• To estimate tropospheric concentration fields at the ground level of air pollutants and greenhouse gases for the participating cities and the whole Europe.
• To develop and validate different data fusion, pattern recognition and image processing methods for extracting optimal pollutants concentration fields from multi-source data.
• To develop an approach for directly linking changes in emissions of local sources to concentrations using source apportionment.
Task 3.1 Atmospheric modelling and data fusion
This task aims at deriving tropospheric concentrations of air pollutants and greenhouse gases for the whole of Europe and for the participating cities. Anthropogenic emissions will be translated into urban and regional air quality estimates by coupling different modelling approaches, including (a) selected meteorological and Eulerian atmospheric transport models, (b) artificial intelligence algorithms for urban scale pollution estimation and (c) data fusion algorithms for satellite, airborne and personal sensing data. The above very different information sources will be integrated optimally by applying different data fusion and data assimilation schemes so as minimize the residual error in the pollutant concentration values. The ICARUS information fusion framework will be used to integrate the various state-of-the-art air quality sensing techniques and atmospheric pollution transport models towards a novel paradigm for environmental monitoring. According to this paradigm, environmental quality assessment will not be based on any one particular technology alone. Instead, it would rely on (a) extracting higher quality information from a variety of environmental sensing techniques, and (b) merging this information into a data assimilation framework. This task will encompass the following sub-tasks:
- Subtask 3.1.1 Atmospheric modelling
The Eulerian (grid) photochemical dispersion models CAMx and CMAQ will be used to provide hourly concentrations of PM10, PM2.5, NO2, ozone, and benzo(a)pyrene. Both of them will be applied at the regional scale and within a nesting approach at the urban scale in the ICARUS participating cities. CAMx and CMAQ will receive meteorological input fields from the meteorological models MM5 and WRF respectively. They include multiple gas phase chemistry mechanism options multiple gas phase chemistry solver options, multiple dry deposition options, ozone and particulate source apportionment technology and reactive tracer source apportionment for air toxics. Beyond state-of-the-art parameterization schemes for criteria pollutants such as BaP and gas-particle portioning will be used in the models. In addition, the Operational Street Pollution Model (OSPM) will be used to evaluate the transport contribution in traffic corridors to the overall tropospheric pollution burden. The emissions required will be computed via the COPERT methodology, using as input traffic volume and vehicle-specific emission factors. Speed-dependent expressions for vehicle-specific emission factors are already available to the ICARUS team.
- Subtask 3.1.2 Remote sensing
Further to the application of atmospheric dispersion models we will make use of Earth Observation data to derive the tropospheric concentration of air pollutants, through the calculation of synoptic indicators of air pollution, such as atmospheric aerosol optical depth (AOD). The pollution effects in satellite data of High Resolution sensors (HRS) are more apparent on certain spectral band combinations than on others. This permits a first delineation of polluted areas and the localization of emission sources, through photo-interpretation of HSR images (Landsat and SPOT). An overall quantitative evaluation of the pollution levels will be carried out by examining the scatterograms of the images and by applying local statistical analysis. The optical thickness of aerosol scattering in the green visible spectrum (at 0.55 m) will be used as quantitative spatially resolved indicator designating pollution loading. Normalizing the AOD values against the horizontal profile of the atmospheric mixing layer the relevant Mie scattering coefficient can be derived and correlated to near-surface fine aerosol concentrations providing concentration maps at very high spatial resolution (up to 10x10m) . Data from Moderate Resolution Imaging Spectroradiometer (MODIS) aboard NASA’s Earth ObservingSystem (EOS) satellites Terra and Aqua satellites will be also used to derive concentration fields at medium spatial resolution which can range from 500 up to 10,000 meters at nadir.
With regard to GHGs tropospheric concentration data will be obtained from the Greenhouse Gases Observing SATellite (GOSAT) for column-averaged dry-air mole fractions of CH4, and CO2 with ground pixel resolution of 10.5 km at nadir, and from the NASA Orbiting Carbon Observatory (OCO-2) for CO2 with ground pixel resolution of 1.3 × 2.3 km2 at best. EO data from the recent TROPOMI sensor aboard Sentinel-5P platform for column-averaged dry-air mole fractions of CH4, and CO2 which has a spatial resolution of 7 km at nadir will be used too.
