NSF-IRES: U.S.-Bangladesh-Canada collaboration to improve air quality

Overview

This project will provide U.S. undergraduate and graduate students a ten-week summer opportunity to gain international research experience in emerging air quality issues in the megacity of Dhaka, Bangladesh and Montreal, Canada. 

The project will take place in collaboration with students and faculty at the University at Albany, Stanford University, the Bangladesh University of Engineering and Technology (BUET), the University of Dhaka, Concordia University and McGill University.

Students will get hands-on experience on low-cost sensors and remote sensing technologies to advance their understanding of air quality problems and potential solutions. Students will work on individual research projects and have unique opportunities to enhance their professional growth, peer-mentoring skills, networking as they develop their competencies in intercultural communication, self-awareness, professional adaptability and leadership in a culturally diverse group.

The experiences and lessons learned from a megacity’s impact on air quality can be valuable for the IRES students to study large U.S. urban centers confronting similar air pollution challenges.
 

Program structure

The project will train 24 U.S. students under a ten-week summer program over the 3-year period. Students will be recruited during the Fall semester. Each year a cohort of 8 students (5 undergraduates and 3 graduate) will participate in four sets of activities:

• pre-departure preparation during the Spring  (virtual) for research meetings and language training
• pre-departure virtual training activities (2 weeks)
• 7 weeks research activities in the IRES site
• 1 week (in-person) post-trip follow-up activities at UAlbany

Key activities include a streamlined two-week air quality short course, travel workshop and research ethics compliance training during the 2-week pre-departure virtual training. The pre-departure training during the Spring will include bi-weekly research meetings to introduce students to low-cost sensor technologies, performance evaluation and air quality sensor data analysis, as well as weekly meetings for culture and language instruction. The 1-week post-trip follow-up will include report writing, presentation of their research findings and international experiences and field trips.
 

Financial Support

• Stipend: $3,500 for undergraduates and $4,200 for graduate students
• Travel: support for international and local travels
• Subsistence: Housing and meals provided

Dhaka, Bangladesh

Bangladesh is known for its vulnerability to climate change, consistently ranked as one of the most polluted countries. With more than 20 million people, Dhaka, the capital of Bangladesh is an example of the most polluted megacities in the world.

The atmospheric and oceanic sciences department at McGill university is one of the top atmospheric science research departments in Canada and has a worldwide reputation. With thirteen faculty members the department is a diverse and vibrant group, with wide ranging research interests.

Recently a number of faculty members within the department have started various projects around wildfires meteorology and smoke transport, so the student would be able to attend seminars, group meetings, and discussions with students and faculty working on similar problems.

The department also has strong ties to several other research institutions nearby, including the Canadian Meteorological Center and the Ouranos regional climate group. 
 

Eligibility

This opportunity is available to undergraduate and graduate students in STEM fields at U.S. institutions with following requirements:

1. U.S.-citizen or permanent residents
2. Grade point average (GPA) of 2.7 for undergraduate and 3.0 for graduate applicants
3. Must commit to being in Canada for 7 weeks during summer 2025
4. Available to attend Spring 2025 (virtual) for pre-departure preparation (research meetings and language instructions)
5. Agree to participate in a follow-up research activity for Fall 2025

Preference will be given to juniors/seniors and graduate students major in environmental engineering or a related discipline (e.g., atmospheric and environmental sciences, chemical engineering and mechanical engineering).
 

How to Apply

1. Complete an online application form.
   
   Undergraduate Application
   
   Graduate Application
    
2. Submit copies of transcripts (a copy of unofficial transcript is fine).
3. Provide one reference (e.g., your academic advisor). If shortlisted, we will contact for recommendation letter.
4. Upload your resume.
5. Provide a statement of purpose. Describe your academic and career goals, research interests, and how you believe the IRES program will help achieve your short- and long-term goals.
6. Provide demographic information

