale processes. In most European cities, the hot spots regarding pollution levels are found in street canyons with high traffic density. The buildings represent a blocking for the dispersion of the pollutants emitted from the traffic, leading to high concentration levels inside the street and thereby high air pollution exposure of pedestrians, car drivers etc. The activities under MOD1 are devoted to the development of models for the description of pollution levels in single streets and development of simple human air pollution exposure models for use in various effect studies.
The worse pollution conditions in streets are observed at low wind speeds, where the traffic induced turbulence often is limiting for the pollutant concentrations. Plume models with simple parameterizations of the dispersion inside the street, were in many cases shown to reproduce the observed concentrations well. However, the traffic induced turbulence is not well determined, and in SATURN special attention will therefore be paid to the parameterization of this phenomenon.
Another process which will be studied and which is not fully explored is the possible impact of thermal effects inside street canyons on the dispersion conditions. These effects are expected to be mainly of importance in South Europe.
Emission factors for the various vehicle categories are usually determined by laboratory experiments. These experiments cannot take into account the variations in the car fleet from one area to another. For this purpose a backward calculation procedure has been applied to determine emission factors using traffic counts and air pollution monitor data in Copenhagen. A similar procedure will be applied on monitor data from St. Petersburg and other cities.
Many of the activities in MOD1 will be closely linked to activities in the TRAPOS project, which will provide the possibility for exchange of young scientists between a number of European institutions of which many are participating in SATURN.
Urban scale processes. Air quality in urban areas is affected by local emissions, but also the contributions from regional scale transport play an important role. Within MOD2 a number of different grid models will be further developed and applied to specific cities in order to determine flow and dispersion conditions as well as pollution levels in the urban background air. These models that have a generally coarse grid resolution from more than 100 kilometres down to about 100 metres are using various degrees of nesting in order to resolve the conditions on the urban scale. Special attention will be given the estimation of the error made when several levels of nesting are applied in order to resolve the processes on a scale of 100 metres or even a few tens of metres.
Subgrid processes can be resolved by means of various types of nesting inside models of coarse resolution, but also by means of using results from one model as input to another (model nesting).
Seven urban scale model systems are currently in development in Europe, and all are represented in SATURN. The work in MOD2 therefore involves many groups and many models. In order to harmonize the development and to co-ordinate the work, a workshop will therefore be organized in the early 1998.
Integration of models on various scales. The activities under MOD3 are devoted to the development of the various models and modules that are specific to the urban atmosphere range of scales and to the simulation methods/procedures needed for studying the urban atmospheric processes taking into account the canopy processes. It includes especially the study of the sensitivity of the different nesting methods: self-nesting, model nesting, model chaining.
One of the aims of the modelling activities will be to develop operational systems for implementation under INT. Many of the modelling groups involved in SATURN are working on similar model tools. Exchange of detailed information about the applied parameterizations will be an important part of the work in MOD3. In special cases also source code of subroutines as well as total model systems will be exchanged after specific agreement between the involved institutions. One way to exchange models within SATURN will be visits of scientists at another institution, bringing a model or being introduced to a model at the institute he/she is visiting. In this case new developments in the model will be available for both involved institutions.
Processes regarding heterogeneous processes and chemical transformations. The residence time plays a major role to identify the relative importance of the chemical reactions that take place in the urban area. In a single street the residence time of the air mass can be as short as a few seconds. Then, only very fast chemical reactions have time to take place, and the mechanisms may be simplified considerably and still describe well the governing processes. An example is the system of NO, NO2 and ozone, for which it has been shown in North European streets that it can be described taking into account only the reaction between NO and ozone and the photo-dissociation of NO2.
On the other hand, during anticyclonic pollution episode situations with very low winds, the pollutant residence times within the canopy are at least of the order of several minutes to hours. Pollutants can even stagnate in the lowest layers of the urban atmosphere during many hours in the same area: some photochemical peak pollution episodes have been identified to last over several days, with air mass convection extending over a few kilometres only. This is also the case in some valley systems where air mass is convected back and forth several times over the same area. In these cases, a larger number of chemical reactions play a significant role, but the chemistry is different from the rural and marine atmosphere due to the presence of high pollution levels in the city and its close surroundings. For instance, in episodes lasting several days, simulating the nocturnal chemistry can be of importance to determine the urban background air content. Chemical mechanisms for application in the various models will be developed under MOD4.
During recent years the importance of particles for e.g. human health effects has been given increasing attention. The processes governing particle concentrations are complex especially in urban areas, where the resuspended material in the streets plays a major role. These processes are not well described in present models and will therefore be given high emphasis under MOD4. This part of the activity is highly complex and will need intensive studies that may run beyond the time schedule of SATURN.
