Subscriber access provided by La Trobe University Library
Environmental Measurements Methods
Using Indoor Positioning and Mobile Sensing for Spatial Exposure and Environmental Characterizations: Pilot Demonstration of PM2.5 Mapping Kai-Chung Cheng, Ching-Hao Tseng, and Lynn M. Hildemann Environ. Sci. Technol. Lett., Just Accepted Manuscript • DOI: 10.1021/acs.estlett.8b00694 • Publication Date (Web): 04 Jan 2019 Downloaded from http://pubs.acs.org on January 5, 2019
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 14
Environmental Science & Technology Letters
1
Using Indoor Positioning and Mobile Sensing for Spatial Exposure and Environmental
2
Characterizations: Pilot Demonstration of PM2.5 Mapping
3
Kai-Chung Cheng,1* Ching-Hao Tseng, Lynn M. Hildemann1
4
1Civil
5
*corresponding author:
[email protected] & Environmental Engineering Dept., Stanford University, Stanford, CA 94305
6 7
ABSTRACT
8
New indoor positioning technology makes it possible to use air measuring devices carried by
9
occupants to characterize spatiotemporal patterns of exposure and air quality inside buildings.
10
This pilot study investigated the potential of a mobile monitoring method to map highly variable
11
PM2.5 distributions inside an occupied one-bedroom apartment, coupling a new ultrasound
12
localization method with a pair of collocated research-grade and low-cost sensors (SidePak and
13
PTQS). Measuring the position of the mobile sensing node every 1 s, down to the centimeter
14
scale, this method generated detailed occupant moving trajectories and location histories on a
15
floorplan map, allowing identification of microenvironments causing transient peak exposures.
16
Utilizing a random walking approach throughout the apartment, it identified smoke intrusion
17
sources and captured the spatial distributions of PM2.5 using interpolated 2-D concentration
18
fields. The correlations between SidePak and PTQS were evaluated when used as portable
19
devices for (i) occupant exposure and (ii) indoor environmental mapping applications. This new
20
mobile sensing method is most effective with a rapidly responding air monitor (e.g., SidePak
21
photometer).
22
INTRODUCTION
23
Indoor air pollution levels close to sources such as cooking and smoking are substantially
24
higher than farther away – this “proximity effect” leads to high spatial variance and sizable
25
differences between personal and stationary measurements (e.g., 1-5).
26
Stationary research-grade sensors have evaluated spatial distributions of fine particulate
27
matter (PM2.5) inside two residences6 and a hospital5 with a smoker, and for a smoke source in
28
an office and a garage.7 Low-cost sensor arrays have measured PM in two households using
29
solid fuels,8 a shop with wood dust emissions,9 and a vehicle manufacturing facility.10 Using
30
sufficient numbers of sensors and suitable interpolation methods (e.g., kriging), concentration
31
fields can be determined with high spatial resolution. 1 ACS Paragon Plus Environment
Environmental Science & Technology Letters
32
The stationary mapping method typically involves deploying sensors near expected source
33
locations. However, for buildings with unknown emissions (e.g., smoke intrusion), sensor array
34
placement becomes less straightforward.
35
With recent advances in indoor positioning technologies, portable monitors potentially can be
36
used for indoor spatial mapping. Occupants carrying monitors could serve as mobile sensing
37
nodes, allowing unknown sources inside buildings to be pinpointed.
38
Using a new indoor positioning technique involving ultrasound, this study measured locations
39
of an occupant-worn SidePak to ±2 cm resolution. This mobile monitoring method was
40
investigated for (i) tracking occupant time-location patterns and PM2.5 exposures and (ii)
41
mapping PM2.5 distributions for smoke intrusion.
42
By integrating this indoor positioning with wireless PM2.5 sensing, we evaluated how well a
43
low-cost sensor carried by the occupant can capture, in the field, the time-varying location of
44
exposure and spatial distribution for PM2.5, compared with a research-grade monitor (SidePak).
45
MATERIALS AND METHODS
46
Equipment. A pair of collocated AM510 SidePak (TSI Inc., Shoreview, MN, USA) and
47
PTQS1005 (Plantower Co., Ltd., Beijing, China) monitors sampled PM2.5 in real time (Figure
48
1(a)), with short response times (~0.8) slightly lower than previously reported values (>~0.9)13 comparing 3 low-cost sensors
129
vs. SidePak for incense burning under well-mixed conditions. Besides differences in sensors and
130
sources, our study involved more rapid decays (ventilated vs. well-sealed chamber) and shorter
131
averaging times (1 versus 30 s). The ratio varied greatly between sources (Table S2), consistent
132
with previous studies.17-19 These ratios were used to rescale PTQS measurements for each source
133
emission and subsequent decay period. For concentrations comparable to background levels,
134
daily pre-emission scaling factors (1.14-1.28) were used.
