About

Michael Howard Bergin is the Sternberg Family Professor of Civil & Environmental Engineering at Duke University. His research focuses on environmental and human health data analytics, low-cost air environmental sensors, air pollution, and renewable energy. He is involved in hands-on research experiences for undergraduate students and has contributed to understanding the discoloration of the Taj Mahal. As a faculty member at the Pratt School of Engineering, he is dedicated to advancing knowledge in civil and environmental engineering through innovative research and education.

Research topics

• Atmospheric sciences
• Environmental engineering
• Environmental science
• Meteorology
• Physics
• Geology
• Engineering
• Geography

Selected publications

• Detection and Quantification of Enteric Pathogens in Aerosols Near Open Wastewater Canals in Cities with Poor Sanitation

  UNC Libraries · 2026-04-08

  articleOpen access

  Urban sanitation infrastructure is inadequate in many low-income countries, leading to the presence of highly concentrated, uncontained fecal waste streams in densely populated areas. Combined with mechanisms of aerosolization, airborne transport of enteric microbes and their genetic material is possible in such settings but remains poorly characterized. We detected and quantified enteric pathogen-associated gene targets in aerosol samples near open wastewater canals (OWCs) or impacted (receiving sewage or wastewater) surface waters and control sites in La Paz, Bolivia; Kanpur, India; and Atlanta, USA, via multiplex reverse-transcription qPCR (37 targets) and ddPCR (13 targets). We detected a wide range of enteric targets, some not previously reported in extramural urban aerosols, with more frequent detections of all enteric targets at higher densities in La Paz and Kanpur near OWCs. We report density estimates ranging up to 4.7 &times; 10² gc per m_air³ across all targets including heat-stable enterotoxigenic <em>Escherichia coli</em>, <em>Campylobacter jejuni</em>, enteroinvasive <em>E. coli</em>/<em>Shigella</em> spp., <em>Salmonella</em> spp., norovirus, and <em>Cryptosporidium</em> spp. Estimated 25, 76, and 0% of samples containing positive pathogen detects were accompanied by culturable <em>E. coli</em> in La Paz, Kanpur, and Atlanta, respectively, suggesting potential for viability of enteric microbes at the point of sampling. Airborne transmission of enteric pathogens merits further investigation in cities with poor sanitation.

  PublisherDOI

• Fine particulate matter levels and HEPA filtration in Los Angeles homes during a wildland-urban-interface fire

  npj Clean Air · 2026-01-14 · 2 citations

  articleOpen access

  Wildland and wildland–urban-interface (WUI) fires substantially elevate fine particulate matter (PM2.5) concentrations in surrounding communities. Portable high-efficiency particulate air (HEPA) purifiers are widely recommended to reduce indoor PM2.5 exposure during such events, yet this guidance largely derives from studies of traffic exhaust, secondhand smoke, or indoor sources, rather than real-world WUI fire episodes. To address this gap, we leveraged data from ongoing randomized crossover trials of long-term use of portable HEPA purifiers in Los Angeles residences. During the Eaton Fire (January 2025), 11 homes were under HEPA intervention and 16 under non-HEPA control. Continuous indoor and outdoor PM2.5 monitoring before, during, and after the Eaton Fire (over 6 weeks) showed outdoor PM2.5 levels rose 148% (19 to 47 µg/m3) and indoor levels 91% (10 to 19 µg/m3) during the fire. HEPA homes had indoor PM2.5 levels 3 µg/m³ (15%, p = 0.01) lower than non-HEPA homes, while outdoor concentrations were comparable. Indoor PM2.5 reductions were unaffected by pre-filter use or clean air delivery rate (CADR)-to-room-volume ratio (0.8–4.9). These findings indicate that portable HEPA purifiers provided statistically significant but modest reductions in PM2.5 levels during WUI fire events. Complementary building-level and behavioral interventions remain critical to reduce indoor exposure in fire-affected communities.

