EARTH OBSERVATION FOR ENVIRONMENTAL AND HEALTH IMPACT ASSESSMENT A METHODOLOGY WITH SYNERGIES FOR EUROPEAN POLICIES Andreas N. Skouloudis (1), David G. Rickerby (2), Peter Pärt (3) (1) European Commission, DG-JRC, IES TP.272, Italy, Email: andreas.skouloudis@jrc.it (2) European Commission, DG-JRC, IES TP.272, Italy, Email: david.rickerby@jrc.it (3) European Commission, DG-JRC, IES TP.263, Italy, Email: peter.part@jrc.it ABSTRACT Although much progress has been made in improving the quality of the environment, it is estimated that up to one third of the global burden of disease might be attributable to environmental factors. This is not easily verifiable with classic epidemiological studies, due to the uncertainties in attributing accurately quantifiable health effects to environmental stressors. This work shows how reliable satellite monitoring can be utilised in assessing exposure to children up to four years old living in the vicinity of roads with heavy traffic and identifying hazards from occupational exposure to toxic substances to agricultural employees. Data sources from various disciplines are utilized in a methodology, which applies the latest satellite observations at high spatial and temporal resolutions to establish links with health effects. 1. POLICIES WITH NEW EARTH OBSERVATION SYNERGIES In the early 8s the European environmental policies for sustainable development were formulated by taking into account only the effect of emissions on the environment without taking into consideration the transportation of pollutants and the natural background. The relevant European directives are the 85/23/EEC on air-quality standards for nitrogen oxides and 8/779/EEC on limit values and guide values for SO 2 and suspended particles. These were based on models focusing purely on emission logistics, which were then used in carrying out forecasts (with RAINS, GAINS, etc) with the expectation that they meet emission standards and economic objectives. The applicable legislation can be consulted at [1]. However, these policies have not achieved the environmental objective of reducing pollutant concentrations in the atmosphere, mainly because they were decoupled from important processes such as the dispersion and chemical transformation of pollutants. These were followed in the 9s by the directives for transport and the environment (directives 1998/69/EC and 1999/96/EC see [2]) as well as the Daughter Directives (1999/3/EC, 2/69/EC, 22/3/EC and 24/17/EC) as described at [1]. These accounted for transport and chemical transformation but were based on data from a sparse monitoring network of expensive stations that could not provide a harmonised picture of atmospheric pollutions at high spatial resolution. It was realised that ideally, policies should focus on effects (for example, on human health or on the economy) and should cover reasonably long time periods since some of the effects are partially cumulative in nature, and therefore appear only in several consecutive years. The European Commission has recently introduced a third level of complexity by implementing an action plan [3] for reducing the disease burden caused by environmental factors in the EU and for identifying and prevent new health threats caused by environmental factors. Efforts therefore are being made to set up information systems which link health effects to environmental causes through a set of monitoring parameters which are generally referred to as indicators. These indicators have been utilised until now in compliance monitoring, which was the primary source of information in epidemiological studies. Unfortunately, this is inadequate to identify concretely the health and environment effects. In addition, the peak of health effects occurs at specific locations (hot spots) during specific episodes. Hence, large area averaging and reporting over the whole year are not sufficient and result in doubtful epidemiological conclusions. Effect assessment has also to take into consideration the fact that both the human and environmental populations consist of individuals every one of us reacts in a unique way to a challenge or stressor. The key concept is advanced high spatial resolution monitoring of populations susceptible to specific environmental hazards. A wider use of earth observation services may assist in facilitating and harmonising the larger number of monitoring and reporting requirement of the European Union Environment and Health Action plan. A new methodology is proposed in this paper, which provides a synthesis of satellite observations, together with new technologies for environmental monitoring that offer new opportunities of exploiting health data with the scope of: Proc. Envisat Symposium 27, Montreux, Switzerland 23 27 April 27 (ESA SP-636, July 27)