135
Exposure Characterization. Figure 2(a) shows 5 clusters of spikes from the collocated
136
mobile SidePak and PTQS (rescaled). While the SidePak had many maxed-out readings
137
(replaced with 20 mg/m3 in the plot), the PTQS (maximum limit ~5 mg/m3) had none. This
138
reflects the SidePak’s more rapid response to concentration changes. The ratios of SidePak to
139
PTQS (unrescaled) time-averaged concentrations over each emission period were systematically
140
higher than scaling factors involving gradually-varying concentrations (Table S2). Overall time-
141
averaged exposure measured by PTQS was 32% (rescaled) and 52% (unrescaled) lower than
142
SidePak.
143
To identify the spatial regions (microenvironments) for these 5 clusters, PM2.5 time series
144
(Figure 2(a)) were connected with occupant location data measured by ultrasound beacons.
145
Figure 2(b) and 2(c) show occupant locations (second by second) as circles on the floorplan, with
146
PM2.5 levels (from SidePak and PTQS, respectively) on a color scale. The plots visualize the
147
patterns for occupant movement around the furniture and between rooms, along with the
148
corresponding exposures.
149
Coupling SidePak with positioning measurements (Figure 2(b)), specific indoor locations
150
(kitchen stove, living room desk, and chair near main entrance) were identified for the
151
concentration spikes (Figure 2(a)). The peak for the spraying activity (chair near main entrance,
152
Figure 2(b)) is less visible due to its much shorter emission period (~5 s versus 10-43 min for
153
cooking/vaping). Higher concentrations along trajectories between the stove and living room
154
table were measured during the decay period after cooking brunch (1st cluster in Figure 2(a)). 5 ACS Paragon Plus Environment
Environmental Science & Technology Letters
155
Despite PTQS shortcomings in capturing concentration variations, it identified the
156
microenvironments with elevated exposures, except for spraying at the main entrance chair
157
(Figure 2(c)). This is not surprising – for emission period (~5 s) ≤ sensor response time (≤10 s),
158
detection will be unlikely.20
159
Rescaling PTQS increased its correlation with SidePak PM2.5 along the moving trajectories
160
(R2 = 0.57 versus 0.49) - this is because it normalized for varying sensor responses to different
161
emissions. The correlation for rescaled PTQS is much lower than in chamber experiments with
162
gradually varying well-mixed concentrations (R2 = 0.78-0.93 here; 0.8-1.0 (typically) in previous
163
studies13,16,17). The more rapid concentration fluctuations weaken the correlation due to different
164
monitor response times. This is supported by the high SidePak to PTQS (rescaled) ratio (1.86) –
165
while the SidePak was able to sense transient peaks, the PTQS was not. Excluding all periods
166
with transient spikes gave a stronger correlation (R2 = 0.90) with a slope (1.05) close to the
167
expected value (1).
168
Environmental Characterization. The moving trajectories of the occupant (blue lines) and
169
measurements at 751 different locations (black dots) in Figure 3(a) covered the three rooms. All
170
3 hidden sources (red stars, Figure 3(a)) were identified by the occupant (either by SidePak
171
display or smell of smoke); therefore, more data points are in proximity to these sources as well
172
as near the air purifier.
173
A 2-D linear interpolation for each room used MATLAB’s (MathWorks Inc., Natick, MA,
174
USA) griddata function, and 2×2 cm grid size (consistent with positioning resolution), to test
175
whether a linear approach (simpler than Kriging) can capture high concentration locations, given
176
much more dense spatial data. Figure 3(b) and 3(c) plot the SidePak and PTQS (rescaled) spatial
177
distributions, using the MATLAB surface function.
178
Figure 3(b) visualizes the locations of three incense sources and the air purifier (localized
179
minimum in bedroom). Concentration gradients close to the sources reveal air flow patterns
180
indoors (e.g., plume moving from the open window to the living room sofa).
181
The PTQS also captured the three emission source locations and the air purifier (Figure 3(c)).
182
However, high concentration regions looked more dispersed (e.g., plumes in the bedroom).
183
Unlike the exposure experiment, in this experiment the sampling point was constantly moving.
184
Due to PTQS time response (≤10 s), the effects of peak concentrations persisted after the
185
occupant had passed the source location, smearing the plumes near emissions. 6 ACS Paragon Plus Environment
Page 6 of 14
Page 7 of 14
Environmental Science & Technology Letters
186
The correlation of rescaled PTQS with SidePak concentrations at all interpolated grid points
187
was R2 = 0.50, similar to grid points ≤2 m from sources (0.48). Excluding these proximity grid
188
points, the correlation became much stronger (R2 = 0.74).