  PublisherOA PDFDOI

• An Rf-Cnn Pipeline for Predicting Pm2.5 Concentration in Sri Lanka

  SSRN Electronic Journal · 2025-01-01

  preprintOpen access
  PublisherDOI

• Big, noisy data: how scalable Gaussian processes can leverage personal weather stations to improve spatiotemporal coverage of urban climate networks

  2025-05-21

  preprintOpen access

  Urban temperature varies dramatically across space and time, yet capturing this variability requires a dense, reliable sensor network—something that is rarely available in practice. Spatiotemporal gaps in data coverage make it difficult to connect localized urban heat stress to health outcomes and energy demand. In this work, we demonstrate how personal weather stations (PWSs) and machine learning can bridge these gaps to improve urban climate monitoring.To show this, we analyze PWS data collected in Durham County, North Carolina, from 2019 to 2024—a network of over 200 sensors recording hourly temperature data, totaling more than 15 million observations. This dataset presents two key sources of bias that must be addressed to ensure reliable urban heat estimates. First, it is preferentially sampled, with a higher density of weather stations in wealthier (and often cooler) neighborhoods. Second, faulty radiation shields on low-cost sensors may positively bias sensor measurements on sunny days.To address these challenges, we explore Gaussian Process Regression (GPR), a flexible machine learning technique that, when defined with a carefully designed covariance structure, can account for non-uniform sensor placement and measurement noise. However, exact GPR is computationally intractable for large spatiotemporal datasets (i.e., &gt; 10,000 observations). To overcome this, we leverage the Variational Nearest Neighbor Gaussian Process (VNNGP), a scalable approximation that enables the application of complex covariance structures to arbitrarily large datasets.Our approach demonstrates that the VNNGP model allows for complex spatiotemporal dependencies to be learned, making them well-suited for urban temperature modeling. Additionally, we show that abundant but noisy PWS data, when integrated with these models, can further improve spatial coverage. Together, these advancements highlight how combining large, imperfect datasets with sophisticated modeling techniques can enhance urban climate monitoring, leading to better heat exposure assessments and more informed environmental policies.

  PublisherDOI

• A paradigm shift: Low-cost sensors for effective air quality monitoring and management in developing countries

  Environment International · 2025-05-10 · 18 citations

  articleOpen access

  • Air quality monitoring challenges in Pakistan and South Asia were discussed. • Low-cost sensor integration with regulatory networks offers a viable air quality monitoring solution. • PM 2.5 pollution hotspots, including Faisalabad, were highlighted in Pakistan. • BlueSky Low-cost Sensor validated against the Beta Attenuation Monitor showed Normalized Root Mean Square Error and Normalized Mean Absolute Error values of 4% and 3%. • The successful use of LCS monitoring in Pakistan highlights its promising potential. This study gives an overview of the air quality monitoring challenges faced by South Asian countries, with a specific focus on Pakistan, and explores the potential application of low-cost sensors (LCS) to address these issues. Currently, 89% of the 4.2 million premature global deaths attributable to ambient air pollution occur in low- and middle-income countries, underscoring the urgent need for improved monitoring and abatement measures. In Pakistan, these challenges result in significant public health and economic consequences due to institutional and financial constraints, limited data availability, and transboundary pollution. The situation is exacerbated by the absence of an effective air quality regulatory network. This study proposes a shift by establishing a hybrid monitoring network that integrates conventional regulatory instruments with LCS. The current PM 2.5 pollution scenario in major Pakistani cities is analyzed using data from 61 LCSs installed nationwide, with one TSI BlueSky sensor validated against a regulatory BAM (Beta Attenuation Monitor) in Chakwal. Results show that the unadjusted BlueSky values demonstrate a strong correlation (Pearson correlation 0.85) with the reference BAM instrument, with acceptable error margins (NRMSE and NMAE of 8% and 6%, respectively), indicating the sensor’s potential for reliable monitoring. Though adjusted values aligned better, the analysis focuses on unadjusted data for broader applicability. These findings suggest that combining LCSs with advanced data analytics can serve as a technically and economically viable solution for accurate air quality monitoring and effective management in Pakistan and other developing countries facing similar high PM 2.5 concentrations. This paradigm shift enhances monitoring capabilities and supports informed policy-making and public health initiatives.

