Urban hybriD models for AiR pollution exposure Assessment (UDARA)

Lead Research Organisation: University of Manchester
Department Name: Earth Atmospheric and Env Sciences

Abstract

Air pollution affects the health of the world's population. The WHO estimate that 3.7 million deaths are annually attributable to outdoor air pollution (6.7% of all deaths globally) and urged Member States to develop air quality monitoring systems and health registries to improve surveillance for all illnesses related to air pollution. Establishing and maintaining air quality monitoring infrastructure can be costly. Furthermore, air quality monitoring alone does not provide the basis for robust analysis of health outcomes or onward air quality management.

Health risks due to air pollution in densely-populated Indonesian urban areas are increasingly associated with transport activities, whilst forest fire emissions further exacerbate health impacts associated with air-pollution. This represents a major health concern for Indonesia with both economic and welfare implications. In order to enable effective control measures that are feasible within wider development, welfare and economic goals, much improved quantification of the exposure of the population to pollution is required. To enable such quantification, the contributions from long-range and local sources pollutants and the availability of suitable health data to the determine health impacts of pollutant exposure must be established.

The UDARA project will be a multidisciplinary collaboration between the University of Manchester and 3 Indonesian Institutions; the Institut Teknologi Bandung, National Nuclear Energy Agency of Indonesia and the University of Padjadjaran involving exposure scientists, atmospheric scientists and health scientists. The project will develop a new approach for providing reliable exposure estimates for analysing the impact of air pollution on health outcomes in Indonesian cities. It will employ a new methodology developed within the EU ESCAPE project in response to the lack of European-specific data on exposure-response relationships and the uncertainty introduced into health outcome analyses by the use of varied exposure models. ESCAPE developed standardised protocols for developing and testing LUR models, while still allowing sufficient flexibility to incorporate application specific factors. The further refinement to incorporate remote sensing and chemical transport model outputs has improved model performance in complex pollution climates such as experienced in Indonesia. The ESCAPE project involved the development of both classic and hybrid models but their performance in the sort of pollution climate seen in Indonesian urban areas has yet to be investigated.

Within UDARA, we will collect new data in Jakarta, a heavily populated city with a pollution problem dominated by urban emissions, and a second city in Sumatera substantially influenced by seasonal biomass burning. The highly spatially resolved measurements of both gaseous and particulate pollutants and data from two 3D pollution models will be used within the ESCAPE methodology to derive pollutant exposure estimates for the two cities. We will additionally conduct an evaluation of the quality and quantity of Indonesian health outcome data that can be used for subsequent future epidemiological studies.

UDARA will thereby provide a new methodology and robust basis for evaluating the health outcomes from air pollution exposure in Indonesia.

Planned Impact

The principal beneficiaries of the research will be those responsible for public health and air pollution control in Indonesia. These include the following government departments and agencies: Ministry of Environment and Forestry (Directorate of Air Pollution Control), Ministry of Health (Directorate General of Public Health), Provincial Environmental Protection Agencies, Local Authorities, Local Environmental Agency, City Health Offices, City Transportation Office, Ministry of Transportation (Directorate General of Land Transportation), Meteorological, Volcanology and Geophysical Agency (BMKG), NGOs (KPBB, Pelangi), Clean Air Initiative (CAI) - Asia, University's Research Centres, Statistical agencies (BPS) and the Geo-Spatial Information Agency (Badan Informasi Geospasial, or BIG). Clearly the ultimate beneficiaries will be the general public of Indonesia who will benefit from the engagement of UDARA researchers with the listed stakeholders and directly through the listed outreach activities.