1) the creation of comparable standardised maps, 2) easing their subsequent interpretation and use across Member States, 3) exploiting environmental resource data bases in a more timely and cost-effective way, 4) integration of recommendations for harmonised monitoring methods, 5) testing through the integrated health statistics, 6) development of suitable data and information management systems, and 7) analysing occupational exposure effects such as toxic substances in agricultural areas. In addition, the advancement of biological knowledge and technologies over the last 2-3 years has made available a completely new array of techniques. This work examines the remaining challenges for research in areas where: data are available and impact assessments are feasible; attribution to environmental sources can be carried out already. Of immediate value is the temporal characterisation of human activities with satellite observations obtained from several sources. Ideally, it can be expected that this will change significantly as the use of new technologies for providing more reliable satellite monitoring will become available, allowing real-time information regarding the location of vulnerable population groups and the identification of the causes of environmental problems. Such images include night lights sources [4] as in Fig. 1, maps of land use [5] and digital elevation [6]. The spatial attribution of population [7] is then utilised in association with other layers of data for identifying specific population groups that might be subjected to risks from environmental causes. The methodology used consists of the following steps: 1. Generate or obtain population density maps with spatial resolution of less than 1x1km 2. 2. Based on age-pyramid data (country or regional basis) calculate the population density map of the specific age group to be examined. 3. Obtain suitable road network layers or layers of areas that are important hot spots for pollutant emissions. 4. Produce strips at 5m, 2m and 35m distance from each side of these roads. 5. Overlay the maps from steps 2 and 4 and then intersect the data to calculate the population affected in the buffer area. 6. Sum the totals for the whole country or the region of interest. 7. Repeat the process by changing the areas of interest in step 3 (e.g. ports) 8. Repeat the process when new data on the road infrastructure or population data become available (ideally every five or ten years). 2. THE METHODOLOGY Epidemiological statistics incorporate uncertainties due to difficulties of associating the health effects in population groups to specific environmental causes. To eliminate these uncertainties, we utilised several types of satellite images to attribute the population in a grid of a resolution of 1x1 km 2. Figure 2. Population density in Greater London and the road network with a buffer of 35m from each side of the road. In this way geographical areas and population groups are identified that are expected to show evidence about the environmental burden of diseases. Figure 1. Image of night light sources from [4]. Fig. 2 is a composite image of population density and the road network with heavy traffic in the Greater London Area. Roads with high traffic are considered to be: motorways, national roads with double lanes, national roads and other principal roads as classified at
the GISCO by Eurostat [8]. For reasons of simplicity, population densities less than 1 inhabitants per km 2 are represented in white. Current knowledge of health consequences from atmospheric pollution could be then used for calculating the consequences for sensitive population groups. One such group is considered to be the children up to four years old who live either in the vicinity of roads with high traffic flows or near ports or major ferry lines. This information can also be superimposed on maps of atmospheric pollution and noise levels in order to identify hotspots for these particular problems. These variations in the future will help to provide an assessment of where health statistics diverge from the national mean and indicate if improvements due to reducing air-pollution levels would be feasible. For such comparisons, pollution maps (e.g. PM 1, PM 2.5, ozone and noise) of high spatial resolution are needed. These could be obtained from measurements, modelling or satellite observations with instruments such as the Scanning Imaging Absorption Spectrometer for Atmospheric Cartography (SCIAMACHY), see [9]. Maps obtained from the monitoring networks are not so suitable due to the low spatial resolution provided by monitoring sites and the scarcity of the number of stations in the various Member States, despite the European directives concerning the monitoring and reporting of atmospheric pollution. Furthermore, the data from such stations frequently have temporal availabilities below 85%. Maps produced by modelling are consistent geographically and can in theory provide information in resolutions of less than 1x1 km 2 but are equally uncertain due to the emission inventories used. They require spatial and temporal validation, there are uncertainties in the chemical mechanisms, and significant computational times are required for producing near real-time maps. This causes problems even on the combination of the previous two methods utilising assimilation techniques. The third option, of utilising satellite observation with instruments like SCIAMACHY, is realistic and uses comparable and standardised tools. It has three different viewing geometries: nadir, limb, and sun/moon occultation, which yield total column values as well as distribution profiles in the stratosphere and (in some cases) the troposphere for trace gases and aerosols. However, the technique provides total column concentration in resolutions which are coarse for identifying health effects (limb vertical 3x132km 2, nadir horizontal 32x215km 2 ). Furthermore, it is not yet possible to acquire observations of several images per day over the same location. For these reasons, this study has still the limitation of not having suitable environmental maps to overlay over the population and infrastructure data, for comparable time periods with comparable spatial resolutions. Hence, health effects cannot be directly calculated at this stage. 