  PublisherDOI

• Sustaining Low-Cost PM2.5 Monitoring Networks in South-Asian Countries: Technical Challenges and Solutions

  Environment International · 2025-01-01

  articleOpen access

  Map Credits: AQLI, Google Maps 2024, and TSI Link, 2024. • LCS network is a viable solution for South-Asian countries. • LCS-related problems can be fixed with routine maintenance and debugging. • The lifetime of LCS can be extended up to many years with timely diagnostics and regular maintenance. The need to monitor South Asia’s air quality stems from its significant negative effects on human and environmental health. Traditional, regulatory-grade air quality monitoring systems have proven costly to operate and very difficult to maintain in most South-Asian countries. Low-cost sensor (LCS) networks have been touted as a viable alternative, but the challenges to sustain them have not been evaluated or thoroughly documented. The acceptance of such monitors, specifically by regulatory agencies, across South-Asian countries is still lacking. Lack of acceptance is due to prevailing myths (especially, in the regulatory circles of South-Asian countries) about their accuracy, precision, consistency, dependability, maintenance, and calibration concerns. The present study fills that knowledge gap through a systematic multi-country empirical analysis while also providing evidence-based solutions to enhance the longevity of LCS across diverse operational environments. Specifically, this study describes strategies and maintenance plans for operating large LCS networks of TSI BlueSky (8143) Sensors across several South-Asian countries, with a focus on problems caused by power outages, power surges, weather conditions, and continued exposure to high amounts of dust and pollution. The article provides further support that incorporating LCS networks into the regulatory framework can facilitate the enforcement of environmental regulations and legislation against polluters. The goal is to develop a more reliable and long-lasting air quality monitoring system that will assist regional environmental regulatory authorities in reducing air pollution-related health hazards and consequent socio-economic disruptions.

  PublisherDOI

• An RF-CNN pipeline for predicting PM2.5 concentration in Sri Lanka

  Journal of Hazardous Materials Advances · 2025-06-09 · 2 citations

  articleOpen access

  • The RF-CNN pipeline can serve as a robust and comprehensive method for precisely forecasting the spatial and temporal fluctuations in PM 2.5 concentrations. • RF-CNN pipeline yielded satisfactory outcomes by employing the RF component as an error model. • The RF-CNN pipeline is intended to enhance air quality forecasting and guide policymakers in mitigating air pollution impacts worldwide. Air pollution is a considerable global public health threat, requiring efficient monitoring and forecasting to guide decision-making. This study introduces a cascaded model of enhanced Random Forest with Convolutional Neural Network (RF-CNN) that predicts spatiotemporal fluctuations in PM 2.5 concentrations throughout Sri Lanka. The K-Nearest Neighbors method is employed to impute missing data, and the model utilizes data from 24 low-cost PM 2.5 sensors that are distributed throughout the country. The Convolutional Neural Network (CNN) derives spatial features from four-band PlanetScope satellite images (3m/pixel resolution, 1km 2 spatial coverage), while the Random Forest (RF) component models the relationship between PM 2.5 levels and four meteorological parameters. These features, combined with meteorological, spatial, and temporal inputs, produce the final forecasting results. The dataset comprises 1934 satellite images that were collected between December 2022 and February 2024, with an average PM 2.5 concentration of approximately 15 μg/m 3 . The RF-CNN model exhibited robust performance metrics across a variety of climate zones, including a normalized root mean square error of approximately 32.4%, a mean absolute percentage error of approximately 25.7%, a normalized mean absolute error of approximately 22.8%, a Spearman r of 0.871, and a Pearson r of 0.873. Two metrics: Input Data Quality Score (IDQS) and Testing Data Quality Score (TDQS) were implemented to evaluate the effects of imputation. Performance was minimally impacted by imputation within acceptable ranges, while exceeding limits resulted in increased uncertainty. This research emphasizes the efficacy of the RF-CNN approach, which integrates satellite imagery and low-cost sensor data, as a scalable solution for predicting spatiotemporal PM 2.5 variations. It provides valuable insights for regions that lack extensive monitoring.