As the development of aerosol models is still in an early stage, concerted efforts are possible within MOD4. As a first step a summary/literature review on the state of art knowledge on aerosols will be carried out serving as the theoretical basis for the development of harmonized European aerosol modules for implementation in models on various scales.
When model validation and sensitivity studies are carried out, it is highly important to be aware what the model results represent. Similarly, it is important for the validation to know what the monitor data represent. As an example, special attention has to be paid when data from a monitor station representing a specific point are compared to results from a grid model. Contributions from the modelling activities will therefore be made to the activities under VAL. These activities will take place late in the process (after year 2000).
The objective of this MG is the development and provision of a framework for validating urban scale models. This will be accomplished by work in four tasks: VAL1, development of an evaluation strategy; VAL2, specifications for data relevant to model evaluation (including data for comparison with model results); VAL3, urban scale emission inventories; and VAL4, model validation exercises.
Development of an evaluation strategy. There is increasing demand for more objective and formalised procedures in order to evaluate the quality (fitness-for-purpose) of models. Currently similar concerns and subsequent activity are evident in many other disciplines.
Evaluation procedures are essentially a consensus of those workers active in a field (including model developers, model users, regulatory bodies etc) as to what is a reasonable and essentially pragmatic strategy for determining model quality and communicating it to interested parties. Interested parties often include less-skilled users who themselves are under increased pressure to provide quality assurance.
The definition of the evaluation concept as applied to urban models is the responsibility of VAL1. This must be related to the nature of the models being considered and many forms of model categorisation are possible, e.g., model purpose, mathematical approach, space and time scales of interest and several others. Substantial cross-referencing of the various forms of categorisation is apparent.
The use of broad, generic categories allows simplified subsequent treatment of these categories such as their general areas of applicability and limitations. One useful categorisation that was considered was into local scale models and urban scale models or (the possibly equivalent) obstacle resolving and obstacle non-resolving models. The development of an evaluation strategy is, however, most obviously linked to the purpose of the model but the above distinctions could be re-interpreted in a "purpose" context. To one of these pairs might be added Air Quality Management Systems (AQMS) as a whole thus providing a direct link between VAL and INT.
A proper evaluation strategy follows a consideration of, for instance, (i) what input variables? (ii) what output variables? (iii) how to compare model and experiment? (iv) what statistical measures to use? (v) what can be concluded from these statistical measures as to the fitness-for-purpose of an individual model?
It was noted that when considered in detail many of these questions are not easily answered. The wide experience of the participants will be essential here.
Although model evaluation involves a scientific assessment, user-oriented assessment, code verification and model validation, it was thought that VAL1 would only consider the last of these which may also be broken down into (i) comparison with data, (ii) code intercomparison, (iii) sensitivity studies and (iv) residuals analysis. During discussion it was pointed out that software exists that allows a degree of code verification and this possibility should not be omitted.
VAL1 will also produce a classification of data sets with respect to their completeness and usefulness for model evaluation purposes (in accordance with specifications developed under VAL2). This may involve data stratification. Data incompleteness will be resolved, where possible, by EXP for field experiments and by VAL for laboratory experiments. A useful starting point for some of these activities is the Inventory of Models and Data Sets produced under COST 615.
The plan for VAL 1 is to develop a draft strategy/protocol, circulate this throughout SATURN for criticism and comment and by an iteration procedure produce a consensus document.
Specifications for data relevant to model evaluation. The data relevant to model evaluation, input and output, are to be considered by VAL2 (with the exception of emission data treated in VAL3). This must address the existence of data sets, quality statements on the data particularly with respect to model evaluation demands and a consideration of what the data can be used for i.e. what are they testing. Additionally VAL2 will consider what is preferred or necessary in future data for model evaluation purposes and what techniques might be applied or developed to satisfy these requirements.
The complementary role of laboratory and field data were discussed with both providing insight; the former control, flexibility and comprehensiveness, and the latter completeness and operational creditability. Also discussed were data uncertainty (the need for error estimates), spatial (and temporal) smoothing of model output, the importance of the gridding technique for CFD models, the determination of data input parameters e.g. zo by remote sensing and the computational generation of simulated meteorological input data.
Urban scale emission inventories. VAL3 is strongly linked to the EUROTRAC-2 subproject GENEMIS-2, in order to ensure compatibility in the methodological approach and the data sets for emission inventories. Specifically the plan is to provide (i) a harmonised method for the generation of urban emission data, (ii) suitable emission factors, (iii) tools for making emission inventories (emission models), and (iv) quality control of resulting emission inventories. The concepts for the first three will be developed on the basis of existing international guidelines and databases. That is, the CORINAIR recommendations and the results of ongoing research projects like MEET will be taken into account. A link will also be established to the Topic Centre on Air Emissions of the EEA.
