ECOSIM deliverable

Programme name

Sector

Project acronym

Contract number

Project title

Telematics Application Programme

Environment

ECOSIM

EN1006

Ecological and environmental monitoring and simulation system for management decision support in urban areas

Deliverable number

Deliverable title

Deliverable version number

Work package contributing to deliverable

Nature of the deliverable
(PR/RE/SP/TO/OT)

Dissemination level
(PU/LI/RP)

Type of deliverable (PD/X)

D06.01

Report on validation (Berlin)

1.0

06

RE

PU

PD

Contractual date of delivery

Actual date of delivery

Month 28: April 1998

November 1998

Author(s)

Project co-ordinator

W. Reichenbächer, SSUB, P. Mieth, S. Unger, GMD

Dr Kurt Fedra, ESS GmbH

tel: +43 2253 633 05

fax: +43 2253 633 059

email: kurt@ess.co.at

List of contents

Technical abstract 3
Executive summary 4
1 Introduction 5
2 Validation methodology 6
    2.1 Functional testing and impact analysis of applications 6
    2.2 Reference cases 8
    2.3 Criteria for success 8
    2.4 Demonstration stage 9
3 Case study Berlin 10
    3.1 he demonstrator domain Berlin/Brandenburg 10
    3.2 Berlin test user profile: the Ministry of Urban Development,
    Environmental Protection and Technology of Berlin 10
    3.3 Air quality and emission conditions in Berlin 12
    3.4 Demonstrator installation and interfaces 14
    3.5 The base case 15
    3.6 Emission reduction scenarios 16
    3.7 Analysis and results of the automatic daily ozone forecasting in
    Berlin/Brandenburg 20
4 Test user conclusions and summary remarks 25

Appendix A: Figures of time series comparison

Appendix B: Berlin questionnaire

Technical abstract

Deliverable number

Title of deliverable

D06.01

Report on validation (Berlin)

ECOSIM is a support system for the investigation and forecasting of pollution levels in urban areas. Allowing the effects of industrial developments or new roads to be quickly and easily examined, ECOSIM can aid public authorities to ensure that their urban plans fully consider environmental impacts. ECOSIM is also able to forecast air pollution levels - typically for the following 24 hours.

The project develops and demonstrates an integrated environmental monitoring and modeling system for management decision support in environmental planning for urban areas. Traffic generated air pollution including photochemical smog, coastal water quality, and groundwater are the initial application domains.

ECOSIM is based on a set of computer models ranging from very simple screening tools to sophisticated 3D dynamic models such as those which allow ozone levels to be calculated from road traffic emissions. It combines these models with up-to-date measurements of current meteorological conditions and pollution by linking directly to local databases and pollution measurement stations. ECOSIM also links the various domains such as surface water, coastal water and air to ensure that as complete a picture as possible of environmental conditions can be predicted .

ECOSIM makes use of very high performance computers whenever it needs to and uses the latest methods in handling maps and similar data to ensure that its results can be easily translated into practical measures.

Executive summary

Deliverable number

Title of deliverable

D06.01

Report on validation (Berlin)

This validation report presents the ECOSIM applications which were executed at the validation site of Berlin by the users and the user support partner. The applications focuses primarily on the use of historical and real-time meteorological and pollution data and environmental models which monitor and forecast the impact of human activities (e.g. traffic and industry emissions) on the environment.

An ECOSIM demonstrator with the user required functionality has been installed at the SSUB office in Berlin. The demonstrator was coupled with the local concentration data base of the BLUME net, and interfaces to analyze the observations were established. The ozone analyzing and forecasting capabilities of the ECOSIM demonstrator have been proven for 4 selected episodes suffering from high surface ozone levels.

The validation report identifies the impacts of ECOSIM applications for the demonstrator Berlin/Brandenburg. It presents the criteria which have been used to select the impacts for validation, it provides a brief description of the methods used to measure them and it includes the questionnaire for evaluation at the demonstration site Berlin.

1 Introduction

ECOSIM is a support system to investigate and forecast pollution levels in urban areas. By allowing the effects of industrial developments or new roads to be quickly and easily examined, public authorities can use ECOSIM to ensure that their urban plans fully consider environmental impacts. ECOSIM is also able to forecast air pollution levels - typically for the following 24 hours.

The project develops and demonstrates an integrated environmental monitoring and modeling system for management decision support in environmental planning for urban areas. Traffic generated air pollution including photochemical smog, coastal water quality, and groundwater are the initial application domains.

ECOSIM is based on a set of computer models ranging from very simple screening tools to sophisticated 3D dynamic models such as those which allow ozone levels to be calculated from road traffic emissions. It combines these models with up-to-date measurements of current meteorological conditions and pollution by linking directly to local databases and pollution measurement stations. ECOSIM also links the various domains such as surface water, coastal water and air to ensure that as complete a picture as possible of environmental conditions can be analyzed and predicted.

