{"id":757810,"date":"2021-09-13T13:43:25","date_gmt":"2021-09-13T20:43:25","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?post_type=msr-project&p=757810"},"modified":"2022-02-08T09:59:20","modified_gmt":"2022-02-08T17:59:20","slug":"air-pollution-sensing-and-causal-modelling","status":"publish","type":"msr-project","link":"https:\/\/www.microsoft.com\/en-us\/research\/project\/air-pollution-sensing-and-causal-modelling\/","title":{"rendered":"Air Pollution Sensing and Causal Modelling"},"content":{"rendered":"
\n\t
\n\t\t
\n\t\t\t\"heatmap\t\t<\/div>\n\t\t\n\t\t
\n\t\t\t\n\t\t\t
\n\t\t\t\t\n\t\t\t\t
\n\t\t\t\t\t\n\t\t\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\n\n

Air Pollution Sensing and Causal Modelling<\/h1>\n\n\n\n

Enabling cost-efficient collection of granular spatiotemporal pollution data through drive-by sensing<\/p>\n\n\t\t\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t<\/div>\n\t\t<\/div>\n\t<\/div>\n<\/section>\n\n\n\n\n\n

Over 91% of the world\u2019s population lives in areas that exceed WHO guideline limits for air quality. Globally, up to 10 million deaths annually are attributed to ambient air pollution \u2014 higher than malaria and HIV. Unfortunately, India registers a large portion of these deaths, approximately over 2 million deaths \u2013 a number greater than all the deaths in 2020-21 due to Covid-19, Tuberculosis, Malaria, and AIDS combined. There are number of studies that have shown direct causality of air pollution with cardiovascular, cerebrovascular, pre\/neo-natal diseases and pediatric health. Air pollution not only poses a significant health risk to humans but also is a major factor leading to climate change. To ensure quality of life for the people, present and future, it is critical we reduce air pollution across the world.<\/p>\n\n\n\n

There have been a lot of public policy measures taken to address pollution in the country but for it to be effective we need a nuanced data-driven approach to understand and predict air pollution. We need to be able gather air pollution data and strengthen the ability to monitor air quality across locations, especially in areas close to hospitals, schools, and workplaces. Low-cost sensors and other emerging technologies can help improve and expand air pollution monitoring in areas that are currently underserved. Further, we need models that can help determine the factors causing air pollution to identify the right policy approaches for effective interventions.<\/p>\n\n\n\n

At the lab we have been working on the problems of enabling better collection of granular spatiotemporal air pollution data and developing models that can help determine the causal factors of air pollution. Through our efforts we have designed and implemented improved approaches to air pollution data collection in Delhi and Bangalore. We have also developed site-specific models to infer causal factors and predict air pollution in Delhi.<\/p>\n\n\n\n

Learn more about our work:<\/p>\n\n\n\n

\n
Air pollution sensing<\/a><\/div>\n\n\n\n
Causal modelling<\/a><\/div>\n<\/div>\n\n\n\n\n\n

Drive-by air pollution sensing<\/h2>\n\n\n\n

Enabling cost-efficient collection of granular spatiotemporal pollution data through drive-by sensing<\/em><\/p>\n\n\n\n

\"heatmap<\/figure><\/div>\n\n\n\n

A crucial step in tackling air pollution is to measure air quality at a fine spatiotemporal granularity. Access to quality data is the first step to perform wide-ranging analyses that include \u2014 identifying sources of air pollution, monitoring compliance of air quality standards, measuring the efficacy of various interventions, etc. Multiple \u2018smart city\u2019 initiatives around the world have touted to solve this problem by having a static deployment of either thousands of low-cost sensors or a few tens of reference-grade expensive monitors. This approach, however, comes with its own challenges \u2013 1) installing these sensors is capital intensive, making it difficult to scale to smaller cities and towns; 2) these static sensors do not capture the spatial variation in pollution within monitored regions.<\/p>\n\n\n\n

A promising approach for several smart city projects, called drive-by sensing, has been explored to address these challenges. Drive-by sensing leverages vehicles retrofitted with different sensors (pollution monitors, etc.) to capture the desired spatiotemporal pollution data at a fraction of the cost. Most of the drive-by sensing efforts have also often used fixed-route vehicles such as public transport buses. However, this approach still leaves many spatiotemporal gaps in data collection due to it being limited to specific routes and times.<\/p>\n\n\n\n

\"Drive-by<\/figure><\/div>\n\n\n\n

To address these gaps, we have been exploring an approach leveraging taxis for drive-by sensing. We have developed a system to bootstrap drive-by sensing deployment by taking into consideration a variety of aspects such as spatiotemporal coverage and budget constraints. Our system significantly outperforms the baselines when a fleet comprises of both taxis and fixed-route vehicles such as public transport buses.<\/p>\n\n\n\n

