2019-063 City of Denton Support for UNT Air Monitoring GrantDate: April 5, 2019 Report No. 2019-063
INFORMAL STAFF REPORT
TO MAYOR AND CITY COUNCIL
SUBJECT
Provide information about the University of North Texas request for support of grant proposal.
BACKGROUND
Air quality has been a priority for Denton for many years. The City of Denton has supported local
air quality awareness and engaged citizens to provide education about actions to improve local air
quality for close to 20 years.
Partnerships with the North Central Texas Council of Governments (NCTCOG), and the
North Texas Commission in the late 1990’s and early 2000’s to educate about alternative
transportation.
Longstanding leadership roles on both on the North Texas Clean Air Steering Committee
and Regional Transportation Committee.
Incorporation of air quality goals into our first Sustainability Plan in 2012
Ongoing Greenhouse Gas Reporting
Greenhouse Gas Contribution Analysis
Air Quality Awareness Events and activities
On March 20, 2019 staff received an email from Lu Liang, a professor at University of North
Texas (UNT) regarding a grant proposal for air monitoring. On March 25th, we received additional
information regarding the proposal and project team and a request for a no cost letter of support
from the City of Denton to be submitted with the grant proposal to the Sloan Foundation. The
deadline for this submission was April 1, 2019.
The proposed research aims to meet several goals. A few key objectives include, exploration of
multiple calibration methods for low cost sensors including in-lab and machine learning
techniques, and the use of fine scale particulate matter data as a part of larger mapping and
modeling analyses. Results of this integrated monitoring, modeling, and mapping analysis includes
information that may be of value to the City’s future understanding of local air pollution dynamics.
UNT assembled a diverse team of personnel that crosses disciplinary (1 environmental / geospatial
science, 2 computer science, 1 data science, 1 atmospheric science 1 media promotion specialist)
and gender (3 female, 3 male) boundaries. The Primary Investigator (PI), Lu Liang, Ph.D., is a
geospatial scientist with 14 years’ experience in monitoring, mapping, and modeling of
environmental changes. Co-PI Dr. David Lary is an atmospheric scientist with expertise in using
observation and automation to facilitate air quality and health research. He has worked on
deploying a network of airborne allergen sensors for a smart city asthma and allergy early warning
system. Co-PI Dr. Qing Yang is a computer scientist and has intensive experience in building
prototype of sensors and sensor networks. Senior Personnel Dr. Yan Huang will oversee system
and platform development activities. She has been awarded several NSF grants for research on
spatial databases, geo-stream processing, big data as well as engaging local governments, teachers
Date: April 5, 2019 Report No. 2019-063
and students in cyber infrastructure. Consultant Laura Jana, MD, has nearly two decades of
experience using traditional and new media platforms to broadly translate and disseminate research
and evidence-based recommendations to lay audiences. She will co-lead media promotion and the
outreach/recruitment activities. Ph.D. student Constant Marks will participate in sensor
deployment and data assimilation and validation aspects of this research project together with
Liang.
See Attachment 1 for full project outline
DISCUSSION
There is no cost to the City of Denton to support the grant. If the grant is awarded, UNT will
coordinate citizen participation. The City may be asked to assist with locations for citizen
workshops, similar to support we provide and receive from UNT for multiple sustainability-related
projects.
Staff provided information to the Committee on the Environment on April 1, 2019, and the
Committee recommended providing a letter of support. Based on this recommendation, staff
submitted a letter of support for the project. Staff will keep the council informed of any
developments regarding grant award.
ATTACHMENTS
1. An Integrated Low-Cost, Multi-Scale Air Pollution Monitoring, Mapping, and Modeling
Platform
STAFF CONTACT:
Katherine Barnett
Sustainability and Customer Initiatives Manager
(940)349-8202
Katherine.barnett@cityofdenton.com
Date: April 5, 2019 Report No. 2019-063
Attachment 1
An Integrated Low‐Cost, Multi‐Scale Air Pollution Monitoring, Mapping, and
Modeling Platform
1. What is the core research question and why is it important?
Fine particulate matter with an aerodynamic diameter of less than 2.5 µm (PM2.5) is a major
component of atmospheric air pollution in cities worldwide1-4. The particle pollution is
surprisingly variable inside neighborhoods which can be up to five to eight times within an
individual city block5-7. Although all major U.S. cities have monitoring stations that provide
continuous measurements of criteria air pollutants, there are too few of these stations to capture
within-city air pollution gradients. For instance, the Dallas-Forth Worth metroplex (DFW), home
to ~7 million people, has only five PM2.5 stations, and most are located in old city cores (Fig 1).