2.1 Frequency of updates Updates of the methodology could be carried with a frequency of: Ten years for actual population census data. Annual updates for estimates of population increase. Annually for estimates of population increase and the attribution of children at the age group of -4 years old. Annually, for air pollutant concentrations and statistics of health effects. Annually, for traffic statistics. 2.2 Data quality The identification of the location of children is based on the population density maps from campaigns that are repeated every ten years. The attribution of children in various age groups is obtained from the national average age pyramids. These numbers could be differentiated in future, on a per NUTS-3 (regional) basis. Further differentiation could be carried out according to gender and social condition. Further improvement can be achieved with: spatial updates incorporating different distances from roads, ports, ferries or industrial installations, other population groups considered according to the health effects, better identification of roads with different traffic loads (vehicles per hour). The temporal evolution is expected to vary significantly only when new census data become available (every 1 years in most countries) or when the road network is significantly modified. However, when these data layers are overlaid with pollution maps, annual revisions will become necessary. 3. HEALTH EFFECTS FOR CHILDREN Although health consequences cannot be determined due to the absence of reliable pollution maps it is still possible to compare the absolute numbers of children in the fifteen EU member states (prior to 24)
affected by two major sources of atmospheric emissions, and to identify where regulatory and research efforts should focus. The composite information shown in Fig. 3 refers to the children living within proximity of 5m from main traffic sources in EU member states before 24. The country codes in this figure are according to ISO 3166 International Organisation for Standardization, see [9]. The percentages refer to the overall number of children in this age group for 2) [11 and 12]. Fig. 4 shows similar data with reference to ports and ferry lines that are close to populated areas. Children of age -4 years old Children of age -4 years old 9, 8, 7, 6, 5, 4, 3, 2, 1, 2.7% 3.4% 2.2% Distance from each side of road: 2.2% 2.5% 1.1% 1.6% 2.1% 1.9% 2.% 3.6% 2.% 5m 1.7% 1.3% AT BE DE DK ES FI FR GR IE IT LU NL PT SE UK 2.% Figure 3. Country comparison of children within a proximity of 5m from heavily trafficked roads. 1,2 1, 8 6 4 2 Distance from ports/ferry lines: -.2%.9%.135%.9%.137%.27% 5m.42%.116%.52% -.1%.95%.228%.22% AT BE DE DK ES FI FR GR IE IT LU NL PT SE UK Figure 4. Country comparison of children within a proximity of 5m from ports & ferry lines. Fig. 5 is an inter-comparison of children in the same age group living within areas of different proximities (5m, 2m and 35m) in three urban areas. The size of these areas is 1x1 km 2 as defined by the AutoOil-2 programme of the European Commission [13]. These are the city areas with the highest NO 2 concentrations [9]. Based on this figure, it is evident that in a buffer zone of 35m+35m around these roads live a significant number of children between the age of zero and four years old. Children of age - 4 years old 1 9 8 7 6 5 4 3 2 1 Distance from each side of the road: 5m 2m 35m London Milan Utrecht Figure 5. Inter-comparison of children living within a proximity of 5m, 2m and 35m from roads with heavy traffic in areas affected by significant NO 2 concentrations. Based on the last three figures the following conclusions can be presented for the analysis carried out for the year 2: 2% of children between zero and four years old appear to be exposed to risks due to the vicinity of main roads at a distance up to 5m from each side of the road. Countries with higher population appear to have more children exposed to risks due to proximity to roads. In countries with dense road network, more children are affected in the specific age group examined. Countries (LU, BE, ES) with high average population densities appear to have a high percentage of children exposed to traffic sources. Countries with long coasts and islands (SE, FI, DK and GR) appear to have a higher percentage of children between and 4 years old exposed due to port and ferry activities. At distances of up to 35m from each side of roads with heavy traffic, live more than 5, children of -4 years in the London, Milan and Utrecht areas. 4. OCCUPATIONAL HAZARDS The same methodology was utilised in a slightly different application. Satellite observations were utilised in order to identify areas where the rate of change for agricultural use is significant over recent years. In fig. 6 is shown an area with a significant rate of construction of greenhouses. The population density and the mortality rate have been examined for the same area. In fig. 7 is shown the mortality rate per 1 inhabitants for the semi-urban population from all