In countries with well-developed capacities for evaluation, the health and consequent economic impacts of exposure to atmospheric pollutants have been found to be substantial. Mitigation measures can be implemented only where data are available for assessment and compliance. UDARA will provide the capability for evaluation of air pollution exposure in Indonesia in addition to an assessment of the availability and quality of health data for effective epidemiological studies. This information will provide a powerful decision making tool, delivering the basis for the implementation of air quality control measures at local, regional and national scale. The information provided by the methodology developed within UDARA will be able to identify the optimum control measures to mitigate against adverse health outcomes in Indonesia at a range of scales.
The Directorate General of Public Health will be able to use exposure assessments and eventually the results from any resultant health outcome prediction capability as a planning tool to enable direction of resources.
The Directorate of Air Pollution Control and Provincial Environmental Protection Agencies will be able to give advice to Government to enable planning decisions and pollution control measures to be implemented.
The general public in the target cities will benefit from the knowledge of their exposure and eventually from control measures implemented through the measures above. Increased public awareness will result from the project outcomes.

Publications

10 25 50
 
Description We have demonstrated the utility and limitation of the use of low cost sensors for continuous monitoring of air pollution by their continuous deployment in Jakarta. We have further developed a land-use regression (LUR) model of the city to predict the geographical distribution of air pollution exposure during the non-biomass burning and biomass burning seasons. We have further developed (and published; see publications) a tool to generate a LUR model that can take a hybrid approach, incorporating regional air quality model and satellite data and applied it to Indonesia. Finally, we have developed a methodology to use existing Indonesian Family Lifestyle Survey (IFLS) data to predict exposure and health effects in geographical regions where data are lacking (submitted for publication, Journal of Epidemiology & Community Health - Manuscript ID jech-2021-216832). All other aspects are at the state of preparation for publication.
Exploitation Route All techniques have been or will be made available for open use and application for prediction of exposures and health effect predictions
Sectors Environment

 
Description There has been some modest impact in that our Indonesian partners have been appraising local stakeholders (local government, general public working groups etc.) of the ouputs from the project in lieu of our anticipated workshops and other engagement activities. However, whilst Covid has had only a modest impact on our scientific outputs, it had a more substantial impact on these stakeholder engagement activities, since a substantial amount of activity was planned around a joint workshop and series of visits which we were unable to arrange. We endeavoured to rearrange these activities for when international travel restarted, though it had not resumed prior to completion of the award and no further activities were possible in the context of UDARA
First Year Of Impact 2020
Sector Environment
Impact Types Societal

 
Description Training of participants in the use of new LUR tool and sensor operation
Geographic Reach Local/Municipal/Regional 
Policy Influence Type Influenced training of practitioners or researchers
Impact We have provided training to develop expertise in the operation of, and analysis and quality assurance of data from, low cost optical particle detectors for particulate matter (PM) measurement. We have similarly provided the LUR model tool (see publications) and the associated training for running. Without this expertise, the delivery of such measurement and air pollution exposure modelling would be reliant on international collaboration.
 
Title Data collection app 
Description A mobile phone app for recording of pollutant measurement locations and collection of geolocated data at each sampling site has been constructed and deployed in Jakarta. 
Type Of Material Improvements to research infrastructure 
Year Produced 2018 
Provided To Others? Yes  
Impact The app has been provided to our partners at ITB 
 
Title Optical particle counter package construction and supply 
Description We have constructed 30 optical particle counter packages, each comprising a Alphasense OPCN2 units, a Raspberry pi0, WiFi connection, logging card SPI interface and power supply. Bespoke software has been written, automatically resuming realtime data collection after power interruption. These have been supplied to ITB and deployed alongside gas sensors to provide reasonably comprehensive low cost pollution measurement packages. 
Type Of Material Improvements to research infrastructure 
Year Produced 2018 
Provided To Others? Yes  
Impact The sensor units are continuously available to our ITB partners and have been deployed in Jakarta and are awaiting deployment in the second city (likely Banjarmasin). 
 
Title Construction of the 3 domain nested WRF-Chem setup for Indonesia 
Description The regional coupled WRF-Chem model has been set up for the Indonesian domain (with the finest nests over the two target cities), adopting the organic aerosol scheme from the PROMOTE activity (see separate entry). 
Type Of Material Computer model/algorithm 
Year Produced 2018 
Provided To Others? Yes  
Impact This model will be used within UDARA to predict the regional contributions to AQ degradation in the 2 chosen cities for evaluation of the background contribution to adverse health outcomes. It can be used by our Indonesian partners. 
 
