Associate Research Scientist · Northeastern University

Khanh Do

Computational atmospheric scientist building the models that tell us what we breathe, and who breathes the worst of it.

I fuse machine learning, GPU computing, and deterministic chemical transport models to make air quality prediction faster, sharper, and more equitable: from CUDA kernels inside CMAQ to deep-learning climate downscaling, hyperlocal mobile sensing, and cumulative environmental justice scoring for Greater Boston.

Download CV Browse 20 publications Get in touch
0
Publications
& manuals
0
Conference
presentations
0
Ph.D. students
mentored
0
Supercomputing
CPU hours awarded
01 / About

Large datasets, hard physics, and the people downstream

My research focuses on predicting and investigating air quality using machine learning methods and deterministic models.

I work with large datasets to recognize patterns and explore the relationships between meteorology, emissions, and exposure, then apply machine learning and technological advances such as GPU computing to make deterministic models dramatically more efficient. The through-line: a model is only as valuable as the decision it enables, and the community it protects.

That has meant replacing the slowest module of a regulatory chemical transport model with CUDA kernels, downscaling coarse global climate simulations to hyper-local forecasts with deep learning, driving a mobile laboratory across the coastal-urban interface of Greater Boston, and building a cumulative environmental justice score system for disadvantaged neighborhoods in that same city.

Air quality & modelingMachine learningComputer visionPersonal exposureData assimilationHigh-performance computingGPU computingClimateModel developmentDownscalingBias correction
02 / Research

Five programs, one question: how do we predict air quality well enough to act on it?

01
NSF CDS&ENOAADOEiSUPER

Chemical transport model enhancement

Making regulatory-grade atmospheric chemistry fast enough to be useful.

  • Implemented GPU computing in the CMAQ gas solver, migrating CMAQ's Rosenbrock integrator from Fortran to CUDA Fortran. The resulting CMAQ-CUDA completes a chemistry time step in 35% to 51% of CMAQv5.4's time across the RACM2, CB6R5, and SAPRC07 mechanisms.
  • Used machine learning to accelerate chemical transport models by substituting their most time-consuming modules, and to bias-correct their output for higher accuracy.
  • Contributed to development of the early version of NOAA's UFS-SRW App, and evaluated UFS-AQM forecasting biases during intense wildfire events.

Published in ACS ES&T Air · presented at CMAS 2022–2024 and IWAQFR 2025

02
iSUPERSloan FoundationCARB

Environmental justice & health assessment

Quantifying who carries the pollution burden, and what would relieve it.

  • Built a new cumulative environmental justice score system for Greater Boston from 59 low-cost sensors and nine reference monitors, scored on median household income, share of minority residents, and share of limited-English-speaking households. Chelsea and Everett rank among the most burdened.
  • Linked 2023 CDC health surveys to criteria pollutants and socioeconomic status at census-tract resolution nationwide: risk climbs sharply once PM2.5 passes 14 µg/m³ and ozone passes 40 ppb, and residents within 5 km of oil and gas point sources carry measurably worse outcomes.
  • Investigated personal PM2.5 exposure in the Inland Empire, explaining how habits, income, and communities of color shape exposure at personal resolution.
  • Quantified how mobility and housing characteristics drive daily PM2.5 exposure and well-being in West San Bernardino, delivering individualized resilience plans to community collaborators.

Published in Environmental Pollution, Environmental Research Letters, Journal of Aerosol Science

03
iSUPERSCAQMD

Machine-learning-assisted air quality prediction

Better predictions from fewer sensors and messier data.

  • Optimized low-cost sensor placement: minimizing the number of sensors needed to cover the most land area while maximizing ML prediction accuracy.
  • Determined the meteorological drivers of ground-level ozone and PM2.5, building empirical models from historical records to project air quality trends.
  • Assembled and operated BAM 1020 reference monitors and ran collocation calibration for low-cost sensor networks.

Published in Environmental Science: Atmospheres, Atmospheric Environment, GMD

04
NCAR/NWSCDeep learning

High-resolution climate downscaling

Thirty years of future climate, resolved down to the neighborhood.

  • Dynamically downscaled 2060–2089 climate projections with the Pseudo Global Warming method and WRF across Shared Socio-Economic Pathways.
  • Delivered nested meteorological products at 36 km (CONUS), 6 km (Eastern US), and 1 km (North Carolina & Virginia), feeding downstream health assessment.
  • Projected daily maximum 2 m temperature rising 1.6 °C to over 4.4 °C by 2060–2089 across SSP1-2.6, SSP2-4.5, and SSP5-8.5, with “Danger” and “Extreme Danger” heat days emerging under SSP5-8.5. Separated two distinct heat mechanisms: weak synoptic anomalies versus land-atmosphere coupling driven by soil moisture depletion.
  • Developing a self-supervised deep learning downscaling approach that turns coarse global climate simulations into hyper-local forecasts.