causes over several years. It is evident from this figure that the provincial data are three times higher than the national rate and twice the values of the regional data. of composite data sets at the same spatial and temporal resolution. With the emergence of new satellite data at high spatial and temporal resolution, such studies will become possible, allowing policies to be targeted at sensitive population groups, as well as for identifying problems masked within national and regional mean values. Figure 6. Rapid changes of land use for agricultural production. in Greece. Deaths per 1 inhabitants 18 16 14 12 1 8 6 4 2 Lasithi Crete Greece 1997 1998 1999 2 21 22 23 24 25 Years Figure 7. Differences of provincial, regional and mortality on rural and semi-urban population in Greece. In addition the annual variations and the trends are increasing for the provincial data. By carrying out simple ground visits it was possible to identify waste from extensive use of toxic substances (even DDT) in the proximity of greenhouses as well as inhabited areas. High resolution satellite observations taken at regular intervals can help in identifying wastes from such activities, to calculate potential exposure to nearby populations or directly to workers in the fields. Areas where such monitoring should be carried out are those where extensive changes in the use of agriculture land has occurred as well as in areas where mortality rates are significantly different from national and regional averages. 5. PROSPECTIVES AND FINAL REMARKS No European policies are currently focusing on children s health separately from other population groups, mainly because this requires the construction Figure 8. Canisters of DD near agricultural areas. Over the short and medium term, it is possible to identify the synergies between exposure and environment and attribute effects on population groups according to age or occupation. Merging relevant layers of environmental data is also feasible for assessing the environmental burden of diseases. There is unfortunately a temporal lag in integrating layers of information for environmental monitoring and this can affect cumulative population doses. Environmental monitoring with satellites can significantly enhance the consistency of reporting of environmental problems and the improvement of geographical representativeness by using remote sensors. We should seek combinations with other methods on the ground which will enhance the spatial attribution of
human populations in a more dynamic way (energy consumption, telecommunication use, traffic counts etc). This together with near real-time overlaying of environmental information will allow identification of important acute health effects. Finally, for occupation hazard assessment, satellite observation can be of valuable assistance in areas where local health rates are significantly different from the macro-scale rates. 6. REFERENCES 1. European Commission, (accessed Mar. 27), DG Environment, Air-Quality Existing Legislation, http://ec.europa.eu/environment/air/existing_leg.ht m 2. European Commission, (accessed Mar. 27), DG Environment, Transport and Environment, http://ec.europa.eu/environment/air/transport.htm http://www.esa.int/esaeo/sem34nkpzd_index _.html#subhead8 1. GISCO ((accessed Mar 27), Database Manual - part 1 chapter 8 http://eusoils.jrc.it/gisco_dbm/dbm/p1ch8.htm#top 11. World Health Organization Regional Office for Europe (27). European health for all database, (HFA-DB), http://www.euro.who.int/hfadb. 12. U.S. Census Bureau, (24 Aug. 26), International Data Base (IDB), http://www.census.gov/ipc/www/idbpyr.html 13. Skouloudis A.N. & Suppan P., (2). Methodology of the AutoOil-2 Programme of the European Commission DG-ENV, Eur 19556, http://autooil.jrc.ec.europa.eu/documents/aq%2 Modelling%2METHODOLOGYv4.pdf 3. European Commission, (24). DG Environment, Environment & Health Action Plan 24-21, COM-24-416, http://ec.europa.eu/environment/health/action_plan.htm 4. National Geophysical Data Centre (accessed Mar 27) DMSP-OLS Night time Lights Time Series, version 2. http://www.ngdc.noaa.gov/dmsp/global_composite s_v2.html 5. European Environment Agency (25), Corine land cover 2 (CLC2) 1 m - version 8, http://dataservice.eea.europa.eu/dataservice/metad etails.asp?id=822 6. European Environment Agency (25), Digital elevation model of Europe - version 1, http://dataservice.eea.europa.eu/dataservice/metad etails.asp?id=65 7. European Environment Agency (25), Population density disaggregated with CLC2, http://dataservice.eea.europa.eu/dataservice/metad etails.asp?id=83 8. Eurostat, European road network, Version 4. GISCO Layer RD Roads, http://eusoils.jrc.it/gisco_dbm/in/rd/dbm/inrd.htm 9. European Space Agency (18 Nov 25), Global air pollution map from the Scanning Imaging Absorption Spectrometer for Atmospheric Cartography (SCIAMACHY),