189
To summarize, as a mobile device, the low-cost PTQS can resolve general spatial distributions
190
of exposures (Figure 2(c)) and concentrations (Figure 3(c)) indoors. However, it cannot represent
191
pollutant levels as reliably as the SidePak when rapidly varying concentrations occur or source
192
types are unknown for scaling adjustments. To resolve detailed variations close to a source, and
193
detect transient emission peaks, it is advisable to use a research-grade monitor that can more
194
quickly respond to rapid changes in concentrations.
195
IMPLICATIONS
196
Using research-grade photometers, studies (e.g., 16,17,21,22) have evaluated low-cost
197
particulate matter sensors in stationary laboratory settings. This study assesses low-cost sensors
198
when used as mobile devices in a real indoor environment with sources, where sporadic
199
concentration spikes occurred as the occupant walked around.
200
Previous studies (e.g., 23-25) have used GPS and portable monitors to map locations and
201
personal exposures outdoors, and indoors at the whole building scale. This study shows this
202
mobile mapping method can now be applied by analogy to track locations inside buildings, using
203
an indoor positioning system (IPS). In conjunction with traditional activity logs for personal
204
exposure assessment, it can track or identify omitted location changes (e.g., leaving kitchen
205
temporarily while cooking) and provide more detailed exposure assessments (e.g., distances and
206
directions from sources).
207
Stationary monitors have been used for indoor air quality spatial mapping. Unlike this
208
“Eulerian” approach (tracking concentrations at discrete fixed points), the “Lagrangian” (mobile)
209
monitoring method tested here offers a way to measure continuous spatial profiles of
210
concentration. Our mapping experiment detected leakage of incense emissions indoors, when the
211
investigator lingered nearby. Future studies examining how well this mobile sensing method can
212
identify sources in real-world situations (e.g., walking by a mild secondhand smoke intrusion
213
location) would be valuable.
214
The methodology presented could potentially be useful for large occupational indoor settings
215
where spatially tracking workers’ exposures or accidental air toxic leaks are critical to ensure
216
occupational safety and health. It can also benefit future research characterizing occupant7 ACS Paragon Plus Environment
Environmental Science & Technology Letters
217
environment interactions - this could be useful for smart building applications, such as
218
ventilation, light, and appliance automation based on recursive patterns, to achieve energy
219
conservation and human health protection in the built indoor environment.
220
ACKNOWLEDGMENTS
221
The authors thank the TomKat Center for Sustainable Energy at Stanford University for funding
222
this seed research and Dr. Ram Rajagopal for his expert advice on the wireless sensor system.
223 224
REFERENCES
225
(1) McBride, S. J.; Ferro, A.; Ott, W. R.; Switzer, P.; Hildemann, L. M. Investigations of the
226
Proximity Effect for Pollutants in the Indoor Environment. Journal of Exposure Analysis and
227
Environmental Epidemiology 1999, 9, 602-621.
228 229
(2) Ferro, A. R.; Kopperud, R. J.; Hildemann, L. M. Elevated Personal Exposure to Particulate
230
Matter from Human Activities in a Residence. Journal of Exposure Analysis and
231
Environmental Epidemiology 2004, 14, S34-S40.
232 233
(3) Cheng, K. C.; Acevedo-Bolton, V.; Jiang, R. T.; Klepeis, N. E.; Ott, W. R.; Fringer, O. B.;
234
Hildemann, L. M. Modeling Exposure Close to Air Pollution Sources in Naturally Ventilated
235
Residences: Association of Turbulent Diffusion Coefficient with Air Change Rate.
236
Environmental Science and Technology 2011, 45, 4016-4022.
237 238
(4) Acevedo-Bolton, V.; Cheng, K. C.; Jiang, R. T.; Ott, W. R.; Klepeis, N. E.; Hildemann, L.
239
M. Measurement of the Proximity Effect for Indoor Air Pollutant Sources in Two Homes.
240
Journal of Environmental Monitoring 2012, 14, 94-104.
241 242
(5) Zhao, T.; Nguyen, C.; Lin, C. H.; Middlekauff, H. R.; Peters, K.; Moheimani, R.; Guo, Q;
243
Zhu, Y. Characteristics of Secondhand Electronic Cigarette Aerosols from Active Human
244
Use. Aerosol Science and Technology 2017, 51, 1368-1376.
245
8 ACS Paragon Plus Environment
Page 8 of 14
Page 9 of 14
Environmental Science & Technology Letters
246
(6) Acevedo-Bolton, V.; Ott, W. R.; Cheng, K. C.; Jiang, R. T.; Klepeis, N. E.; Hildemann, L.
247
M. Controlled Experiments Measuring Personal Exposure to PM2.5 in Close Proximity to
248
Cigarette Smoking. Indoor Air 2014, 24, 199-212.