ECOSIM makes use of very high performance computers whenever necessary and uses the latest methods in handling maps and similar data to ensure that its results can easily be translated into practical measures for pollution control.

ECOSIM involves participants from Austria, Germany, Greece, Italy, Poland, and the UK, and works with the cities of Berlin, Athens, and Gdansk as validation sites and initial end users.

After having undergone final refinements based on the insights gained in the validation phase, the ECOSIM system can become an important tool relevant to various local and regional authorities thus generating significant European Added Value.

2 Validation methodology

The objective of the validation phase is to validate the operation of the ECOSIM demonstrator within the context of a variety of environmental domains.

The described validation activity comprised two elements: verification and demonstration, and took place at the demonstration site Berlin. The verification process sought to confirm that the demonstrator implements the specified user requirements. The demonstration activity allowed urban authorities to use the Berlin demonstrator for their specific task. This facilitated feedback on the validity of the agreed requirements, their method of implementation, new requirements and the general acceptability of the system for operational use. It proves the technical reliability of the system, the geographic information system, data managing and analysis functions and model interfaces. It also provided opportunities for publicizing the demonstrator.

In cooperation with the local users, the detailed criteria for success of the verification activity have been defined. These include comparisons of the ECOSIM results with historical and observation data for selected episodes and plausibility considerations analyzing emission reduction scenarios. Additionally, a questionnaire has been produced to give a more precise picture of the relation between user requirements and the demonstrator functionality for designated tasks.

The monitoring network for air pollution in Berlin and the meteorological measurement data network have been partly connected with the ECOSIM server. The general functionality of the ECOSIM server was tested and a user training was performed.

The scenario analysis consists of in comparison of a pre-defined standard case with an emission-reduction scenario using REGOZON model system. The standard case serves as a reference situation, defining the main characteristics of a typical mid-summer day in the region Berlin-Brandenburg. To this aim, weather conditions were selected which normally lead to high ozone concentrations and which have a high ozone-formation potential (moderate wind velocity, high temperatures and insolation). Because traffic is the main cause for high near-surface ozone concentrations in the Berlin-Brandenburg region, the emission-reduction scenario considered traffic controlling measures which influence the amount and the composition of traffic-emitted ozone precursor substances. The time-period for simulations will be 24 hours. The necessary input data were pre-processed in collaboration with the end-user SSUB. SSUB was also consulted in reference to the concrete measures, their effects on emission behaviour and the choice of an appropriate episode. The verification of the simulation results was performed using the measurement data of the monitoring network connected with the ECOSIM server. Based on the results of the scenario analysis, the influence of traffic controlling measures on the regional ozone concentration has been evaluated and decision alternatives can be derived.

    2.1 Functional testing and impact analysis of applications

Functional testing involves determining that the technical functionality (functional design) of ECOSIM is correct and that the output from the modules is in agreement - within defined limits - with experimental and/or field data. Since the ECOSIM project did not focus on model development itself, only limited resources (amount of data, computing resources) were available for this task (and the results must be interpreted with these limitations in mind).

The functional design ensures that the software requirements have been understood and can be implemented. The functional design specifies the software at the functional level, i.e. what the software will do, and is not concerned with implementation details. It includes functional tests whereby the software can be tested at the functional level.

Functional testing of applications was performed to measure the quality of the results derived from ECOSIM and was the predominant activity of the verification stage. Functional testing primarily involved measurement of ozone levels and planning decisions.

Within the context of this project user acceptance is considered to be a reflection of the extent to which ECOSIM fulfills direct user requirements. ECOSIM's appeal is largely a product of 'how well the system works' which is dependent on a set of criteria common to almost all the applications, defined as:

    • presenting the information required by users in order to fulfil their day-to-day tasks (e.g., relating to environmental planning issues);
    • presenting information in a format that is convenient to users (i.e., at the correct time and frequency and superimposed on other data such road maps;
    • the 'look and feel' of the interface (e.g., is it simple to use, is the GUI acceptable, etc.);
    • robustness;
    • flexibility.

Acceptance tests verified that the software was properly installed and operating correctly. An example of an acceptance test was the comparison of ozone level predictions from ECOSIM with actual measurements taken from "ground truth" sites. (Due to the complexity of the involved phenomena in some cases these comparisons may only be possible in a qualitative manner).

Information was gathered during the verification phase, by surveying the opinions of a representative sample of users. Acceptance testing did not include the opinions of individuals or groups indirectly affected by ECOSIM. Unfulfilled requirements had been identified and, if necessary, a few modifications were made to ECOSIM before the demonstration stage; remaining user requirements and problems should be solved if a commercial product is built based on the ECOSIM demonstrator.

Impact analysis was conducted on the validated impacts selected after consultation with ECOSIM partners.