An eco-friendly NASA-awarded Light Manned Aircraft (LMA) specialized in aerial mapping, air quality & environmental monitoring will be also used in pilot applications in three of the ICARUS participating cities (Athens, Thessaloniki and Ljubljana). The aircraft will provide air quality data at different height profiles as well as remote sensing images using hyperspectral sensors at very high spatial resolution (1x1m). The aircraft which is designed as an aerial platform of sensing environmental data, will operate at different height profiles above European cities providing measurements of air pollutants and emission data using sensor technologies as well as special sampling instrumentation.
The ICARUS LMA will also provide remote sensing images in order to detect urban gaseous emissions; using an aerial thermal camera it will provide urban heat mapping of the ICARUS cities informing local authorities who are planning urban vegetation and infrastructure programs.
- Subtask 3.1.3 Data fusion
Data from ground-based monitoring stations, provided by local and national monitoring networks, AIRBASE and R&D campaigns, atmospheric dispersion modelling results (Sub-task 3.1.1) and Earth Observation data (Sub-task 3.1.2) will be critically evaluated to: (a) assess data gaps and overlaps across different geographical and temporal scales; (b) propose the proper mechanisms for data integration with regard to filling gaps and smoothing data disruptions due to overlapping or discontinuous datasets. This information will guide the development of advanced data fusion and model nesting methods for the different data classes involved.
We shall start from a detailed error analysis and result intercomparison based on the data collected from the ICARUS participating cities. Error estimation across the computational domain in each city will be used as input to decision theory-based data fusion algorithms designed to reckon the best possible estimate of atmospheric composition following a scheme that is flexible enough to make use of the best possible information across the observation domain. In ICARUS data fusion will be done using a Kalman filter designed to account for the relative error introduced by each of the information sources. Such a framework will ensure that data singularities generated for example from point disturbances of the optical parameters field captured by the satellite sensor are eliminated. At the same time, atmospheric model weaknesses, such as the high dependence of result accuracy on the adequacy of the emissions inventory and boundary/initial values can be dealt with via fusion with the satellite-derived estimates and ground-based measurements. Other data fusion approaches such as the stationary variational method (3Dvar and its one and two-dimensional variants) and Optimal Interpolation (OI) will be investigated. In addition, artificial intelligence systems such as artificial neural networks (ANN) have been proven remarkable tools for addressing urban air quality for selected pollutants. Beyond purely forecasting tools, the utility of these types of modelling systems is of interest when: (a) higher spatial resolution is required and the existing monitoring network is not sufficiently dense and (b) there are significant gaps in the time series of the data available. The Broyden-Fletcher-Goldfarb-Shanno (BFGS) quasi-Newton algorithm will be used based on previous experience that demonstrated the BFGS algorithm to be the most efficient in solving unconstrained nonlinear optimisation problems. Reduction in error in air pollution assessment at high spatial resolution and with time deficient data is a typical problem of this kind.
As a last step, the data fusion scheme more appropriate for each ICARUS participating city will be devised and applied. The choice of the best algorithm will be influenced by the structure and quality of the local air pollution datasets and by the local user requirements with regard to relevance for effective policy support.
Task 3.2 Source apportionment
The objective of this task is to use source apportionment methods and – as far as necessary – measurements of concentrations of trace constituents in the atmosphere to:
a) further evaluate model results (in cooperation with task 3.1)
b) develop an approach for estimating changes in concentrations caused by changes in emissions of certain sources (e.g. wood combustion, road transport) without the necessity to use atmospheric transport modeling.
The necessary data will be primarily taken from the public monitoring networks, data from measurement campaigns and airborne/satellite and other remote sensing data. Additional measurements will only be performed, if relevant data are missing. This is expected especially for non-CO2 greenhouse gases. Beyond the measurements to fill data gaps, extensive monitoring for source apportionment will be executed in 6 cities (Athens, Thessaloniki, Madrid, Stuttgart, Ljubljana, Brno) for a period of one year, including two urban sites (urban background and traffic) and one regional. These will be selected among locations where monitoring stations already exist, completed by the additional instrumentation required in each monitoring station. Monitoring data will be collected for pollutants like NOX, O3, BC, SO2, benzene, PM10, PM2.5 but also the chemical composition of PM e.g. ions (ammonia, nitrate and sulfate), heavy metals, OC/EC (organic carbon/elemental carbon), BaP and selected NMVOC species. Ad-hoc targeted measurements campaigns will be used for assessing the impact of GHG policies applied in the participant cities during the project lifetime.