  PublisherDOI

• Estimating the effects of vegetation and increased albedo on the urban heat island effect with spatial causal inference

  Scientific Reports · 2024-01-04 · 28 citations

  articleOpen access

  Abstract The urban heat island effect causes increased heat stress in urban areas. Cool roofs and urban greening have been promoted as mitigation strategies to reduce this effect. However, evaluating their efficacy remains a challenge, as potential temperature reductions depend on local characteristics. Existing methods to characterize their efficacy, such as computational fluid dynamics and urban canopy models, are computationally burdensome and require a high degree of expertise to employ. We propose a data-driven approach to overcome these hurdles, inspired by recent innovations in spatial causal inference. This approach allows for estimates of hypothetical interventions to reduce the urban heat island effect. We demonstrate this approach by modeling evening temperature in Durham, North Carolina, using readily retrieved air temperature, land cover, and satellite data. Hypothetical interventions such as lining streets with trees, cool roofs, and changing parking lots to green space are estimated to decrease evening temperatures by a maximum of 0.7–0.9 $$^{\circ } \hbox {C}$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mrow> <mml:msup> <mml:mrow/> <mml:mo>∘</mml:mo> </mml:msup> <mml:mtext>C</mml:mtext> </mml:mrow> </mml:math> , with reduced effects on temperature as a function of distance from the intervention. Because of the ease of data access, this approach may be applied to other cities in the U.S. to help them come up with city-specific solutions for reducing urban heat stress.

  PublisherOA PDFDOI

• Refining Citizen Climate Science: Addressing Preferential Sampling for Improved Estimates of Urban Heat

  Environmental Science & Technology Letters · 2024-06-17 · 5 citations

  article

  Studies of urban heat are often limited by their ability to measure air temperature; data are collected either at a few locations over time or at many locations at one point in time. Citizen science approaches to observing temperature provide a way to overcome these limitations, by capturing data over long time scales, at many locations. However, citizen scientists are more likely to be wealthier, making certain neighborhoods better observed than others. Because urban heat islands are more prevalent in poorer neighborhoods, heat extremes are less likely to be observed by citizen scientists. In spatial statistics, this is known as preferential sampling. When we adjust citizen science data for this effect, we obtain results that better agree with NOAA’s urban heat island data, which are not preferentially sampled. Using this adjustment, estimates of the July 2021 average evening temperature are almost 1 °C warmer in unobserved neighborhoods in Durham, North Carolina, than if they were not adjusted. We demonstrate that adjusted citizen science data allow for better characterization of heat risk at any time of interest and may be used for almost any neighborhood in the United States.

  PublisherDOI

• Development and Optimization of the Low-Cost Optical Monitor for Real-Time Monitoring of Atmospheric Black and Brown Carbon

  SSRN Electronic Journal · 2024-01-01

  preprintOpen access
  PublisherDOI

Recent grants

• Characterization of the Physical and Chemical Properties of Water Insoluble Atmospheric Aerosol

  NSF · $274k · 2001–2006

• Collaborative research: Spatial and temporal variability of surface albedo and light absorbing chemical species in Greenland.

  NSF · $46k · 2015–2015

• Collaborative Research: Direct radiative forcing over central Greenland; Assessment of the coupled effect of light absorbing aerosols and snow albedo variability

  NSF · $210k · 2010–2013

Frequent coauthors

• Michael Valerino

  Pratt Institute

  18 shared

• Lucas Rocha-Melogno

  Meridian Institute

  14 shared

• Marcos Andrade

  12 shared

• Marc A. Deshusses

  Duke University

  12 shared

• James J. Schauer

  11 shared

• Roby Greenwald

  Georgia State University

  10 shared

• Jack E. Dibb

  10 shared

• Aniket Ratnaparkhi

  Indian Institute of Technology Gandhinagar

  10 shared

Labs

• Bergin GroupPI

  List of current and past lab members in the Bergin Group in the Department of Civil & Environmental Engineering at Duke University

Education

• PhD, CEE

  Carnegie Mellon University

  1995