General measurement techniques comprised 'hard' and 'soft' methods of assessment for measuring the selected impacts for each application. Hard methods comprised comparison of ECOSIM modeling and forecast results with the results of independently validated data to determine the quality (i.e., timeliness, accuracy, regularity, etc.) of the results generated by ECOSIM. Soft methods comprised qualitative or subjective information gathering methods such as interviews and questionnaires although this information may, in some cases, be expressed quantitatively.

    2.2 Reference cases

Reference cases are made to determine the relative worth of ECOSIM applications with respect to those that currently exist (or to establish their worth in cases where no other applications exist at present). To facilitate comparisons between applications, reference cases will be structured in the same way as far as possible. Their structure will comprise:

    • comparison of application quality i.e., timeliness, accuracy, level of detail, regularity etc.;
    • analysis of the implications of differences in application quality;
    • comparison of the cost of existing and ECOSIM applications;
    • analysis of obstacles to implementing ECOSIM applications;
    • conclusions about the value of ECOSIM applications.

Reference case studies took place during the demonstration stage when ECOSIM applications was considered to be at a pre-operational stage of development. Reference case studies were to some extent constrained by the limited availability of data and by the lack of appropriate networks.

    2.3 Criteria for success

The overall objectives of the ECOSIM project were to implement an integrated environmental management decision support system and in particular, to:

    • demonstrate that greater insight has been obtained by the validation users into their own particular environmental issues;
    • verify that the system suits the user requirements;
    • obtain a system which is capable of beeing up-grade (perhaps within 2 years) to a commercial product or set of products.

These objectives comprise the general criteria which will be used to determine the success of the project as a whole. The criteria for success used during the verification stage of the validation process was focused on the performance of ECOSIM and user acceptance. They included:

    • efficiency improvements over existing applications;
    • reliability;
    • ease of use of ECOSIM;
    • information quality improvements over existing applications;
    • the flexibility vis-a-vis modelling of information from different environmental domains.

The success criteria which have been used during the demonstration phase relate to the measurable, direct impacts of ECOSIM applications on end-users and indirect beneficiaries. These criteria were application-specific. Thus, in Berlin ECOSIM should allow the authorities to forecast the levels of ozone and classical air pollutants on a day-to-day basis mainly arising from traffic emissions and use similar results to inform urban planning decisions (e.g. new roads) and traffic control measures to reduce regional ozone concentrations (e.g. limiting inner city traffic, forcing use of catalytic converters).

More "global" criteria such as willingness of users to use ECOSIM (perhaps compared with existing applications) to fulfill their day to day objectives also offered a meaningful way of measuring the success of the demonstration stage of ECOSIM.

    2.4 Demonstration stage

The assessment objectives of the demonstration stage were to determine the value of the ECOSIM application impacts selected for validation. Impact analysis involved the use of qualitative and quantitative methods to estimate the economic and social benefits of ECOSIM applications and tried to compare them to existing information systems. To this end, a questionnaire was developed, distributed and filled out by the users.

3 Case study Berlin
    3.1 The Demonstrator domain Berlin/Brandenburg

The main focus of the environmental authorities in Berlin /Brandenburg is on short term daily ozone forecast in conjunction with the possibility to estimate effects of emission reduction strategies as well as the use of the ECOSIM system for strategic considerations and observational data analysis. The water quality management is not a main issue for the authorities involved in the Berlin case study and was therefore not considered.

The model domain for the demonstrator B is the conurban area of Berlin. This model area of 100 km x 100 km extension is flat and shows a relatively weak orography. The selected size of the computational grid was 2 km x 2 km. The area is well suited for investigating effects from anthropogenic emissions of an urban area because the relatively densely populated city of Berlin lies in a sparsely populated domain.

    3.2 Berlin test user profile: the Ministry of Urban Development, Environmental Protection and Technology of Berlin

The Department VI D of the Ministry of Urban Development, Environmental Protection and Technology of Berlin is responsible for the main planning issues of air pollution control in the city of Berlin.

In Berlin, a monitoring network for air quality has been run by SSUB for more than 20 years (BLUME, see Figure 1). There are currently 45 measuring stations for different air pollutants and two additional stations for meteorological measurements. Most of the stations are arranged on a grid of approximately 4km x 4km. Measurements are made of levels of dust, benzene, VOC, SO2, NOx, CO and O3. In addition to its position, each measurement station has certain key features recorded within a database: its general location (e.g. inner city/suburban), type of district (e.g. residential, industrial), traffic levels (in bands of vehicles per day), type of private heating (in terms of level of SO2 emissions). BLUME measures, stores and analyzes the most important air pollutants continuously in a spatially fine resolution. The public is being kept informed about the statistical analyses of these measurements in regular annual reports and up-to-date data via the Internet. The data from the measuring network has for many years formed the basis for management measures of the authorities to improve the air quality. The concentrations due to the classical air pollutant sulfur dioxide has been improved considerably because of the collapse of many industrial companies of the former GDR. At present, the main air quality problems in the region of Berlin are:

    • the summer ozone load at the outskirts of the city
    • the nitrogen dioxide load in the city center
    • the soot and benzene load in the street canyons of the city center
    • the fine dust load (PM10) in the city and in the closer surroundings.