For the greenhouse gases CO2, BC, CH4, N2O, SF6 measurements will be made using a specially designed greenhouse gas analyzer, TRACE 1300 GC from G.A.S (Global Analyzer Solutions) THERMO Scientific, which consists of two parallel channels for simultaneous detection of greenhouse gases in air and is coupled with pneumatic control modules and automated port valves. The analyser is installed in NCSRD for high-precision concentration measurements of methane, carbon dioxide, nitrous oxide and SF6. Air samples from all participating cities will be collected in canisters and analyzed by NCSRD.
All atmospheric pollution data used in ICARUS will follow a strict quality control and assurance protocol. The QC/QA protocol will take in consideration three aspects:
(a) whether the data have been collected via a reference method according to Annex VI of Directive 2008/50/CE and annex V of Directive 2004/107/CE. The type of analysis or sampling instrumentation will be reviewed and their traceability and quality control and assurance will be assessed based on the standards.
(b) whether the monitoring networks in the participating cities meet the Data Quality Objectives set by Directive 2008/50/CE and Directive 2004/107/CE with reference to expanded uncertainty either on limit or objective values, minimum data capture and minimum time coverage.
(c) whether the data were obtained using non-reference analytical methods such as particles means automatic analysers; in this case, the correction factor vis-a-vis the reference method will be used to correct the respective dataset.
For source apportionment a number of well-established methods are available. The simplest approach is the spatial increment approach. The so-called Lenschow approach, which is based on the spatial increment approach, can especially be used to distinguish between local traffic, urban and regional sources. The first step of the Lenschow approach applies the increment approach. In further steps of the approach, the chemical composition of PM at traffic sites, industrial sites, urban, rural and regional background are included. A further rather simple method, the mass closure method, which is a tracer based approach, will be applied in the cities of Bristol, Athens and Thessaloniki. This approach is strong in calculating the contribution of marine aerosols, caused by sea spray emissions, mainly comprising of sodium and chloride, to the entire PM pollution in these cities. It will also be applied in southern cities like Madrid, Milan, Athens and Thessaloniki with silicon and/or aluminum as tracers for mineral dust from soil erosion especially during dry summer months. Identifying traffic emission sources, EC is used as tracer for PM exhaust emissions as well as Cu, Ba and Sb for the PM emissions caused by abrasion of brake and clutch ware. The weakness of this approach is, the difficulty to distinguish between different sources emitting the same tracer. Therefore well-established statistical receptor models will be used further for source apportionment. Depending on the knowledge about emission sources there is a wide range of receptor models available from multivariate models like PCA (principal component analysis) and PMF (positive matrix factor), if the knowledge about the emission source is limited, up to regression models and CBM (chemical mass balance) models, if the knowledge about the emission sources is complete. Multivariate models need more data to disentangle the sources and source contributions based on data intrinsic correlations between different compounds. With the help of auxiliary data e.g. heavy metals, PAH, organic tracers etc. as well as the temporal courses of these compounds, e.g. the seasonal, weekday/week-end and diurnal courses are necessary beside the PM concentrations to verify the source assignments. The Openair package for R will be used to evaluate spatial and temporal changes in source contributions in relation to meteorological and synoptic atmospheric scenarios. Assessment of chemical mass balance will rely on air pollution monitoring data coming from the regulatory monitoring networks, remote sensing (satellite and aerial) data or eventually modelled data in order to fill data gaps. Although a wide variety of pollutants (e.g. CO, NOx, O3, BC, SO2, benzene) will be included, focusing on data of PM chemical speciation (ions, heavy metals, OC/EC); the latter will give plenty of information on the contribution of the various sources present in cities. The tiered use of the methods above, starting from analyses of routinely available data and gaining higher complexity with increasing data availability and information, will deliver a complex data base with far reaching information about the sources causing ambient air pollution. The last step of this subtask will lead into the development of user friendly tools to authorities and other stakeholders grounded on readily available data. A complete guidance tool based on all results obtained will be drafted. The guidance will include recommendations on the application of source apportionment for the assessment of the effectiveness of abatement measures of air quality and CFP.