There was considerable discussion on the provision of information with high temporal and spatial resolution to reflect local scale model demands. The use of emission factors based on back calculation was also addressed.
Model validation exercises. The evaluation strategy developed in VAL1 would be used to combine the models and the data sets, developed under VAL2 and VAL3 through model validation exercises under VAL4. This will involve intercomparison exercises based on models (and model modules) drawing upon current activities at JRC on modelling and statistical analyses of the output. It was thought that intercomparison exercises might be based on one or a selection of urban areas (cities) for which relevant data is available.
A quite distinct validation exercise was also to be undertaken in which comparison with observational data would be made. This would also enable an assessment of whether the overall strategy for model evaluation could be implemented.
In principle, any research group will have the possibility to participate in model intercomparison and model validation activities. Groups outside of SATURN will, however, be requested to provide specified model documentation needed for the quality assurance measures planned in SATURN.
Finally, it was reassuring to note that the many short presentations provided by participants in SATURN while of interest in themselves contributed comprehensively to the activities of VAL.
The specific aim of this MG is to create proper validation data sets from observations and experiments. EXP includes the following tasks: EXP1, local scale experiments (street level); EXP2, urban scale experiments (urban background); EXP3, implementation of complete campaigns; and EXP4, particle studies.
Experimental studies are organised or planned in several countries in the frame of SATURN: Finland, Denmark, Russia, Italy, Sweden, Austria, Switzerland, Greece, Portugal and Hungary. Although the experiments in the cities do not cover the whole set of anthropogenic pollutants and the above scales, a broad set of constituents will be measured at different sites, e.g. NOx, SO2, O3, BTX, VOCs, HONO, nitrate-radicals, PM2.5, PM10 and elemental composition of tropospheric aerosol particles. Advanced field and laboratory techniques are used for the studies, including DOAS, LIDAR, PIXE and XRF.
Future field and laboratory studies will be designed and prepared in a very close co-operation with PIs active in MOD, VAL and INT in order to collect experimental data needed for model development and validation regarding spatial and temporal resolution, meteorological parameters, traffic counting, topography etc. Special attention should be paid to the modelling of particles of different size ranges.
A special application of air quality data is determination of emission factors from the actual car fleet by backward model calculations using traffic counts and local air quality data. This procedure can be applied for all pollutants emitted from vehicles, e.g. NOx, VOCs and particles.
Quality control and quality assurance will be provided and documented through all the experiments. It will include laboratory inter-comparisons, ring tests, parallel measurements with independent methods, parallel sampling at same site and also split sample analyses by different laboratories. Standard or accredited measurement and analytical methods will be used when possible.
At the workshop it was stated that many experimental data sets are already available (e.g. from London, Copenhagen, Oslo). It is expected that not all of them will meet the requirements prescribed; however, each data set will have certain value. Organisation of joint campaigns depends on funding. However, it should be noted that all national funding agencies support national teams for local studies, which makes it difficult to have a real international monitoring campaign funded. Forthcoming national measurement campaigns should as far as possible be adjusted to the applications in SATURN.
An inventory of all experimental data within SATURN and attached experimental studies will be prepared - based on a questionnaire - in a co-operation between the Main Groups.
SATURN should also benefit from other ongoing projects, including the EUROTRAC-2 subprojects AEROSOL and LOOP, the TMR project TRAPOS and the German BERLIOZ campaign. It should also be taken into consideration that only a limited number of model classes can be applied within SATURN. However, the group of experts and the experimental data set to be stored are an enormous potential for any future applications regarding urban air quality studies.
This MG has been set up in response to the situation that efficient air quality management tools are lacking in most cities in Europe. The majority of the population live in cities. For urban authorities to tackle successfully their air pollution problems, local abatement policies are necessary, and for that purpose, efficient air quality management software tools are necessary. Some integrated tools are available, and cooperative work is necessary between scientists and software development groups in Europe to further develop the family of such tools and increase their operationality and applicability for users (e.g. local authorities), based upon a close cooperation with the users. The total work in SATURN should make such tools more acceptable to the user community, since the integrated tools will be based upon modules with harmonized and validated contents, whether it concerns, emission inventories, dispersion models or the utilization of monitoring data. The tasks INT 2-4 are designed to further this work: INT2, integrated monitoring/modelling systems; INT3, comprehensive air quality management systems; and INT4, exposure assessment.
Framework Project. The task INT1, the "Framework Project", has a special role. It should develop the common application-oriented framework for all contributions to SATURN, providing results that are potentially useful for all air quality management systems and integrated assessment methods. The concept of the Framework Project was inspired by the Application Project of the first phase of EUROTRAC. Since the amount of work foreseen goes beyond the limited resources of the Steering Group, it is intended to set up the Framework Project as a financed project. Currently, possibilities for financing it are explored.