Title PROMOTE Organic Aerosol Module (sub-components also adapted and used for UDARA) 
Description We have put in place a new module for the description of primary and secondary organic aerosol within the WRF-Chem model, based on the widely used "volatility basis set" (VBS) approach. The primary organic aerosol is considered as either anthropogenic or biomass burning in origin and carried as two separate volatility distributions in the module. Secondary organic aerosol is derived from two biogenic and two anthropogenic surrogate precursors and each is considered in its own volatility distribution. All these components are carried in each of 8 size fractions in the MOSAIC aerosol module. 
Type Of Material Computer model/algorithm 
Year Produced 2018 
Provided To Others? No  
Impact The module is fully developed and tested. We are using it to create a Gaussian process emulator that will enable the parameters in the module to optimally reproduce the measured organic aerosol. The module will then be passed to our collaborators at IITM for adoption by the SAFAR operational air quality system. The same module will be used within the UDARA project on the Indonesian domain (though not using the emulator optimised parameters for India). 
 
Description Collaborative partnership with partners at Bandung Institute of Technology (ITB) for measurement, data analysis, modelling and interpretation within UDARA 
Organisation Bandung Institute of Technology
Country Indonesia 
Sector Academic/University 
PI Contribution A joint meeting was held in April 2018 to kick off the collaboration. Since then we have jointly carried out most UDARA investigations, with specific tasks taken on by each partner where the skillset is most suitable. In Manchester we have conducted extensive screening of Indonesian Family Lifestyle Survey (IFLS) data across the country in collaboration with Indonesian partners. We have used GIS tools to identify optimum locations for measurements in Jakarta and created a mobile phone app for recording and collection of measurement data. We have constructed and provided 30 sets of particle measurement instrumentation for deployment alongside other samplers in Jakarta. We have constructed a regional model for deployment over Indonesia for prediction of regional contributions to pollution and degraded health outcomes. We have continued the collaboration to participate in the joint analysis of data in each workpackage. We have developed a tool for the construction of a hybrid land-use regression (LUR) model, that ingests measurement, regional air quality model and satellite data for application in the project (published in the Journal of Open Source Software, see publications). We have further developed, with our partners, and approach to use IFLS data to predict exposure and health effects in geographical regions where data are lacking (submitted for publication, Journal of Epidemiology & Community Health - Manuscript ID jech-2021-216832)
Collaborator Contribution We were hosted at ITB Bandung by partners, who have jointly screened and analysed IFLS data ready for writing the joint publication (submitted to Journal of Epidemiology & Community Health - Manuscript ID jech-2021-216832). Guided by our GIS activity, partners have deployed the pollution measurement packages (including the particle sensors we constructed and provided) across Jakarta for several months for air quality data collection and data are under QA. Our partners have collected the continuous measurement dataset for use in the LUR model and are in the process of quality assurance and data analysis. They have taken delivery of the LUR model and are preparing it for application. We have continually metonline, though Covid has prevented our anticipated visits to Indonesia and our face-to-face stakeholder engagement activities.
Impact One paper has been submitted for publication to the Journal of Epidemiology & Community Health - Manuscript ID jech-2021-216832
Start Year 2017
 
Title XLUR: A land use regression wizard for ArcGIS Pro 
Description A Python toolbox for ArcGIS Pro that enables the development and application of land use regression models. 
Type Of Technology Software 
Year Produced 2020 
Open Source License? Yes  
URL https://zenodo.org/record/3877328
 
Title XLUR: A land use regression wizard for ArcGIS Pro 
Description Notable changes Warning message appears at start up of BuildLUR, if no Spatial Analyst license is found. 
Type Of Technology Software 
Year Produced 2020 
Open Source License? Yes  
URL https://zenodo.org/record/3889604