HazeMon
IoT เพือ่การเฝ้าระวงัหมอกควนั
Nisarat Tansakul, Sathita Fakprapai, Sararat Nontachai
and Jutamas Kunphan
at 34th WUNCA
Outline
• HazeMon ?
• Background and Objective
• Hardware and Sensors
• Related Research
– Improvement of Wireless Sensor for Biomass
Burning Smoke Air Pollution Monitoring
– Assessment of exposure to biomass burning
smoke for School children in hotspot areas
of Chiang Rai province, Thailand
– Assessment of Health effects of Biomass
Burning Emission in Chiang Rai Province,
Thailand
HazeMon?
• Low-cost Real-time Monitoring of Haze
Air Quality Disasters in Rural
Communities in Thailand and
Southeast Asia.
• Sponsored by a grant from the French
Ministry of Foreign Affairs and
International Development
• Using Wireless Sensors (IoT)
Background
• In Thailand and Southeast Asia, haze air
quality disasters every year.
• Haze disaster affects the health of people
living in a very large area.
• Government weather stations that report
haze conditions are costly and located
mostly near to large urban cities.
• Rural citizens and farmers still lack timely
capabilities to get the information
regarding haze air quality disaster.
Background
• Low-cost sensors was deployed at
Community Network sites in Tak
Province.
6
Early Warning
Objective
• To build and validate low-cost haze
monitoring sensors that can be
deployed in remote rural areas to
warn rural citizens when too much
haze is present.
Team and Partners• Thailand : Asian Institute of Technology (AIT)
– Internet Education and Research Laboratory (intERLab)
– School of Environment Resources and Development
(SERD)
• France :– Laboratoire Informatique de Paris 6 (LIP6), UPMC, Paris,
France
– Institut national de la santé et de la recherche médicale
(INSERM), Paris, France
• Asia :– Institute of Computer Science, University of the Philippines
Los Baños, Los Baños, The Philippines
– MIIT-University of Kuala Lumpur, KL, Malaysia
Hardware and Sensors
• Hybrid Architecture
– Microcontroller (for sensing)
– Arm-like (for system operations)
• Initial production based UDOO
– WPA-2 Enterprise
– Linux Ubuntu v.s. OpenWRT
Hardware and Sensors
• Temperature
• Relative humidity
• Air pressure
• PM1
• PM2.5
• PM10
• Formaldehyde
Web Interface• Developed with UPMC
– Running on Amazon Docker
– Able to scale
– Easy to use
Site Selection
• 3 difference areas in Chiang Rai
Related Research
• Improvement of Wireless Sensor for
Biomass Burning Smoke Air Pollution
Monitoring
• Assessment of exposure to biomass
burning smoke for School children in
hotspot areas of Chiang Rai province,
Thailand
• Assessment of Health effects of Biomass
Burning Emission in Chiang Rai Province,
Thailand
Related Research
For Biomass Burning Smoke Air Pollution Monitoring
of Wireless Sensors
Improvement and Application
By : Sathita Fakprapai
Haze pollution
88of 3.7 million premature death % Air pollution monitoring
(WHO, 2012)
Objective and Scope
Improve AIT existing wireless
sensors with more sensors & to
test with three different fieldsapplications.
Adding more suitable sensors.
Evaluate the improved sensor nodes.
Field measurement and real time publicizing.
Selection & Assembly
Phase3
Phase 1
Phase 2
Evaluation with referent instruments
Apply to Field & Publish data
NO2, VOCs, PM, CO, CO2
Meteorological parameters &GPS
Particulate matter (PM)
30 times larger than the largest fine particle
(EPA, 2017)
(Thairath, 2016)
(Alen Air Purifiers, 2016)
Volatile organic compounds
(VOCs)
(Libelium, 2017)
Nitrogen dioxide (NO2)
(PCD, 2016)
(NASA, 2017)
Carbon monoxide (CO)
Carbon dioxide (CO2)
(Yokohamathailand, 2017)
Microcontroller Arduino YUN
Transceiver 3G/Wifi
GPS module Grove GPS
Software Arduino Software (IDE)
NO2 sensor 1,390
VOCs sensor 840
Dust sensor 790
CO sensor 120
CO2 sensor 1,250
Temp & Humidity 690
GPS module 1,250
Acrylic Box 200
3G Air card 1,000
SD card 786
Arduino YUN 2,800
Base shield 390
Total 11,506
Assembly sensor box
Literature Review
Sensor Selection
Suitable Sensors
Assembly Sensor Nodes
Evaluation
Apply to Field Monitoring and Publicizing Data
Discussion and Conclusion
Methodology
Sensors evaluation
AIT/Rice Field Burning PCD monitoring stations Chaingrai
AIT/Rice Field Burning
PCD monitoring stations
Sensor nodes will consist of suitable VOCs, NO2,
CO, CO2, PM2.5, PM10 and BC sensors.
The calibration results will be acceptable value.
The nodes will be apply at field and report real-time data.