IRES participants will work on research projects collaborating with the foreign mentors and students. The Principal Investigator and foreign mentors will offer numerous research topics at undergraduate and graduate levels. This helps students choose their individual research topics based on their interests. The nature of proposed research activities falls under a unifying research topic i.e., improving air quality in a megacity by leveraging recent advances in low-cost sensor and remote sensing technologies. The projects will be structured in such a way that both undergraduate and graduate students can feel confident that they can complete the proposed tasks successfully in due time. A brief description of sample research projects is given below:

• Sample project 1: Development of a calibration framework for low-cost PM2.5 monitoring system in a megacity of Bangladesh. 
  While low-cost sensor-based air quality monitoring shows great potential for spatially dense and community-scale air monitoring with a reasonable cost, their calibration and data inversion are the most critical aspect. The response of the low-cost sensors is highly sensitive to ambient temperature, humidity, pollutants mixture. High pollution loadings, local weather, and infrastructure conditions in polluted cities in developing countries present unique challenges for low-cost sensors calibration, deployment, and maintenance. The proposed work will develop calibration models and a standard operational procedure for low-cost PM2.5 sensors under a megacity-specific conditions.
   
• Sample project 2: Coupling low-cost air sensors and satellite data for mapping brick kiln pollution in Bangladesh. 
  Bangladesh is the fourth largest brick producer globally with about 7000 brick kilns and produces about 23 billion bricks annually. Traditional energy-inefficient brick manufacturing kilns in Bangladesh and across South Asia kill thousands of people every year. Accurate mapping of the pollution scenario is the first step for developing science-based solutions and tools for addressing the brick kiln pollution in Bangladesh. We propose to combine low-cost fine particulate mass sensor and satellite-based observations for mapping surface PM concentrations in the brick-kiln pollution impacted areas in Bangladesh.
   
• Sample project 3: Developing a low-cost approach for evaluating kiln level emissions in Bangladesh. 
  Each year the global radiative forcing generated by the black carbon and greenhouse gases emanating from brick kilns in South Asia is equivalent to the climate impact generated from the entire US passenger car fleet. Recent estimates suggest that air pollution generated by brick kilns results in over 60,000 premature adult deaths annually in South Asia. We propose a set of iterative field measurements collaborating with Stanford University researchers to work towards developing a practical low-cost approach to evaluating pollution generated by individuals and groups of kilns.
  Saving Lives and Combatting Climate Change - YouTube
   
• Sample project 4: Air pollution mapping in Dhaka City using low-cost sensors. 
  Air pollutants in urban areas vary spatially and temporally. Numerous studies have identified that a central monitoring approach cannot capture the pollution gradient in urban areas and pollution concentrations can vary substantially within a few hundred meters. In recent years, the mobile and distributed monitoring with data science and machine learning approaches have been shown a great promise for high spatial resolution urban air pollution monitoring and exposure estimates. We will develop an empirical model and high spatial resolution air pollution mapping using lower cost sensor measurements and advanced statistical modeling approach.
   
• Sample project 5: Understanding residential indoor air quality exposure in Dhaka using multipollutant sensors. 
  Climate change can affect the air we breathe in both ambient and indoor environments. While government and regulatory agencies have focused to tackle urban ambient air pollution, little attention has been paid to assess the quality of air in our homes. To fill this gap, IRES participants will measure air pollutants in residential homes to get experience in assessing indoor air quality in a polluted megacity of Dhaka. Several criteria air pollutants e.g., fine particulate matter (PM2.5), carbon monoxide (CO) and climate-driven pollutants e.g., black carbon and caron dioxide (CO2) will be measured using low-cost sensors.
   
• Sample project 6: Monitoring exposure to air pollutants and greenhouse gases in public transportation in Dhaka city using low-cost sensors. 
  Traffic jam is a known issue in Dhaka city due to the lack of proper planning, lack of road as well as public bus, poor traffic managements, old vehicles, reckless driving, and less skilled driver. In Dhaka city, public bus is a crucial means of transport and passengers are espousing huge pollution due to the long hours staying inside the public bus due to traffic jam. We will focus to quantify the exposure of particulate matter and trace gases of the passengers in the public bus for the first time in Dhaka, Bangladesh. We will use multipollutant sensor monitors for measuring particulate fractions (PM1, PM2.5 and PM10), gaseous pollutants (CO, NO2, O3), and greenhouse gases (CO2, CH4).
   