Results of the source apportionment will be delivered to task 3.1 to be used for the evaluation of concentrations and data fusion. In addition, focusing on the most toxic pollutants in EU cities, namely PM2.5, PM10, BC and NO2, we will develop a relationship between changes in emissions of certain emission sources and the change in concentrations based on the source apportionment results. Using this relationship, we can estimate concentration changes caused by policy/measure combinations without having to rely on atmospheric modeling, thus avoiding a source of large uncertainties.
We expect that the methodology can be established at least for large local and urban sources such as road transport. The methods will be supplied to WP 5 for application.
Task 3.3 Radiative forcing and climate modeling
This Task will provide all the climate data and information required to the impact assessments and analyses that will be performed within the project. To this aim, different climate change projections will be considered, according to the most recent CMIP5 emission scenarios. Specifically, high-resolution climate simulations performed within the Euro-CORDEX and Med-CORDEX programs will be used. Furthermore, in order to provide an assessment of the uncertainties related either to the models and to the emission scenarios, ensembles of simulations will be considered and analysed. The climate data will be suitably processed to build and provide indicators according to the case studies requirements. Different future climate scenarios will be considered, such as RCP4.5 and RCP8.5, covering the current century. RCP8.5 is the one with the highest rate of increase in greenhouse gas concentrations within the new set of Representative Concentration Pathways (RCPs). On the other hand, RCP4.5 is a scenario with a moderate rate of increase in GHGs.
Climate data will be provided at the regional (Euro-Mediterranean domain) and local (order of tens kilometers) scale, thus focusing on the selected cities.
Radiative forcing change (ΔF) due to greenhouse gases will be estimated according to the IPCC simplified methodology, which estimates ΔF as a function of changing gas concentration through algebraic formulations specific to each gas. The impact of aerosol on the local radiative balance can be large but also very complex. The effect is the scattering of part of the incoming solar radiation back into space. This causes a negative radiative forcing which may partly, and locally even completely, offset the enhanced greenhouse effect. However, due to their short atmospheric lifetime, the radiative forcing of aerosols is very inhomogeneous in space and in time. Some aerosols, such as soot, absorb solar radiation directly, leading to local heating of the atmosphere, or absorb and emit infrared radiation, adding to the enhanced greenhouse effect. This forcing depends on aerosol concentration and chemistry (the latter determines not only absorption of radiation by soot but also growth by water uptake and increased backscatter) and the aerosol mixing state. In ICARUS aerosol optical depth (AOD) and single scattering albedo derived from satellite measurements, surface albedo and phase function will be used for the calculation of radiative forcing. The value of ΔF consists of the longwave and shortwave parts. Shortwave radiative forcing is the primary target of ICARUS. Characterization of aerosol in terms of global warming potential will be done through the development of a simplified scheme that translates different types of emissions (provided by WP 2) into a first order assessment of radiative forcing. This will be done by comparing emissions of different species with the resulting average aerosol characteristics and the resulting radiative forcing.
The significance of the obtained results in terms of radiative forcing change will be evaluated. In case of significant differences, we plan to simulate the corresponding climate through the CMCC high resolution, fully coupled, regional climate model (COSMO-CLM atmospheric model coupled with NEMO ocean model) in an ensemble framework to provide surface temperature, mass and energy fluxes for the impact assessments and analyses.
|Deliverable Number||Deliverable Title||Lead beneficiary||Type||Dissemination level||Due Date (in months)|
|D3.1||Delivery of the climate data and climate indicators at the regional
and local scale, together with a technical report
describing the dataset.
|17 – CMCC||Report||Public||6|
|D3.2||Report on data fusion methodology||1 – AUTH||Report||Public||15|
|D3.3||Report on AQ and GHGs concentration at the ground level in the ICARUS cities||1 – AUTH||Report||Public||18|
|D3.4||Report on results of source apportionment in all participating cities||2 – USTUTT||Report||Public||18|
|D3.5||Technical report on the evaluation of the changes in the surface radiative
forcing due to implementation of mitigation strategies at local level
|17 – CMCC||Report||Public||42|