Published in Weather and Climate Extremes · submitted to Earth's Future · 500,000 CPU + 1,500 GPU hours, NCAR-Wyoming Supercomputing Center

05
iSUPERNew

Mobile measurements

Taking the instruments to the street, at the scale people actually breathe.

  • Mobile laboratory mapping of fine-scale air quality and greenhouse gas gradients across the coastal-urban interface of Greater Boston.
  • Hyperlocal mobile sensing to map air pollution concentrations across Greater Boston at block-by-block resolution.
  • Recognized with the iSUPER Mobile Air Monitoring Field Research Contributions award (2026).

Presented at AGU 2026 and AAAR 2025

Toolbelt

Models

CMAQ · WRF · WRF-Chem · UFS-SRW · UFS-AQM · RRFS-CMAQ · MPAS

Methods

Deep learning & CNNs · Self-supervised downscaling · Physics-guided ML · Generalized additive models · Bias correction · Data assimilation · Spatial interpolation · Pseudo global warming

Computing

CUDA / GPU acceleration · MPI & HPC workflows · Cheyenne · Derecho · NERSC · Anvil · Discovery

Measurement

Mobile laboratory & hyperlocal sensing · BAM 1020 · Low-cost sensor networks & collocation calibration · UAV vertical profiling · Personal exposure monitoring

Vision & other

GPU-assisted image processing for high-resolution traffic footage · Vehicular detection, counting & classification · Vehicular emission detection · Atmospheric dispersion modeling · M/M/1 & M/G/1 queueing analysis

03 / Publications

20 publications & training manuals

Eight as first author. Plus three technical reports for NOAA, the National Center for Sustainable Transportation, and SCAQMD.

  1. 2026

    Shuhua Lu, Do K, Yang Zhang, Xiaodong Chen, Ruby Leung, Michelle L. Bell. “High-Resolution Projections of Extreme Heat and Thermal Stress in Southeastern U.S.”

    Weather and Climate Extremes 52, 100878

    10.1016/j.wace.2026.100878 ↗
  2. 2026

    Do K, Shuhua Lu, Yang Zhang. “From Coarse Global Climate Simulations to Hyper-Local Climate Forecasts: A Deep Learning Downscaling Approach.”

    Submitted to Earth's Future

  3. 2025

    Do K, Yang Zhang. “Assessing Health Risks and Disparities in Air Pollution and Socioeconomic Status across the United States.”

    Environmental Pollution

    10.1016/j.envpol.2025.126311 ↗
  4. 2025

    Do K, Yang Zhang, Siqi Ma, Daniel Tong. “Assessing Air Pollution Exposure Disparities in Disadvantaged Communities of Greater Boston: A New Cumulative Environmental Justice Score System.”

    Environmental Research Letters

    10.1088/1748-9326/adb16c ↗
  5. 2025

    Duncan Quevedo, Do K, Jose Rodriguez Borbon, Bryan M. Wong, Cesunica E. Ivey. “GPU Implementation of a Gas-Phase Chemistry Solver in the CMAQ Chemical Transport Model.”

    ACS ES&T Air

    10.1021/acsestair.4c00181 ↗
  6. 2024

    Do K, George Delic, Jose Rodriguez Borbon, Duncan Quevedo, Bryan M. Wong, Cesunica E. Ivey. “Graphics Processing Unit Assisted Computation for a Gas-Phase Chemical Solver in a Regulatory Chemical Transport Model.”

    EarthArXiv preprint

    eartharxiv.org/repository/view/6323 ↗
  7. 2024

    Torres I, Do K, Delgado A, Mourad C, Yu H, Ivey CE. “Indoor and Ambient Influences on PM2.5 Exposure and Well-being for a Rail Impacted Community and Implications for Personal Protections.”

    Environmental Research Letters

    10.1088/1748-9326/ad90f5 ↗
  8. 2024

    Do K, Daniel Schuch, Yang Zhang. “Practice Instruction for Online-Coupled WRF-Chem Version 4.6.0: Training Manual for the Hands-on Training Session on WRF-Chem.”

    World Meteorological Organization · E-learning & Training Resources Portal

    etrp.wmo.int · training manual (PDF) ↗
  9. 2024

    Zihan Zhu, Do K, Cesunica E. Ivey, Don R. Collins. “Assessing CMAQ Model Discrepancies in a Heavily Polluted Air Basin Using UAV Vertical Profiles and Sensitivity Analyses.”