The two first problem areas already existed 3 years earlier, when the project ECOSIM was planned, while the last two problem areas have been considered more intensively in the recent years because the critical limits in Germany and Europe decreased. To analyze the ozone formation and nitrogen dioxide loads in Berlin with appropriate simulation tools, an emission inventory of all pollutants, in particular organic compounds, relevant for the ozone formation, nitrogen dioxide, carbon monoxide and sulfur dioxide were created in 1994. A 3-layer model (REGOZON) for the simulation of the ozone formation in the lower atmosphere, adapted for the region of Berlin, can be driven by these input data.

Figure 1: The Air quality measuring net in Berlin (BLUME)

Extensive model tests were carried out for an ozone episode in July 1994. The chosen period was part of a measuring campaign of the Senate of Berlin and the Ministry of Environmental Protection of the county Brandenburg (FLUMOB). Two aircrafts and several mobile measuring stations measured ozone and its precursor substances. Due to a longer period of fair weather with high irradiation and temperatures, the mean ozone level was already relatively high for this region. The reliability of the ozone simulation results was proven by a large comparison of simulated and observed data.

The developed software tools within the ECOSIM project (ECOSIM demonstrator) should allow to support all work procedures of the Berlin authorities, necessary for analyzing the reasons for high loads of ozone and other air pollutants. In particular:

1. Temporal and spatial representation of all measured values of BLUME network data, in particular ozone and nitrogen dioxide.

2. Spatial representation of the emissions of ozone precursor substances for the individual groups of emission sources

3. Execution of ozone simulation calculations with different emission and weather scenarios.

4. Presentation of the results of the simulation calculations, in particular for ozone and the precursor substances.

5. Supply of current ozone prognoses via Internet.

    3.3 Air quality and emission conditions in Berlin

Berlin has been confronted with a considerable increase in motor traffic since the reunification of Germany. The number of motor vehicles registered in Berlin has risen by about 30% since 1989 to a present total of 1,280,000. A further growth in motor vehicle traffic is projected for the future, especially for heavily polluting freight transport. These changes are not yet concluded. Traffic increases result from the expansion of the Berlin/Brandenburg residential and economic area, from the rapid growth of international economic relations and particularly from the strengthening of links between Berlin and Eastern Europe.

Motor vehicle traffic has become the greatest source of air pollution in Berlin. The most significant pollutants emitted by motor vehicles in terms of quantity are carbon monoxide, hydrocarbons, nitrogen oxide and carbon dioxide. The pollutant quantities of diesel particulates, tire abrasion and benzene are much smaller, but they are important because of their effects on human health. Vehicle pollution is especially high in the inner city, where over 1 million people inhabit an area of 100 km2. The future functions of the inner city will clearly increase traffic and air pollution in this area.

The main ozone precursors in the urban scale are nitrogen oxides and hydrocarbons. Nitrogen oxide emissions are caused primarily by combustion processes in power plants and heavy industry firing plants as well as in motors. Nitrogen oxides develop from the nitrogen and the oxygen of the combustion air and to a lesser extent also through the oxidation of nitrogen-containing elements of the fuel. The formation of nitrogen oxides greatly increases with the combustion temperature. The primary polluters in the city of Berlin are motor vehicle traffic with 19,000 t at approximately 50% and the power plants as well as other licensed facilities with 16,000 t, or a share of more than 40%. Nitrogen monoxide comprises more than 90% of nitrogen oxide emissions. Above all it is the reaction with ozone in the atmosphere which forms the nitrogen dioxide so much more damaging to human health. Since these reactions use up ozone, conurbations display, on average, a slighter higher ozone concentration in the near-ground air than in rural regions. Beside ozone, combinations with hydrocarbon molecules also contribute to the transformation of nitrogen monoxide into nitrogen dioxide. Through further oxidation and combination with hydrocarbon molecules acidic aerosols develop from nitrogen dioxide, which deposit on surfaces and/or precipitate from the atmosphere as acid rain. Under the influence of intense sunlight reactions occur in the atmosphere simultaneously, by which nitrogen dioxide decays again into nitrogen monoxide and oxygen atoms. In this process, in which likewise water - and hydrocarbon molecules are involved, more near-ground ozone is produced than consumed. The ozone concentration in the air can rise thereby greatly.

Hydrocarbon (HC) emissions, along with nitrogen oxide, play a significant role as ozone precursors. Other hydrocarbons, e.g. benzene, require particular attention because of their carcinogenic effects.