Given the lack of sufficient government surveillance, the expanded use of low-cost sensors is
recommended by Environmental Protection Agency (EPA) as the “new paradigm for air quality
monitoring.” Despite the wide application of low-cost sensors in environmental monitoring, the
quality and reliability of the collected data for characterizing air pollution at the granular level is
yet not well investigated.
The goal of this proposal is to answer “how can detailed, localized data collected from a
network of low-cost sensors fulfill calibration, and be integrated with satellite observations
to improve local and regional understanding of air pollution dynamics?” via an integrated
“monitoring-mapping-modelling” framework. The three research objectives are:
1. Objective 1 (O1): Build a fine-grained, widely deployed ground monitoring network with
cost-effective sensors, mobile air quality monitor fleet, and data-driven calibration
algorithm.
2. Objective 2 (O2): Deploy very fine spatio-temporal mapping of intra-urban variability in
PM2.5 pollution concentrations to retrieve detailed, localized air pollution pattern.
3. Objective 3 (O3): Develop regional scale modeling of PM2.5 dynamics through the
integration of ground sensor network with satellite observations.
This multidisciplinary project will generate technology and knowledge necessary to enhance
air quality research. For example:
1. The project will improve self-calibration capacities of in-situ low-cost sensor network to
enable long-term, consistent, and reliable data;
2. Using low-cost sensors to capture variability in environmental system will satisfy a broad
range of community interests to meet a diverse set of needs;
3. Integration of in-situ and satellite data will fill the technical and knowledge gap of how to
augment and enhance earth observation data in predicting PM2.5 concentration.
This project will also exert significant societal impacts, such as:
1. A promotion of public participation in scientific research and increase of public awareness
in a “generating and applying” strategy that goes beyond dissemination of research
evidence;
Date: April 5, 2019 Report No. 2019-063
2. By providing substantial real-world scientific learning opportunities to the next generation
of scientists and leaders from diverse backgrounds, the project is anticipated to broaden the
participation of women and underrepresented groups in sensor technology;
3. Several research infrastructure prototypes will be established, for example the development
of mass calibration of low-cost sensors for retrieving high-quality environmental data.
2. What are the current knowledge gaps on this question?
Low-cost sensor calibration: With recent advances in low-cost sensor technologies, the air
quality monitoring paradigm is shifting from government and academic entities to individuals.
Unlike traditional stationary measurements that require high cost and expertise, low-cost sensors
enable flexible and long-term neighborhood monitoring in a mobile manner. To date, most air
pollution monitoring research using the sensor-based method is largely devoted to applications;
for example, human health and clean vehicle policy. Studies examining the problem of data
integrity, especially the limitations of collecting and using high-frequency, fine-resolution sensor
data for monitoring ambient PM2.5 levels, are limited in both scope and depth. Thus, an
overlooked yet essential task is to fulfill the automated calibration, bias-detection, and
uncertainty estimation of these low-cost sensors to ensure the measurement quality. In addition,
while much effort has been recently placed on the connectivity of the large disbursed networks,
little to no effort has been spent on investigating how the quantities and spatial distribution of in-
situ sensors may affect their ability in capturing the significant degree of air pollution variability
in urban areas across space and over time.
Multi-scale data integration: While measurements from low-cost sensors present a set of
scattered points that gauge real-time fluctuations in local air quality, it suffers from the
discontinuity of the point-based signals in the spatial domain. This limitation can be overcome
by integrating sensor data with satellite data that generate baseline air quality information.
However, remotely sensed air quality data also have limitations: among them is the typical
assumption that there is a constant relationship between ground PM2.5 levels and satellite-derived
aerosol optical depth (AOD), a measure of light extinction by aerosol in the entire atmospheric
column8-9. In reality, this AOD-PM2.5 relationship is affected by several factors: 1) PM2.5
measurements are usually of dry aerosol mass while satellite AOD is retrieved under all humidity
conditions; 2) AOD can only be retrieved from cloud/snow-free images for passive sensors
whereas ground PM2.5 is continuously measured under all conditions; 3) satellite AOD retrieval
is subject to larger algorithmic errors over bright impervious surfaces in urban; and 4) the aerosol
vertical profile vary in space and time, which greatly affects the AOD-PM2.5 relationship.