    Environmental Science: Atmospheres

    10.1039/d4ea00004h ↗
  10. 2024

    Do K, Yeganeh AK, Gao Z, Ivey CE. “Performance of Machine Learning for Ozone Modeling in Southern California during the COVID-19 Shutdown.”

    Environmental Science: Atmospheres

    10.1039/D3EA00159H ↗
  11. 2024

    Ziqi Gao, Do K, Zongrun Li, Xiangyu Jiang, Kamal J. Maji, Cesunica E. Ivey, Armistead G. Russell. “Predicting PM2.5 Levels and Exceedance Days Using Machine Learning Methods.”

    Atmospheric Environment

    10.1016/j.atmosenv.2024.120396 ↗
  12. 2024

    Duncan Quevedo, Ziqi Gao, Do K, Roya Bahreini, Don Collins, Cesunica Ivey. “Multidecadal Analysis of Meteorological and Emissions Regimes for PM2.5 Across California.”

    ACS ES&T Air

    10.1021/acsestair.3c00019 ↗
  13. 2023

    Ziqi Gao, Cesunica E. Ivey, Charles L. Blanchard, Do K, Sang-Mi Lee, Armistead G. Russell. “Emissions and Meteorological Impacts on PM2.5 Species Concentrations in Southern California Using Generalized Additive Modeling.”

    Science of the Total Environment

    10.2139/ssrn.4377297 ↗
  14. 2023

    Ziqi Gao, Cesunica E. Ivey, Charles L. Blanchard, Do K, Sang-Mi Lee, Armistead G. Russell. “Emissions, Meteorological and Climate Impacts on PM2.5 Levels in Southern California Using a Generalized Additive Model: Historic Trends and Future Estimates.”

    Chemosphere

    10.1016/j.chemosphere.2023.138385 ↗
  15. 2023

    Do K, Mahish M, Kashfi Yeganeh A, Gao Z, Blanchard CL, Ivey CE. “A Machine Learning Approach to Quantify the Impact of Meteorology on Tropospheric Ozone in the Inland Empire, CA.”

    Environmental Science: Atmospheres

    10.1039/D2EA00077F ↗
  16. 2023

    Md Hasibul Hasan, Haofei Yu, Cesunica Ivey, Ajay Pillarisetti, Ziyang Yuan, Do K, Yi Li. “Unexpected Performance Improvements of Nitrogen Dioxide and Ozone Sensors by Including Carbon Monoxide Sensor Signal.”

    ACS Omega

    10.1021/acsomega.2c07734 ↗
  17. 2022

    Gao Z, Ivey CE, Blanchard CL, Do K, Lee S-M, Russell AG. “Separating Emissions and Meteorological Impacts on Peak Ozone Concentrations in Southern California Using Generalized Additive Modeling.”

    Environmental Pollution, 119503

    10.1016/j.envpol.2022.119503 ↗
  18. 2022

    Ziqi Gao, Yifeng Wang, Petros Vasilakos, Cesunica E. Ivey, Do K, Armistead G. Russell. “Predicting Peak Daily Maximum 8-h Ozone and Linkages to Emissions and Meteorology in Southern California Using Machine Learning Methods.”

    Geoscientific Model Development

    10.5194/gmd-15-9015-2022 ↗
  19. 2021

    Do K, Yu H, Velasquez J, Grell-Brisk M, Smith H, Ivey CE. “A Data-Driven Approach for Characterizing Community-Scale Air Pollution Exposure Disparities in Inland Southern California.”

    Journal of Aerosol Science, Article 105704

    10.1016/j.jaerosci.2020.105704 ↗
  20. 2020

    Ivey CE, Gao Z, Do K, Kashfi Yeganeh A, Russell AG, Blanchard CL, Lee S-M. “Impacts of the 2020 COVID-19 Shutdown Measures on Ozone Production in the Los Angeles Basin.”

    ChemRxiv preprint

    10.26434/chemrxiv.12805367.v1 ↗

Technical reports

  • Do K, Yang Zhang. “NOAA Semi-annual Report for UFS-SRW App.” Grant No. NA22OAR4590515.
  • Ivey CE, Nguyen A, Xu RM, Do K, Hao P, Barth M. “Hyperlocal Monitoring of Traffic-Related Air Pollution to Assess Near-Term Impacts of Sustainable Transportation Interventions.” National Center for Sustainable Transportation, Contract No. DOT 69A3551747114, Feb 2023.
  • Ivey CE, Blanchard CL, Russell AG, Gao Z, Do K. “Ozone Meteorology Study, final report.” South Coast Air Quality Management District, Contract No. 20058, January 2022.