In the context of urban air quality, the Framework Project is in the border area between the science and the application in air quality management:
The workshop discussed the draft workplan for the Framework Project, which contained the following items:
The inventorizing of air quality management practices in Europe might be an important activity, with emphasis on the tool/software part of such practices.
The EUROCITIES organisation might be a suitable representative of the user community, and contacts to set up a connection have already been made. In addition, personal contacts between PIs and individual cities are very important in promoting SATURN and its results, especially those cities where experiments and modelling is performed as part of the SATURN programme.
The idea of the Framework Project was unanimously welcomed and several workshop attendees stressed the important role of INT1 in maintaining close contacts between the PIs of the individual contributions. All PIs are strongly recommended to contribute to the INT Main Group as actively as possible. Prior to this, each PI must of course develop his own impression of the potential contribution of his work to actual application in the wider SATURN / Air Quality Management context.
Integrated air quality management systems. The main objective of tasks INT 2-4 is to organise contributions which have as their goal to develop, validate and demonstrate actual software tools for use in the management of urban air quality. There is a degree of overlap between the tasks INT 2-4, and they were treated as a group in the workshop. Under the further development of SATURN, and each contribution, it is expected that they will find their place within the INT 2-4 structure.
There are 9 contributions in this group, and 7 of them were presented at the workshop (contributions from Germany, Greece, Italy, the Netherlands, Norway, Portugal, Spain). The contributions fall in 2 main groups of application:
Some contributions fit in both these groups. The systems are in different phases of development, and some are (almost) existing as operative tools. Several novel concepts were anticipated/reported to be included, such as population exposure modules, cluster analysis for statistical forecasting, land use mapping by satellite, optimization models and traffic prediction models.
This variety of contributions and their contents give reason to be optimistic regarding the possibility to arrive at efficient software tools for air quality management. Key issues for the PIs in the further development, are among others:
Another topic of importance is cost-benefit analysis, as part of a management tool. This requires integration also of such knowledge and models.
It is important to develop links between the projects and PIs in the INT group and the other Main Groups (MOD, VAL, EXP). The first step from the side of the INT group will be to develop overall specifications for the contents of each of the modules which an air quality management tool must include, mainly modules for emissions, dispersion modeling, GIS, monitoring data treatment, presentation of results, and further data assimilation modules, exposure models etc. This structure is intended (i) to give to the SATURN PIs guidance regarding the framework in which their products could play a role, and (ii) to facilitate the exchange of modules between air quality management systems.
The last session of the workshop started with the presentation of the reports of the four rapporteurs. A general discussion followed which turned out to be lively and very fruitful. It mainly aimed at (i) identifying the needs for effective collaboration between the four MGs and (ii) formulating the main conclusions of the workshop. The latter may be summarised as follows:
Model development. The development of models and individual modules should be focused to 2-3 categories which are considered as the most relevant for formulating source-receptor relationships at the urban scale. Close collaboration should be aimed at between the research groups involved in the different modelling tasks.
Model evaluation. For the above 2-3 model categories evaluation procedures ("protocols") will have to be developed and the specifications of data required for the evaluation will have to be worked out. Moreover, rules for model validation exercises will have to be formulated.
Experiments. A questionnaire action will allow characterising available datasets with regard to their usability in model validation exercises. The SATURN Steering Group should interact with scientists planning new experiments (both in SATURN and in the frame of other projects), in order that the resulting experimental datasets are suitable for future model evaluation activities.
Integration. It is intended to develop a harmonised modular structure for air quality management systems and integrated assessment methods. This could promote the integration of scientific results and could largely facilitate the successful addressing of air pollution problems by urban authorities. As the common application-oriented framework for all SATURN contributions, the Framework Project will be most important for providing scientific results that are potentially useful for air quality management systems and integrated assessment methods.
General issues. The success of SATURN will to a large extent depend on the QA/QC measures applied to ensure the validity of the scientific results. Each PI will be responsible to adhere to the foreseen QA/QC strategy. The subproject may substantially benefit from contacts to several other EUROTRAC-2 subprojects as well as to other projects, agencies and relevant bodies.
The workshop closed with the announcement that the 2nd SATURN Workshop will take place in Hamburg on 27 and 28 August 1998.
This article was largely based on the reports which were prepared for the four Main Groups of activities in SATURN. The author wishes to thank sincerely the four rapporteurs, as well as the four Main Group leaders for their comments on the reports. Moreover, the contributions of several task leaders to the reports are gratefully acknowledged.