Hydrocarbons are released through the exhaust when fuel is unburned or incompletely burned. Considerable amounts also reach the atmosphere due to fuel evaporation. Hydrocarbons evaporate from the fuel tank and other fuel feed elements, such as the fuel line, carburetor, filter, reserve canister, etc. Hydrocarbons also vaporize when fuel station storage depots and motor vehicle tanks are filled.

As we have effective laws and guidelines in Europe and Germany, the individual emissions of each typical passenger vehicle will decrease during the next years. But also in future, the traffic will increase and some pollutant emissions will not show significant reductions. This is particulary so for nitrogen oxides and (diesel) particulates. (Diesel) particulates have the greatest current need for action. Hydrocarbon emissions, however, are expected to decline significantly even without more regulatory intervention because 1) more vehicles are being equipped with 3-way catalytic converters and 2) fuel quality is improving. Technical improvements for passenger cars have had significant effects. The situation for diesel particulates is different. Trucks and buses make up 5% of total travelled distances and are responsible for 10% of total emissions - but they are responsible for 90% of carcinogenic diesel particulate emissions. The main source of ozone precursor emissions in this region is the traffic responsible for at least 60% of the total emission (Table 1).

Table 1: Emission distribution in the model domain Berlin/Brandenburg

emission class

NOx[kt/a]

VOC [kt/a]

CO [kt/a]

traffic (excluding cct)

68.9

40.1

164.0

city centre traffic (cct)

3.3

1.9

8.6

traffic VOC evaporation

-

30.6

-

industry with an emission height <30 m

1.5

40.1

1.0

industry with an emission height 30 m

27.6

1.6

22.3

households

0.5

13.9

4.2

total

101.2

128.3

199.1

The spatial emission distribution of the ozone precursor substances, mainly nitrogen oxides (NO x ) and volatile organic compounds (VOC), reflects the population distribution, showing high emissions in Berlin and low emissions in the surroundings (Figure 2).

Figure 2: Spatial NOx [t * a-1] emission distribution in the model domain
    3.4 Demonstrator installation and interfaces

The demonstrator uses automatic interfaces to meteorological and air quality observation in the region of Berlin/Brandenburg. It includes initial data for ozone and nitrogen oxides concentration, initial meteorological data and prognostic synoptic data for the large scale meteorology. It employs the same interfaces designed and used for an automatic daily ozone forecast. For the Berlin case study, the ozone model REGOZON has been used. The time variation of the background concentrations of ozone and its precursors was computed with a kind of chemical box model simulating the conditions in a rural area with no horizontal resolution but having the same vertical structure as all other model components. The box model obtains all necessary meteorological quantities from the mesoscale meteorological model and uses their averaged values as well as typical emission data of the surroundings of the model domain.

The demonstrator runs on a SUN U2/2170Cr2 2 x UltraSPARC-I. The necessary data, the initial meteorological data sets, as well as the air quality data from the BLUME net are transferred once a day during the night via automatic ftp and stored locally in a data base. This allows calculation and analyzing of previous episodes as well as a real prognostic 24 hour ozone forecast. A model run for a 24 hour forecast consumes about 2 hours of computation time. The calculation is executed automatically as soon as the complete data set arrives.

The local installation for the user SSUB consists of a SUN 5/110TGX microSPARC-II 110MHz. This machine was provided by the GMD to enable the potential end users to test the system. The end user SSUB has no appropriate data connection to run the simulation models by anywhere else and transmit the data. That is why everything was installed locally, the GUI, the BLUME data server as well as the simulation models. Although the model execution time is relatively high on this machine, the principle functionality of the demonstrator was enabled. A model run for a 24 hour forecast consumes about 5 hours of computation time on this SUN workstation.

    3.5 The Base case

For the demonstrator application in the Berlin domain, an episode of 3 days of high ozone levels from 1996 and one day in 1997 has been chosen (6.6., 7.6., 8.6.1996 and 12.6.1997). The period in 1996 is quite difficult to model from the meteorological point of view showing particularly strong changes of the wind direction in the model domain, especially on June 7. On this day, the wind direction changed to the opposite direction in a short period of time which is a challenge for any dispersion model. The measured ozone maxima were in the range of 280 µg/m³.

The computation of the ozone distribution often shows an ozone plume downwind of Berlin as long as the wind direction does not change significantly during the model run, and the wind speed is at least moderate and the meteorological conditions promote a ozone production. An example of the ozone distribution is shown in Figure 3 for June 12 in 1997. It is already clear that the main source for the extraordinary ozone production is strongly related to emissions in Berlin.