Published datasets

Global 5 km daily minimum, mean, and maximum bias-corrected 2 m temperature, released on Zenodo as sole author. Each link is a concept DOI and always resolves to the newest version.

04 / Talks & teaching abroad

34 conference presentations across CMAS, AGU, AAAR, AMS, AEESP & more

WMO

Instructor: “Training Course on Seamless Prediction of Air Pollution in Africa”

Delivered a specialized training course at the Egyptian Meteorological Authority in Egypt, organized by the World Meteorological Organization, instructing participants on running and evaluating the WRF-Chem model over the African domain.

As presenter (15)

  • Poster

    “A Physics-Based Deep Learning Method for High-Resolution Predictions of PM2.5 and O3”

    Gordon Research Seminar · Newry, ME · Aug 2025
  • Poster

    “A Physics-Based Deep Learning Method for High-Resolution Predictions of PM2.5 and O3”

    Gordon Research Conference · Newry, ME · Aug 2025
  • Oral

    “Examining the UFS-AQM's Forecasting Biases during An Intense Wildfire Event”

    11th International Workshop on Air Quality Forecasting Research · Montreal, QC · Jun 2025
  • Poster

    “Evaluation of UFS-AQM During Summer 2023: Impacts of Updated Emissions”

    11th International Workshop on Air Quality Forecasting Research · Montreal, QC · Jun 2025
  • Oral

    “Accelerating High-Resolution Downscaling of Meteorological Variables via Self-supervised Deep Learning”

    Joint MPAS/WRF Users Workshop 2025 · Boulder, CO · Jun 2025
  • Poster

    “Assessing Air Pollution Exposure Disparities in Disadvantaged Communities of Greater Boston: A New Cumulative Environmental Justice Score System”

    HPC DAY 2025 · Amherst, MA · Apr 2025
  • Poster

    “Assessing Health Risks and Disparities in Air Pollution and Socioeconomic Status across the United States”

    23rd Annual CMAS Conference · Chapel Hill, NC · Oct 2024
  • Poster

    “Quantify the Impacts of Air Pollution on Disadvantaged Communities in Greater Boston with a New Cumulative Environmental Justice Score System”

    BARI 2024 · Boston, MA · Apr 2024
  • Oral

    “GPU-Assisted Computation for a Gas-Phase Chemical Solver in CMAQ”

    22nd Annual CMAS Conference · Chapel Hill, NC · Oct 2023
  • Oral

    “Machine Learning with Spatial Interpolation Techniques for Constructing 2-Dimensional Ozone Concentrations in Southern California during the COVID-19 Shutdown”

    22nd Annual CMAS Conference · Chapel Hill, NC · Oct 2023
  • Poster

    “Superior Performance of Convolutional Neural Network for Predicting PM2.5 Concentrations and Exceedances under Sparse Data Availability”

    AEESP 2023 · Boston, MA · Jun 2023
  • Oral

    “A Machine Learning Approach to Quantify the Impact of Meteorology on Tropospheric Ozone in the Inland Empire, CA”

    21st Annual CMAS Conference · Chapel Hill, NC · Oct 2022
  • Oral

    “GPU-Assisted CMAQ Simulations”

    21st Annual CMAS Conference · Chapel Hill, NC · Oct 2022
  • Poster

    “Applying Machine Learning and Chemical Transport Model to Investigate the Influences of Meteorology on PM2.5 and Ozone”

    38th AAAR Meeting · Oct 2020
  • Poster

    “High-Resolution Personal Exposure Monitoring of PM2.5 in Inland Southern California”

    37th AAAR Meeting · Portland, OR · Oct 2019

As co-author (19)

  • Upcoming

    “Development of a Hybrid Physics-Guided Deep Learning Air Quality Forecasting System Over Greater Boston”

    Oral · AMS 2027 · Denver, CO · Jan 2027
  • Upcoming

    “Mobile Laboratory Mapping Fine-Scale Air Quality and Greenhouse Gas Gradients across the Coastal-Urban Interface of Greater Boston”

    Oral · AGU 2026 · San Francisco, CA · Dec 2026

  • “Improving Real-Time Air Quality Forecasting through Bias-Correction and Machine Learning over Southeastern U.S.”