We have validated our approach through pilots with our partners, Ola Cabs and Three Wheel United. In Collaboration with Ola Cabs, one of India\u2019s largest cab aggregators, deployed a drive-by sensing pilot in Delhi by retrofitting 20 cabs with sensors. We have also piloted the system with autorickshaws in Bengaluru in collaboration with Three Wheel United. In the pilot, 17 autorickshaw were fitted with sensors and the drivers were incentivized to take specified routes. Through these approaches we were able to cover areas that were either underserved or inaccessible to public transportation, ensuring we had a cost-effective and wider collection of granular spatiotemporal pollution.<\/p>\n\n\n\n\n\n

Enabling data driven air pollution policy making<\/h2>\n\n\n\n

Inferring site-specific factors of pollution for effective data-driven policy making<\/em><\/p>\n\n\n\n

\"Data<\/figure><\/div>\n\n\n\n

There have been a lot of public policy measures taken to address pollution in the country, but these interventions have not always been highly effective. For effective interventions we need to understand air pollution and analyze both local and global air pollution data to be able to identify causal factors and predict air pollution.<\/p>\n\n\n\n

Air pollution consists of various kinds and sizes of particulate matter, out of which PM2.5 \u2013particulate matter that is 2.5 micrometers and smaller – contributes to over 80% of the deaths. The major constituents of PM2.5 are sea salt, dust, black carbon, sulphates, nitrates, organic carbon, and secondary organic aerosols, with latter five having grave impact on people\u2019s health. The major sources of PM2.5 are biomass burning (crop burning and forest fires), thermal power plants, steel and other such industries, and urban centers. Beyond this there are also factors such as transborder winds and meteorological phenomena such as temperature inversions that can impact pollution. It is critical to understand these a range of factors to understand and predict air pollution.<\/p>\n\n\n\n

We have been working on developing models to predict constituents of PM2.5 pollution at specific locations using available open sourced meteorological and pollution datasets. The model accounts for the various geographical, chemical, economic, and population level activities to identify the root causes of air pollution and predict possible pollution patterns. The aim of the project is to enable model and data-driven decisions to formulate more effective policy design to address air pollution. The model also allows for counterfactual reasoning to verify the effectiveness of policy decisions. The ability to provide casual evidence also can help declutter politics and lay bare the fallacies in policies and illustrate the required steps both long and short term to reduce air pollution.<\/p>\n\n\n\n

Through this project we hope to provide an effective tool to both policy makers and the public to address air pollution and work toward cleaner air and healthier citizens.<\/p>\n\n\n","protected":false},"excerpt":{"rendered":"

Through our drive-by air pollution sensing research project, we are enabling cost-efficient collection of granular spatiotemporal pollution data through drive-by sensing. <\/p>\n","protected":false},"featured_media":773758,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"footnotes":""},"research-area":[13556],"msr-locale":[268875],"msr-impact-theme":[261670],"msr-pillar":[],"class_list":["post-757810","msr-project","type-msr-project","status-publish","has-post-thumbnail","hentry","msr-research-area-artificial-intelligence","msr-locale-en_us","msr-archive-status-active"],"msr_project_start":"","related-publications":[700966,784450],"related-downloads":[785689],"related-videos":[],"related-groups":[],"related-events":[],"related-opportunities":[],"related-posts":[],"related-articles":[],"tab-content":[],"slides":[],"related-researchers":[{"type":"user_nicename","display_name":"Srinivasan Iyengar","user_id":41221,"people_section":"Researchers","alias":"sriyengar"},{"type":"user_nicename","display_name":"Mohit Jain","user_id":38769,"people_section":"Researchers","alias":"mohja"},{"type":"user_nicename","display_name":"Amit Sharma","user_id":30997,"people_section":"Researchers","alias":"amshar"},{"type":"user_nicename","display_name":"Manohar Swaminathan","user_id":35356,"people_section":"Researchers","alias":"swmanoh@microsoft.com"},{"type":"guest","display_name":"Dhruv Agarwal","user_id":791702,"people_section":"Research Fellow","alias":""}],"msr_research_lab":[199562],"msr_impact_theme":["Resilience"],"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/757810"}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-project"}],"version-history":[{"count":11,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/757810\/revisions"}],"predecessor-version":[{"id":818935,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/757810\/revisions\/818935"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media\/773758"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=757810"}],"wp:term":[{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=757810"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=757810"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=757810"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=757810"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}