Expected results
Demonstrate sensor box
https://thingspeak.com/channels/209860/private_show
Link for data monitoring
ASSESSMENT OF EXPOSURE TO BIOMASS
BURNING SMOKE FOR SCHOOL CHILDREN
IN HOTSPOT AREAS OF CHIANG RAI
PROVINCE, THAILAND
Master student of Environmental Engineering and Management
School of Environment, Resources and Development
By : Sararat Nontachai
Asian Institute of Technology
(AIT)
OBJECTIVES
The overall goal is to assess the exposure levels and
potential health effects of emission from biomass burning.
The concrete objectives are as follows:
To assess ambient air pollution in selected study areas
during burning and non-burning periods
To conduct exposure assessment
by using personal exposure monitoring
To comparatively analyze air pollution and exposure levels
during burning and non-burning periods
6/23
SCOPE OF STUDY
Focusing on exposure of school children
to biomass burning in Chiang Rai province
The monitoring is conducted in two period;
non-burning and burning period
Ambient monitoring and personal monitoring;
PM2.5 and VOCs are collected
Survey, questionnaires, time-activity diaries,
direct observation, laboratory experiment
as well as exposure model will be applied
to assess exposure to pollutants from biomass burning.
7/23
FRAMEWORK OF METRODOLOGY
9/23
Chiang Rai: selected area
Burning period Non-burning period
Ambient monitoring:
PM2.5 and BTEXQuestionnaires and spirometric test
of 100 students
Personal monitoring PM2.5, BTEX :
(2 subjects/season)
Personal exposure concentration
Data compilation and analysis
Conclusions& recommendations
Ambient monitoring:
Mirror foundation
Ban Huai Khom School (BHKS)
Personal exposure monitoring:
Student from BHKS
Student from Ban Thung Luang
Exposure model
Health
SITE DESCRIPTION
Chiang Rai
Source: Google Maps (2016)
▣ Mueang district was selected as the one of intensive burning area in
Chiang Rai. ▣ High frequency of hotspots, easy to access, convenience (telephone,
electricity) and safety are the reason for selecting.
10/23
SITE SELECTION
• Mae Yao sub-district has large amount of hotspots arising• This sub-district is easy to access
• The ambient monitoring stations HAZEMON projects will be located there
• PCD monitoring stations is nearby
In this study, Mae Yao sub-
district is selected as a
monitoring area
Source: Google Maps (2016)
12/23
SITE OF MONITORING STATION
Distance from Mirror
foundation to
Ban Huai Khom School
around 1.5 km
13/23
Mirror foundation and
Ban Huai Khom School (BHKS)
are selected for ambient air monitoring
Mirror foundation
Ban Huai Khom School
The selected participants will need to meet certain criteria, including
1)participant’s age vary from 10-12 year-old.
2) the household is not close to the roads.
3)Non-smoking residents.
4) the main cooking fuel is LPG.
SELECTION OF SCHOOLS AND
SCHOOL CHILDREN
• There are many school childrean in Mae Yao sub-district.
• Ban Huai Khom School (BHKS) and
Ban Thungluang school are selected for conducting questionnaires.
• School children are selected as a target group of this study.
• Subjects are selected from surveys and questionnaires.
14/23
SITE OF SELECTED SCHOOL
15/23
Ban Huai Khom School
Ban Huai Khom school
Monitor
PM2.5 : MiniVol and GRIMM
BTEX : Charcoal coconut tube
Ban Thungluang school
Ban Thung luang School Distance from BTLS to
BHKS around 2.8 km
SAMPLING DESIGN AND
SAMPLE SIZE
• The sample size will be calculated by Taro Yamane equation
(Equation 1).
Where:
n: Sample size
N: Total number of population
e: The acceptable error (10%)
▣ This study focuses on children population who are 10-12 years old.
▣ The total number of population in selected group is around 1104.
▣ The sample size should be 100
▣ 2 subjects who meet criteria and not meet criteria selected from surveys and questionnaires
to conduct PEM and SKC charcoal coconut tube.
𝑛 =𝑁
(1 + 𝑁𝑒2)
16/23
SUMMARY OF SAMPLING PLAN
Ambient monitoring
(2 locations)
Personal monitoring
( 2 students)
Equipment
- MiniVol: PM2.5
- CO : online data
- SKC charcoal coconut tube with personal
pump for BTEX
- PEM: PM2.5
- SKC charcoal coconut tube with personal
pump for BTEX
Sampling media
- Quartz filter ∅47mm: PM2.5 mass (Ambient monitoring)
- Quartz filter ∅37mm: PM2.5 mass (Personal monitoring)
- Online measurement: CO
- SKC-charcoal coconut tube: BTEX
Sampling duration - 24 hour sampling - 24 hour sampling
Sampling period
- Non burning period: December 6 – February 5, 2016
- Burning period: February 15 – March 20, 2017
17/23
MONITORING SITE
18/23
Ambient monitoring
Personal monitoring
DATA COLLECTION
Surveys Questionnaires Time activity diaries Direct observation
OTHER DATA COLLECTION
Spirometric test
Air pollution dataMeteorology data
Hospital records
19/23
DATA ANALYSIS
Exposure modeling
▣The total exposure of subjects to each
pollutant was calculated by exposure model
▣ Input data: Exposed concentrations and time
of subjects spend in each environment
(obtained from time activities diaries)
▣ Output data: Exposure concentration of each
pollutant during a period 24 hours.