Figure 3: Surface ozone concentration on June 12, 1997 in the model domain Berlin/Brandenburg

One of the principle problems of a model evaluation procedure is the comparison of the simulated data with the observations. The mesoscale models produce gridded values with a horizontal resolution of few kilometers and a time scale of several minutes, whereas the observations are point measurements reflecting the local conditions. Especially observations in a city are often influenced by fresh NO emissions and the very local conditions at the location of the measuring device. Therefore, it cannot be expected to achieve very close agreement between point observations and simulations. Nevertheless, due to a lack of other verification strategies, a time series comparison for 10 operational ozone measuring stations belonging to the Berlin BLUME measuring net has been documented (see Appendix A, Figure 1-8).

All days show for most of the locations a rather close agreement between simulations and observations. In general, simulations often predict a higher diurnal variation in comparison with measurements. The reasons can be found in the

    • parametrization of the mixing height and stability quantities, causing a slightly late growing of the mixing height in the morning and a rapid falling down at the evening,
    • calculation of the dry deposition for ozone.

Nevertheless, the main intention of the model development was the representation of the concentrations during the time when the ozone maximum normally occurs, normally in the afternoon. This is especially important for any practical air pollution management. During this period of time, the model performs well. Taking into account the input data uncertainties, especially because of the rather unpredictable nature of the precise traffic emissions, the results are mostly satisfactory.

The analysis of the time series comparison sometimes exhibits a poor agreement on some stations although most of the locations perform well. For example, on June 7, 1996, the simulation underpredicts the ozone values at station #10 in the afternoon (Appendix Figure 3). On the other hand, the afternoon measurements at this station show a considerably higher surface ozone concentration than any other station of the BLUME net, at least 30% higher. This concentration peak seems to be a very local effect and therefore probably not resolvable with the model having a horizontal resolution of 2kmx2km.

    3.6 Emission reduction scenarios

The calculations described above serve as reference situations for scenario analysis. For this purpose, additional model runs were executed using the same set of data for the background, the initial meteorological conditions and the synoptic conditions used for the base case but scaling the emission classes. The comparison with the reference case helps to estimate future trends under changing emission conditions or to determine the effect of an emission reduction measure in case of a critical air quality situation.

Two scenarios have been selected for the episodes described in Section 3.5. The assumed emission conditions for scenario 1 represent the expected emissions in the year 2005 without additional regulatory efforts. The percent of emission reduction in comparison with the recent emission strength is given in Table 2.

Table 2: Scenario 1: expected emission conditions for year 2005 without regulatory efforts

emission class

NOx reduction [%]

NOx reduction [kt]

VOC reduction [%]

VOC reduction [kt]

traffic

30

20.7

50

20.1

city centre traffic

45

1.5

58

1.1

traffic VOC evaporation

-

20

6.1

industry with an emission height <30m

30

0.5

30

8.0

industry with an emission height 30m

24

6.6

10

0.5

households

10

0.05

10

1.4

sum

29

29.4

29

37.2

Scenario 2 is based on the same assumptions but takes into account regulatory efforts to reduce the city centre traffic. The expected emission conditions for this scenario are given in Table 3.

Table 3: Scenario 2: expected emission conditions for year 2005 with regulatory efforts

emission class

NOx reduction [%]

NOx reduction [kt]

VOC reduction [%]

VOC reduction [kt]

traffic

30

20.7

50

20.1

city centre traffic

70

2.3

80

1.5

traffic VOC evaporation

-

-

20

6.1

industry with an emission height <30m

30

0.5

30

8.0

industry with an emission height30m

24

6.6

10

0.5

households

10

0.05

10

1.4

sum

30

30.2

29

37.6

There are different possibilities to estimate the effect of an emission reduction measure. It can be based on the influence on measuring sides or on the maximum ozone values. In the field of air quality management, the effect of an emission reduction of ozone precursors on the ozone concentration is often considered in relation to critical limits because these limits are the basis for the authorities to act. The counting of grid hours with ozone concentrations above a limit (in this context a level of 180 µg/m³ was considered to be "critical") gives a visual measure for the area which suffers from high ozone levels during a certain time. The normalized number of grid hours above the critical limit in comparison with the base case is given in Table 4.

A possibly more relevant criterion in order to characterize the reduction of the total amount of ozone above critical ozone levels is the normalized excess ozone (NEO) as an indicator for the efficiency of a measure during the whole episode and the total area. As a normalization factor serves the excess ozone of the reference state. For a successful reduction of high ozone levels the indicator must be below one. A smaller indicator stands for a more effective measure in order to reduce critical ozone levels.