    Poster · AGU 2025 · New Orleans, LA · Dec 2025

  • “Mapping Air Pollution Concentrations in Greater Boston Using Hyperlocal Mobile Sensing”

    Poster · AAAR 2025 · Buffalo, NY · Oct 2025

  • “Future Projections of Heatwave Events Using High-Resolution WRF Downscaling Based on the Pseudo Global Warming Method”

    Poster · Joint MPAS/WRF Users Workshop 2025 · Boulder, CO · Jun 2025

  • “Multi-Scale Air Quality Modeling Using WRF-Chem-GHG over Northeastern Africa”

    Oral · 23rd CMAS · Chapel Hill, NC · Oct 2024

  • “GPU Implementation of a Gas-Phase Chemistry Solver in the CMAQ Chemical Transport Model”

    Oral · 23rd CMAS · Chapel Hill, NC · Oct 2024

  • “Graphics Processing Unit Assisted Computation for a Gas-Phase Chemical Solver in a Regulatory Chemical Transport Model”

    Poster · BASC Symposium, UC Berkeley · Mar 2024

  • “Compounding Risks in Disparately Impacted Communities: California Case Studies and Future Recommendations”

    Oral · AGU Fall Meeting · San Francisco, CA · Dec 2023

  • “Household and Mobility-Influenced Personal PM2.5 Exposures for a Rail-Impacted Environmental Justice Community”

    Oral · AAAR 2023 · Portland, OR · Oct 2023

  • “UAV Measurements in a Heavily Burdened Air Basin to Understand Meteorological and Emissions Uncertainties in CMAQ”

    Poster · MAC-MAQ 2023 · Davis, CA · Sep 2023

  • “Air Pollution Exposure Mitigation for the Protection of Impacted Communities”

    Oral · AEESP 2023 · Boston, MA · Jun 2023

  • “Vertical Ozone Profiles Measurement in Riverside, CA”

    Poster · AGU Fall Meeting · Dec 2021

  • “On the Accuracy of Personal Exposures: Lessons from a Pilot Urban Personal Exposure Study”

    Poster · AGU Fall Meeting 2021

  • “Application of Machine Learning for Future Air Quality Predictions in Southern California”

    Poster · 38th AAAR Meeting · Oct 2020

  • “High-Temporal Resolution Personal Exposure Pilot Study in Inland Southern California”

    Poster · Air Sensors International Conference · Pasadena, CA · May 2020

  • “CMAQ-Enhanced Estimates of Personal Exposure to Diesel Particulates”

    Poster · CMAS Meeting · Chapel Hill, NC · Oct 2019

  • “High Temporal Resolution Microenvironmental Exposure in the Inland Empire: Implications for Exposure Risk of Under-served Populations”

    Poster · AGU Fall Meeting 2019

  • “Field Evaluation and Calibration of a Six-Parameter Low-Cost Sensor System in Northwestern and Southeastern US”

    Poster · 37th AAAR Meeting · Oct 2019
05 / Teaching, mentorship & service

Building the next cohort, and the proposals that fund them

Mentorship & leadership

  • Academic mentor to six Ph.D. students, overseeing progress and guiding research projects
  • Assist Principal Investigators in developing and preparing proposals for NSF, NOAA, and EPA funding opportunities
  • Technical advisor to the Center for Community Action and Environmental Justice (Jurupa Valley, CA) and The Air I Breathe (Colton, CA)
  • Community outreach in San Bernardino, CA

Teaching

  • WMO training instructor: WRF-Chem for the African domain, Egyptian Meteorological Authority
  • Guest lecture, Carnegie Mellon University (2023)
  • Invited talk, Northeastern University CEE Graduate Expo (2024)
  • Teaching Assistant, CHE 117 Separation Processes (enrollment 49), UC Riverside (2022) · Outstanding TA Award

Peer review & service

  • Poster evaluation committee, CMAS 2023 & 2024
  • Journal reviewer: Environmental Science & Technology, Nature Sustainability, GeoHealth, Atmosphere, Applied Sciences, Environmental Research Letters, GIScience & Remote Sensing, Geoscientific Model Development, npj Climate and Atmospheric Science
  • Student assistance, 2019 AAAR conference

Computational allocations

  • NCAR-Wyoming Supercomputing Center: repository UNTE0007, “SSP Downscaling and RRFS-CMAQ Development”: 500,000 CPU hours + 1,500 GPU hours
  • NERSC: 1,500 CPU node hours
  • Production experience on Cheyenne, Derecho, NERSC, Anvil, and Discovery
06 / Contact

Let's model something that matters.

Open to collaborations on air quality modeling, ML-accelerated chemical transport, climate downscaling, mobile monitoring, and environmental health, and to faculty and research-scientist opportunities.

kh.do@northeastern.edu Google Scholar ORCID Northeastern University · Boston, MA
Download full CV (PDF)