Failure to account for the spatial and temporal variability in the AOD-PM2.5 relationship may
underestimate the predictive power of satellite remote sensing for large-scale estimation of
ground level PM2.510, and a feasible way to improve is to calibrate the AOD-PM2.5 relationship
with a wealth of ground information covering a wide range of regional and seasonal conditions.
The potential for low-cost, in-situ sensor monitoring to aid in the generation of continuous and
realistic PM2.5 concentration fields via integration with earth observation data represents an
emerging and critical need in air quality research.
3. What is the proposed research methodology?
Date: April 5, 2019 Report No. 2019-063
This proposal seeks to improve city-scale pollution prediction capacities through an integrated
low-cost sensor monitoring, regional-scale satellite mapping, and multidisciplinary modeling
framework (M3). We will test and implement our integrated M3 framework in the DFW region,
which has been suffering from unhealthy levels of particle pollution for many years and only five
sparsely distributed PM2.5 monitoring stations.
Exploring our first objective, our team seeks to address how low-cost sensors can be used to
collect reliable air quality data at the neighborhood scale. M3 will provide an integrated three-tier
solution to data calibration and assurance within a large network of real time, low-cost devices
for airborne particulates. Before deployment, each sensor will be calibrated in a controlled dust
generation chamber (lab calibration) against an EPA certified reference to obtain the
individual probability distribution function. Those sensors will also be sent back to the lab for
recalibration after a certain operation period. Two zero emission electric cars (field
calibration) carrying the PM reference instrument will be used to routinely drive past all
deployed sensors to provide ongoing routine calibration in the field that can differ substantially
from the controlled lab environment considering the highly variable outdoor conditions.
Considering air pollution tends to have similar changing trends at nearby locations, any drift in a
sensor will be quickly detected as part of a fully automated, real-time workflow where the
trajectory of each sensor’s measurements will be automatically compared to that of its neighbor’s
and to the reference instrument (mass calibration). With the big data, machine learning that has
been successfully tested by our team will be used to correct these inter-sensor biases11.
Besides reliable devices, the successful quantification of intra-urban pollution variability relies
on how to recruit citizens whose residential locations have good spatial representation across
different air pollution stratum. Two recruitment strategies will be used:
1. “School hub” strategy: Student ambassadors trained at UNT will take sensor parts to
schools to train high school students on the installation and usage. The high school students
will take the sensors home and put in their backyards. Whenever lab recalibration is
required, they will bring the sensors back to school and test it against a reference device
that we set up in school. This strategy ensures the high recalibration quality of those
sensors simultaneously at a desired locations and fulfills our educational goal for preparing
the next generation leaders in sensor technology and environmental science;
2. Interested individuals will be asked via traditional and social media to report their house
addresses, and we will select a representative sample covering each pollution strata that
was partitioned based on historical annual average PM2.5 concentrations. We will hold
workshops at local libraries to train participants the installation and usage of sensors.
An integrated M3 management system will be established for data dissemination and
community engagement, which will include three modules:
1. Air quality mapping and visualization platform, which enables the display of real-time data
streamed from all deployed sensors and allows users to cross-reference sensor data and
community information to look for potential local effects;
2. The participant portal that allows individuals to receive daily reports about their air
pollution exposure level to motivate their engagement;
3. The manager dashboard functions as the control panel that supports real-time monitoring of
device activity and pushes automatic notifications to offline or abnormal devices.
Date: April 5, 2019 Report No. 2019-063
We will integrate the real-time in-situ data collected from O1 and satellite-derived Aerosol
Optical Depth (AOD) to model the dynamics of PM2.5concentrations in the DFW region at
detailed scales. Considering that the AOD-PM2.5 relationship is spatially inconsistent and
temporally varying, we will utilize mixed effects models to predict location-specific PM2.5levels
by considering a spectrum of variables, including weather (e.g. humidity, temperature,
precipitation, surface pressure, wind), sociodemographic , traffic, and landscape composition at
horizontal (e.g. land cover and land use) and vertical dimensions (e.g. building and tree height).
Furthermore, we will assess the effects of using citizen science data by comparing the model
results against those modelled with only fixed monitoring station data. Three different AOD
products will be used in the model to explore their applicability and limitations in modeling
ground PM2.5. Finally, we will conduct computer simulation to test to what extent regional-scale
air pollution distribution can be adequately modeled with real-time monitoring data. In
particular, we will explore how many in-situ samples are sufficient and how spatio-temporal
representativeness of citizen science data will influence modeling.