Statistical data analysis
▣ To test the difference in personal
exposure among participants, a paired
t-test will be applied.
▣ To study the relationship between
biomass burning with haze air pollution,
a Pearson correlation coefficient
analysis will be applied.
20/23
QUALITY ASSURANCE AND
QUALITY CONTROL (QA/QC)
Sampling and Analysis
Parameters Measurement method Calibration method
PM2.5 Microbalance USEPA (1998)
BTEX GC/FID NIOSH 1501 (NIOSH, 2003)
21/23
EXPECTED OUTCOMES
A comparison of air pollution concentration in
burning season and non-burning period in Mae
Yao sub-district, Mueang district, Chiang Rai
Exposure concentration of PM2.5 and BTEX
from selected group
The potential health effects between burning and
non-burning period
23/23
ASSESSMENT OF HEALTH EFFECTS OF
BIOMASS BURNING EMISSION IN CHIANG RAI PROVINCE, THAILAND
Environmental Engineering and Management
School of Environment, Resources and Development
Asian Institute of Technology, Thailand
Ms. Jutamas KunphanMaster’s student
To assess health effects of biomass burning
emission in Chiang Rai province, Thailand.
Overall objective
1. To analyze hospital records in relation to air pollution levels
(PM10, CO).
2. To assess air pollution levels during burning and non-burning
seasons in relation to the hotspot counts and meteorological
data.
3. To assess the deposition potential of haze particles in the
human respiratory system.
Specific objectives
OBJECTIVES
SCOPE OF STUDY
The study focus on analysis of hospital records during six
years (2011-2016) in Chiang Rai province, Thailand.
Air pollution level focusing on PM10, CO, and fire hotspots are
analyzed with the hospital records to reveal relationships.
The MPPD model will be applied to assess the
deposition rate of haze particles in human lungs.
The particle size distributions during haze and non-haze periods
required for MPPD modeling determined by a cascade impactor.
METHODOLOGY FRAMEWORK
Phase 2: Data analysis
Conclusions and recommendations
Phase 1 : Data collection
Data analysis
Relationship between air pollution
level and Hospital records
Relationship between hospital records, air
pollution level, meteorological and hotspot data
Input data of MPPD model(Measurement PM by using a cascade impactor
MPPD model
Output data
PCA
Ambient air monitoring data
(CO and PM10)
Hospital records
(Involve with air pollution)
Hotspot data (from MODIS
satellite)
GAM
Meteorologicaldata
Select the study area
(Chiang Rai province, Thailand)
Select the study period (2011-2016)
METHODOLOGY(Data collection)
Ambient air monitoring data (CO and PM10) from 2 PCD
monitoring stations in Chiang Rai
Hospital records from Hospital in Chiang Rai
Satellite data (Hotspot) from Fire Information for Resource
Management System (FIRMS) website
Input data of MPPD model (Particle concentration will be
sampling by using a cascade impactor)
The collected daily data of hospital
records and daily concentrations of
air pollutants (PM10 and CO)
Poisson regression with generalized additive model (GAM)
To reveal relationships between the
pollution levels and daily numbers of
patients in different conditions (haze and non-haze periods)
Using SPSS software to analysis
Generalized Additive Model (GAM)
METHODOLOGY(Data analysis)
ParametersCO (ppm)
PM10 (µg/m3)
Wind speed (m/s)
Surface temperature (°C)
Relative humidity (%)
Air pressure (mb)
hospital records
Biomass burning hotspots
All of these will be analyzed by
the PCA from SPSS software to
reveals the relationships.
Principal Component Analysis (PCA)
Stage Size range (µm)
0 9.0 – 10.0
1 5.8 – 9.0
2 4.7 – 5.8
3 3.3 – 4.7
4 2.1 – 3.3
5 1.1 – 2.1
6 0.7 – 1.1
7 0.4 – 0.7
backup filter 0-0.4
Particle Size Data
Input data of MPPD model
(Particle concentration)
Cascade impactor
● Particle concentration● Breathing frequency● Tidal volume● Inhalation fraction● Pause fraction
Input data of MPPD model
Data collection both burning
and non-burning seasons.
Multiple-Path Particle Dosimetry Model
(MPPD v 2.11)
Input data of MPPD model
(Particle concentration)
MONITORING SITE
At the Mirror Foundation,Chiang Rai
EXPECTED RESULTS
The number of patients has positive correlation with the PM10
and CO concentration level in burning season.
Health incidence rate during burning and non-burning
season are significant difference.
The correlation demonstrate between air pollution
concentration (PM10, CO) and number of hotspot.
Prediction deposition patterns in the human respiratory tract
of particle size distribution between burning and non-burning
season are available.
Conclusion
• HazeMon is an example of research
which is extended to real community.
• We should starting to plan how we can
extend our research to community.