Both indicators show more or less the same tendencies for the scenario calculations on different days although the reduction of grid hours above the critical limit is more effective for the selected episodes in comparison with a reduction of the excess ozone. The effect of an emission reduction measure can be quite different for different meteorological conditions and/or concentration background. For example, the reduction is very effective for June 12, 1997, whereas the reduction is not significant on June 8, 1996. Different meteorological conditions lead to different ozone production efficiencies. This can be explained by different mixing heights, different temperatures and solar radiation and/or different levels of background ozone and background concentration of ozone precursors. Both indicators, normalized grid hours and excess ozone, suffer from the fact that they rely on the selected critical value but give a good qualitative measure for decision makers because they have to take into consideration the air quality with respect to the official critical limits.

Due to the complexity of the processes and the nonlinear nature of the photochemistry, it is not possible to quantify the emission reduction potential of the described scenarios in general. When analyzing excess ozone, changes of ozone flow and grid elements with high ozone levels, one must keep in mind that the reduction of high ozone concentrations can differ dramatically from day to day for the same measure. On the other hand, an additionally reduction of the traffic emissions in the centre of Berlin does not lead to remarkable ozone reduction for all considered episodes. The additional emission reduction in comparison with emission reduction scenario 1 is less than one percent of the total NOx emission (see Table 2 and Table 3). This reduction is too small to produce any noticeable ozone decrease (it is far below the uncertainty range of the emission inventory), and the possible increase of the ozone level in the city centre might compensate the slight ozone decrease anywhere else. But, an emission reduction of precursors in the city centre of Berlin often leads to an increase of ozone in the city but, nevertheless, as the consequence of a reduction of precursors to an improvement of the integral air quality, which is determined by concentrations of ozone, its precursors and other primary air pollutants.

Table 4: Normalized reduction of grid hours with concentration above 180 µg/m³ and normalized excess ozone with a assumed critical value of 180 µg/m³ for the emission reduction scenarios

date

scenario 1

scenario 2

normalized reduction of grid hours with concentration above 180 µg/m³

normalized excess ozone

above 180 µg/m³

normalized reduction of grid hours with concentration above 180 µg/m³

normalized excess ozone

above 180 µg/m³

6.6.1996

0.99

0.88

0.99

0.89

7.6.1996

0.90

0.78

0.90

0.78

8.6.1996

1.02

0.99

1.02

0.99

12.6.1997

0.38

0.28

0.38

0.28

Also, an isolated emission reduction measure in the model domain only cannot reduce the ozone concentration by a significant amount even with a moderate decrease of ozone precursor emissions. The ozone concentration in this model domain is to a high degree determined by the mean ozone concentration. Only the peak ozone values are formed in down-wind regions of Berlin due to the high precursor emissions in the city. Efficient ozone reduction strategies have to base on a larger scale and more long-term oriented ozone precursor emission reduction.

    3.7 Analysis and results of the automatic daily ozone forecasting in
    Berlin/Brandenburg

The second main focus of the ECOSIM demonstrator in Berlin is the testing of an operational ozone forecasting system for this region. A first implementation and test runs for summer 1997 gave a relatively high root mean square error (RMSE) of about 35 µg/m³ for the maximum predicted ozone concentration during the day. This value was even higher than the results of a persistance approach, assuming the same ozone value as the day before (Figure 4). The main reason was the use of an inappropriate or incomplete input data set.

Figure 4: Forecasted maximum ozone concentrations vs. measured maximum ozone concentrations in summer 1997

The new version of the REGOZON-based prototype of an automatic ozone forecasting systems was tested from May to September 1998. The input facilities were completely redesigned. The system used interfaces to the air pollution measuring net BLUME to extract the initial concentrations and to synoptic initial and prognostic meteorological data provided by the Institute of Meteorology of the Free University of Berlin. The ozone simulation was automatically started during the night as soon as the initial data arrived. After finishing the model run, ozone maps showing the near surface ozone concentration and animations were produced and published in the world wide web (http://www.first.gmd.de/ozon/) for public access. Critical loads in the city of Berlin and in the surroundings were determined. Additionally, data are immediately sent to the Senate and to a local radio station.

The first analysis shows a good technical reliability of the system. In less than 5% of all cases the system did not receive actual air quality data because of data transfer problems and was therefore executed with mean background concentrations. The graphic representation of the results of the net was always ensured.

Initial model evaluation was carried out by the Senate of Berlin. According to the regulations, the authorities have to act if a certain critical load has been reached. That is why their evaluation criterion was the comparison of the maximum forecasted ozone concentration with the measured one. Although this might be a questionable evaluation criterion in general, it is the most important quantity for the policy makers, if they are interested in using the system to forecast the surface near ozone for the following day.