4. What will be the outputs from the research project and how will they be disseminated?
Our multidisciplinary team will focus on generating the following key outputs:
1. Publications in high impact academic journal and research summaries for professional
journals.
2. A new data calibration system for real-time, low-cost air quality monitoring devices;
3. Provision of both hyper-local in-situ air quality data and satellite derived products through
M3 data portal.
4. Evidence summaries for particulate matter dynamics in DFW and identification of risks.
5. Two-minute online video and targeted social media campaign communicating the potential
impact of the research findings on real people
6. A TEDex event showcasing our mission and project to local audiences.
To ensure that the outputs maximize the benefit to academia, industry, and the public, we
identify four key audiences and the corresponding strategies to disseminate this research:
1. Research organizations through academic forums (local, regional, and national
conferences) and peer-reviewed publications;
2. Stakeholders through semi-annual project report and annual survey to gain feedback from
this community;
3. The general public via digital media platforms, the project website, and TEDex events
throughout the duration of the project and after completion to provide regular updates on
the project’ progress and publication of deliverables;
4. Direct engagement with the citizen in the neighborhood where the sensors are placed by
hosting regular sensor training workshops in libraries and high schools.
5. What are the proposer and team qualifications
We have assembled a diverse team of personnel that crosses disciplinary (1 environmental /
geospatial science, 2 computer science, 1 data science, 1 atmospheric science, 1 media
promotion specialist) and gender (3 female, 3 male) boundaries. The Primary Investigator (PI),
Lu Liang, Ph.D., will provide overall project direction and management. She is a geospatial
scientist with 14 years’ experience in monitoring, mapping, and modeling of environmental
changes. She has led four citizen-engaged air pollution monitoring campaigns with more than
Date: April 5, 2019 Report No. 2019-063
100 participants in Beijing12. She has successfully mentored underrepresented students in
publication and presentation of scientific research. She will participate in all aspects of research.
Co-PI Dr. David Lary is an atmospheric scientist with expertise in using observation and
automation to facilitate air quality and health research. He has worked on deploying a network of
airborne allergen sensors for a smart city asthma and allergy early warning system. He will
supervise a Ph.D. student and lead the lab calibration and integrated calibration algorithm design.
Co-PI Dr. Qing Yang is a computer scientist and has intensive experience in building
prototype of sensors and sensor networks. For example, supported by Federal Highway
Administration, Dr. Yang designed and implemented a remote, self-sustained sensing system for
monitoring water quality near highways. Dr. Yang is currently managing the connected and
autonomous vehicle (CAV) lab at UNT, and he will lead the development of vehicle monitoring
platform for field calibration in this project.
Senior Personnel Dr. Yan Huang will oversee system and platform development activities.
Dr. Huang has made original contributions of major significance to spatial databases, spatial data
mining, and geo-stream processing. She has been awarded several NSF grants for research on
spatial databases, geo-stream processing, big data as well as engaging local governments,
teachers and students in cyber infrastructure.
Consultant Laura Jana, MD, has nearly two decades of experience using traditional and new
media platforms to broadly translate and disseminate research and evidence-based
recommendations to lay audiences. She will co-lead media promotion and the
outreach/recruitment activities.
Ph.D. student Constant Marks will participate in sensor deployment and data assimilation and
validation aspects of this research project together with Liang. Marks has a decadal long
experience as an engineering consultant developing and applying emission measurement and
control systems. He is now running a citizen science initiative to deploy a low-cost sensor
network in DFW that will be utilized in this project.
6. What other sources of support can the project leverage?
The University of North Texas (UNT) the University of Texas, Dallas (UTD) fully supports
the academic year salaries of Dr. Liang, Huang, Yang, and Lary. Dr. Liang’s laboratory at UNT
will serve as the central site for project coordination, database management, lab calibration of
monitoring sensors, and student training. She also has access to the Center for Spatial Analysis
and Mapping at UNT that will provide instructional support in the areas of geographic
information systems, computer cartography, spatial analysis, and environmental modeling.