• Sample project 7: Evaluation of satellite-based models in estimating the surface PM2.5 over Bangladesh. 
  For assessing PM2.5 exposure and associated health impacts, there is a need to estimate accurate surface PM2.5 concentrations. Many studies have been applied machine learning (ML)-based models to estimate surface PM2.5 concentration by using satellite-retrieved aerosol optical depth (AOD) and ground-based measured PM2.5 concentration. However, there are some uncertainties associated with ML-based models leading to potential bias in estimating surface PM2.5. IRES students will apply ML-based models such as artificial neural network (ANN) and random forest to evaluate the accuracy of estimated PM2.5 concentration in Dhaka using hourly satellite-retrieved AOD data and ground-based measured PM2.5 concentrations.
   
• Sample project 8: Estimation and characterization of waste burning impact in Dhaka, Bangladesh using low-cost sensors. 
  Waste burning (e.g., garbage, agricultural residue) is one of the neglected sources of air pollution in many countries of the world including Bangladesh. Varieties of waste burning are happening in Dhaka city, e.g., open garbage burning, municipal solid waste burning, dry leaves burning, and garments cotton burning, and agricultural residue burning. IRES participants will study the particulate, gaseous and greenhouse gas emissions from waste burning in Dhaka using low-cost multipollutant sensor monitors.
   
• Sample project 9: Understanding indoor air quality in Montreal, Canada 
  (Mentors: Fariborz Haghighat, Zhi Chen, Md. Aynul Bari)
  The proposed IRES project will conduct indoor air quality monitoring using low-cost PurpleAir sensors and measure PM fractions (PM1/PM2.5) in at least 20 homes in Montreal. An intervention measure will be also done, where half of the home participants (n=10) use air purifiers to reduce exposure. The study will also leverage UAlbany’s indoor air quality study and use low-cost PurpleAir sensor data from at least 10 homes to compare impacts on indoor air quality in both Canadian and U.S. homes. IRES students will also compare with PurpleAir sensor data measured in Dhaka homes during 2023-3024 to find differences in indoor PM2.5 expsure between a developing and a developed country.
   
• Sample project 10: Understanding impacts of wildfire smoke events on ambient air quality in Canada and the United States 
  (Mentors: Zhi Chen, Md. Aynul Bari)
  In Canada and the U.S., community air quality observation networks are limited in characterizing the diverse group of air pollutants that affect exposure across regional to neighborhood scales. To increase public awareness and address any smoke-related air quality issues (e.g., wildfires), low-cost air quality sensors will be deployed over one year period in at least 10 fixed ambient sampling sites in Montreal to measure particulate matter fractions (PM1/PM2.5/PM10). The study will leverage U.S.EPA’s enhanced air quality monitoring study for communities (PI: Bari), where UAlbany is developing an enhanced air quality monitoring network from Summer 2025 in and near underserved neighborhoods in NYS. IRES students can use sensor data from 15 fixed ambient sites in NYS and compare air quality and impacts of wildfires events on air quality in both the Canada and U.S. neighborhoods.
   
• Sample project 11: Performance of commercial air purifiers for reducing wildfires impact on indoor air quality 
  (Mentor: Fariborz Haghighat)
  A purifier system proposed by this project can be installed in the indoor air quality Lab to examine the performance of the purifier. This project can examine different scenarios under different building and outdoor meteorological conditions to possibly see how the wildfires will affect the indoor air quality.
   