The analysis of 130 days of summer 1998 (Figure 5) shows the following results:

    • The bias for the simulation was very low, below 1 µg/m³. There is neither a tendency to over- nor to underpredict the maximum ozone.
    • The root mean square error (RMSE), an appropriate measure for the expected averaged error, is about 25 µg/m³. This still seems to be quite high. But taking into account the input data uncertainties and the often very local occurrence of high ozone levels which cannot be resolved in a mesoscale model the result is acceptable.
    • The RMSE of a simple statistical method operationally used by the Senate is slightly lower (23 µg/m³), but exhibits a considerable bias with a tendency to overpredict the maximum ozone level. Additionally, the statistical method is based on data observed at the late morning of the same day and gives therefore results at a time much too late to react if the predicted level exceeds the critical limit. A statistical method does not provide any guidance for ozone management, which is a main feature of the model based forecast approach.
    • The persistence approach (assuming the same maximum value as reached at the day before) gives worse results in comparison with both forecast methods.

Figure 5: Forecasted maximum ozone concentrations vs. measured maximum ozone concentrations in summer 1998

A few days showed a substancial difference between predicted and observed maximum ozone levels. This was mainly caused by wrong or inconsistent input data. As a consequence, all input data used have to be automatically checked before starting the simulation, and improper values should be refused by the program. Also, tendencies of ozone underprediction can be compensated by using the maximum ozone of the previous day of measuring station at the Frohnau tower. This leads to a considerably lower RMSE (23 µg/m³) (Figure 6).

Figure 6: Forecasted maximum ozone concentrations vs. measured maximum ozone concentrations in summer 1998, minimum correction performed
4 Test user conclusions and summary remarks

The tested ECOSIM prototype includes most of the desired features for data analysis and modeling. The graphic representation of spatial distribution of the emission, substance and class specific, has been fully realized. A graphic information system (GIS) allows the production of overlapping maps using pixel or vector oriented data sets. Numerous data analysis functions for monitoring data from the BLUME measuring net were implemented.

The ECOSIM software for the analysis of measurement results has mainly been used for specific tests by the staff of the Department VI D of the Ministry of Urban Development, Environmental Protection and Technology of Berlin, because a software tool for the analysis of the results of measurement is already in operational use in the department in Berlin. One important advantage in comparison with the available software is the possibility of representing the spatial distribution of the results of measurement with isolines or as three-dimensional isosurfaces. It is clear that in the present installation the principle functionality of the systems has been realized rather than the whole range of possibilities. However, in a commercial product derived from the ECOSIM demonstrator, data analysis and modeling functions should allow an access to an enhanced set of substances presently restricted to ozone and nitrogen dioxide.

The ECOSIM demonstrator is not very easy to use for an untrained computer user, even if he has some experience with standard programs under the operating system Microsoft Windows. But a user accustomed to working with standard programs is able to operate this software system after a few hours instruction. In the current stage, the present installation does not always prevent the user from producing a system crash or simply running into a dead end. After a briefing it is however possible to execute reality near simulation calculations for ozone formation in Berlin with this program. The emissions of precursor substances can be changed in a very simple way for the individual groups of emission sources. The simulation calculation can be executed with 4 different preselected real meteorological data records. These meteorological data sets characterize weather conditions with circulating weak winds and considerable high ozone production capabilities. The simulation results show realistic ozone formation in the model domain, comparable to the measured extra ozone production due to emissions in the Berlin area on hot summer days. This component of the ECOSIM demonstrator is already in the present form a valuable support tool for the work of the Berlin authorities in the field of strategic air quality planning and management. An elimination of minor errors in the system and an extension of the functionality would be however desirable for a derived commercial product.

In a future version an interface to the automatically running daily ozone forecast as well as to an extended meteorological data base should be implemented. It should be easier to compare measurements and results of computation in a tabular form, as time series and in its spatial distribution, e.g. with isolines.

Another emphasis of the ECOSIM project in Berlin was the simulation of the ozone formation and the use of the system for routine ozone forecast.

This achieved forecast quality is better than the simple persistence forecast, but the objective to meet a mean square error of 20 µg/m³ was still not achieved. The model reliability and the model results have been considerably improved for the summer 1998 in comparison with the initial tests executed in summer 1997. It should be borne in mind, that the input data inaccuracies are still quite high. Also, the sense of using the maximum ozone concentration as a model evaluation criterion is very limited. Nevertheless, the routine ozone forecast with the REGOZON model is an important aid for the evaluation of the future ozone load during summer high pressure weather conditions. In particular this applies, because it is at present the only objectively determined ozone forecast for the Berlin area, which is already present in the evening of the day before. This enables the responsible persons for proclaiming a summer smog alarm to check the prerequisites for a possible driving ban on the following day according to the federal concentration control law before the evening news. Another objective ozone forecast procedure, which is based on statistical analyses of ozone measurements and operationally applied in our administration, leads for the summer 1998 to slightly better results than the REGOZON model (the mean square error amounted to 23 µg/m³), but this procedure relies on data which has to be measured in the morning of the day, to which the forecast was to apply. However, a driving ban because of summer smog at this time could have only small effects for the reduction of the ozone precursor substances, because the main precursor emission source, the traffic during the daily rush hour, reaches already its maximum.