The CAV laboratory led by Dr. Yang has eight desktop computers running Linux and
Windows systems, one PowerEdge 2900 application server, one PowerEdge 1800 network
storage server with 10 TB disk. In addition, there are two golf carts that are customized to
replicate the perception system on autonomous vehicles consisting of Nvida DRIVE PX2,
cameras, LiDAR, Radar and GPS devices. This lab will be the central site to design air quality
monitoring platform in mobile (e.g., vehicle) and fix-site manner.
Dr. Huang’s research laboratory is ∼600 square feet. It is equipped with four servers, 10 Unix
workstations, numerous Windows-based systems, a 4-node SUN SMP system, terabyte storage,
Date: April 5, 2019 Report No. 2019-063
and a 16-node Beowulf cluster. Various software including Oracle, MySQL, PostgreSQL,
Matlab, GIS software, Clementine, MSDN, and various compilers for different languages. This
lab will serve as the research and development center to develop calibration systems and the
data storage and visualization hub.
The Multi-Scale Integrated Intelligent Interactive Sensing Center (MINTS) managed by Dr.
Lary has powerful hardware and software capacities to assemble, fabricate and calibrate sensing
systems utilizing facilities for additive manufacture. Sensor ensemble and controlled
environment calibration will be mainly conducted here.
References:
1. Götschi, T., Heinrich, J., Sunyer, J. and Künzli, N., 2008. Long-term effects of ambient
air pollution on lung function: a review. Epidemiology, 19(5), 690-701.
2. Health Effects Institute. 2018. State of Global Air 2018. Special Report. Boston, MA:
Health Effects Institute.
3. Huang, J., Pan, X., Guo, X. and Li, G., 2018. Impacts of air pollution wave on years of
life lost: A crucial way to communicate the health risks of air pollution to the
public. Environment international, 113, 42-49.
4. Laden, F., Neas, L.M., Dockery, D.W. and Schwartz, J., 2000. Association of fine
particulate matter from different sources with daily mortality in six US
cities. Environmental health perspectives, 108(10), 941.
5. Apte, J.S., Messier, K.P., Gani, S., Brauer, M., Kirchstetter, T.W., Lunden, M.M.,
Marshall, J.D., Portier, C.J., Vermeulen, R.C.H., and Hamburg, S.P., 2017. High
resolution air pollution mapping with Google Street View cars: Exploiting big
data. Environmental Science & Technology, 51(12), 6999-7008.
6. Dionisio, K.L., Rooney, M.S., Arku, R.E., Friedman, A.B., Hughes, A.F., Vallarino, J.,
Agyei-Mensah, S., Spengler, J.D. and Ezzati, M., 2010. Within-neighborhood patterns
and sources of particle pollution: mobile monitoring and geographic information system
analysis in four communities in Accra, Ghana. Environmental health
perspectives, 118(5), 607.
7. Van Vliet, E.D.S. and Kinney, P.L., 2007. Impacts of roadway emissions on urban
particulate matter concentrations in sub-Saharan Africa: new evidence from Nairobi,
Kenya. Environmental research letters, 2(4), p.045028.
8. Tao, J., Zhang, M., Chen, L., Wang, Z., Su, L., Ge, C., Han, X. and Zou, M., 2013. A
method to estimate concentrations of surface-level particulate matter using satellite-based
aerosol optical thickness. Science china earth sciences, 56(8), 1422-1433.
9. Wang, J. and Christopher, S.A., 2003. Intercomparison between satellite-derived aerosol
optical thickness and PM2.5 mass: implications for air quality studies. Geophysical
research letters, 30(21).
Date: April 5, 2019 Report No. 2019-063
10. Hu, X., Waller, L.A., Lyapustin, A., Wang, Y., Al-Hamdan, M.Z., Crosson, W.L., Estes
Jr, M.G., Estes, S.M., Quattrochi, D.A., Puttaswamy, S.J. and Liu, Y., 2014. Estimating
ground-level PM2.5 concentrations in the Southeastern United States using MAIAC AOD
retrievals and a two-stage model. Remote sensing of environment, 140, 220-232.
11. Lary, D.J., Remer, L.A., MacNeill, D., Roscoe, B., and Paradise, S., 2009. Machine
learning and bias correction of MODIS aerosol optical depth. IEEE Geoscience and
Remote Sensing Letters, 6(4):694-698.
12. Liang, L., Gong, P., Cong, N., Li, Z.C., Zhao, Y., and Chen, Y., Assessment of personal
exposure to particulate air pollution: the first result of City Health Outlook (CHO)
project. BMC Public Health, In revision.