• Sample project 12: Connecting high altitude transport to surface conditions: the meteorological conditions for downward intrusion of aloft smoke plumes 
  (Mentors: Robert Fajbar, Cheng-Hsuan Lu)
  To better understand the meteorological conditions for the downwards intrusion, we will use the Hysplit model to simulate large quantities of virtual Lagrangian particles in areas and seasons that are prone to wildfires, using the Calcul Quebec supercomputing network. These trajectories will then be analyzed to find particles which are mixed downwards into the boundary layer, and then atmospheric reanalysis data will be composited around the times and locations that particles are being mixed downwards to create meteorological composites of the conditions responsible for the downwards intrusions of smoke plumes.
   
• Sample project 13: Development of trustworthy data collection from low-cost sensor-based air quality monitoring tools 
  (Mentor: Suryadipta Majumdar)
  The recent studies show that data integrity is often threatened in low-cost sensor-based solutions. Specifically, the first category of threats is introduced by spoofing attacks where adversaries (by exploiting vulnerabilities in devices, software or communication channels) modify the actual sensor readings. The second category of threats is introduced by masking attacks where adversaries (by exploiting vulnerabilities in devices, software or communication channels) modify drastic pattern changes in sensor readings to hide any specific events (e.g., drastic drop in the air quality). 
  The final category of threats is introduced by faulty devices where adversaries exploit faulty devices that misreport the air quality metrics. There exist a few recent works (e.g., Wei et al., 2023) to detect anomalies in air sensor data. However, none of them study the impact of data integrity and trustworthiness on air quality monitoring tools and address the above-mentioned challenges. In this project, we will first study the impact of wrong data on wild-fire effects on air quality of this region. 
  Second, we will identify/learn the features of sensor data impacting the quality of study. Third, we will explore the applicability of existing security solutions for other sensor-based environments (e.g., smart home) into air quality monitoring tools. On a successful completion of this project, we will expect the following outcomes: (i) finding on the importance of data integrity and measure the tolerance, (ii) the features in sensor data with their relative ranking, and (iii) mechanisms to build trustworthiness.
   
• Sample project 14: Developing resilience in air quality monitoring tools against AI-enhanced adversaries 
  (Mentor: Suryadipta Majumdar)
  With the wide-spread adoption of artificial intelligence (AI) in various applications including analyzing air quality, this project focuses on AI’s security impacts on such tools. Specifically, this project aims at studying the adversarial effects of AI on air quality monitoring tools as well as building resilient techniques for these tools against AI-enhanced adversaries. The recent studies report the imminent threats from both passive (i.e., typical ML issues - lack of training data, labeled data, data quality) and active (i.e., against outliers and data poisoning attacks) challenges. 
  To understand the landscape of AI-based threats, we will first analyze the adversarial machine learning threats, which are applicable to air quality monitoring tools. During this study, we will mainly identify the characteristics, objectives, and requirements of AI-based threats and classify both adversaries and defenders/users based on these findings. We will also define the scope of such AI-based threats, the capabilities of both adversaries and defenders, and other assumptions currently utilized in this domain. The outcome of this research activity will provide a threat model for AI-based adversaries. 
  To detect existing adversarial examples that might have manipulated the learning process of air quality monitoring tools, we will first collect system data and train different models to capture the normal behavior of a system. We will then develop a technique that will detect any anomalous patterns in that model as adversarial examples. The outcome of this research activity will provide a set of existing adversarial examples in air quality monitoring.
   

For questions or any information, please email [email protected].
 

Mentors
 

Md. Aynul Bari
Assistant Professor and Principal Investigator
Department of Environmental and Sustainable Engineering
College of Engineering and Applied Sciences
University at Albany, State University of New York

Abdus Salam
Professor
Department of Chemistry
University of Dhaka

2024

Ishe

Ishe is a junior at Central State University who majors in Environmental Engineering. Ishe was tasked with monitoring air quality in 10 schools across the megacity of Dhaka in Bangladesh for the duration of his IRES research project. Due to his experiences on his first internship and research project, Ishe is interested in continuing air quality research and all the different avenues that this field holds. Ishe is a member of the National Society of Black Engineers which he has been a member of for two years now. In Ishe’s free time he likes to go out on hikes around Ohio, spend time with his family or just watch a soccer game during the weekends.