249 250
(7) Cheng K. C.; Zheng, D.; Hildemann, L. M. Impact of Mechanical Mixing on Air Pollutant
251
Concentration near Sources in Unoccupied Rooms: Connecting Peak Exposure with Energy
252
Input. Environmental Science and Technology 2018 (submitted).
253 254
(8) Patel, S.; Li, J.; Pandey, A.; Pervez, S.; Chakrabarty, R. K.; Biswas, P. Spatio-temporal
255
Measurement of Indoor Particulate Matter Concentrations Using a Wireless Network of
256
Low-cost Sensors in Households Using Solid Fuels. Environmental Research 2017, 152, 59-
257
65.
258 259
(9) Li, J.; Li, H.; Ma, Y.; Wang, Y.; Abokifa, A. A.; Lu, C.; Biswas, P. Spatiotemporal
260
Distribution of Indoor Particulate Matter Concentration with a Low-cost Sensor Network.
261
Building and Environment 2018, 127, 138-147.
262 263
(10) Thomas, G. W.; Sousan, S.; Tatum, M.; Liu, X.; Zuidema, C.; Fitzpatrick, M.; Koehler, K.
264
A.; Peters, T. M. Low-cost, Distributed Environmental Monitors for Factory Worker Health.
265
Sensors 2018, 18, 1441.
266 267 268
(11) Koyuncu, H; Yang, S. H. A Survey of Indoor Positioning and Object Locating Systems. International Journal of Computer Science and Network Security 2010, 10, 121-128.
269 270
(12) Brena, R. F.; García-Vázquez, J. P.; Galván-Tejada, C. E.; Muñoz-Rodriguez, D.; Vargas-
271
Rosales, C.; Frangmeyer, J. Evolution of Indoor Positioning Technologies: A Survey.
272
Journal of Sensors 2017, 2630413.
273 274
(13) Wang,Y.; Li, J.; Jing, H.; Zhang, Q.; Jiang, J.; Biswas, P. Laboratory Evaluation and
275
Calibration of Three Low-cost Particle Sensors for Particulate Matter Measurement. Aerosol
276
Science and Technology 2015, 49, 1063-1077. 9 ACS Paragon Plus Environment
Environmental Science & Technology Letters
277 278
(14) Jiang, R. T.; Acevedo-Bolton, V.; Cheng, K. C.; Klepeis, N. E.; Ott, W. R.; Hildemann, L.
279
M. Determination of Response of Real-time SidePak AM510 Monitor to Secondhand Smoke,
280
Other Common Indoor Aerosols, and Outdoor Aerosol. Journal of Environmental
281
Monitoring 2011, 13, 1695-1702.
282 283
(15) Dacunto, P. J.; Cheng, K. C.; Acevedo-Bolton, V.; Jiang, R. T.; Klepeis, N. E.; Repace, J. L.;
284
Ott, W. R.; Hildemann, L. M. Real-time Particle Monitor Calibration Factors and PM2.5
285
Emission Factors for Multiple Indoor Sources. Environmental Science: Processes & Impacts
286
2013, 15:1511-1519.
287 288 289
(16) Singer, B. C.; Delp, W. W. Response of Consumer and Research Grade Indoor Air Quality Monitors to Residential Sources of Fine Particles. Indoor Air 2018, 28, 624-639.
290 291 292
(17) Manikonda, A.; Zíková, N.; Hopke, P. K.; Ferro, A. R. Laboratory Assessment of Low-cost PM Monitors. Journal of Aerosol Science 2016, 102, 29-40.
293 294 295
(18) Sousan, S.; Koehler, K.; Hallett, L.; Peters, T. M. Evaluation of Consumer Monitors to Measure Particulate Matter. Journal of Aerosol Science 2017, 107, 123-133.
296 297
(19) Zikova, N.; Hopke, P. K.; Ferro, A. R. Evaluation of New Low-cost Particle Monitors for
298
PM2.5 Concentrations Measurements. Journal of Aerosol Science 2017, 105, 24-34.
299 300
(20) Cheng, K. C.; Acevedo-Bolton, V.; Jiang, R.T.; Klepeis, N. E.; Ott, W. R.,; Hildemann, L.
301
M. Model-based Reconstruction of the Time Response of Electrochemical Air Pollutant
302
Monitors to Rapidly Varying Concentrations. Journal of Environmental Monitoring 2010,
303
12, 846-853.
304 305
(21) Austin, E.; Novosselov, I.; Seto, E.; Yost, M. G. Laboratory Evaluation of the Shinyei
306
PPD42NS Low-cost Particulate Matter Sensor. PLOS ONE 2015, 10, e0141928.
307 10 ACS Paragon Plus Environment
Page 10 of 14
Page 11 of 14
308
Environmental Science & Technology Letters
(22) Kelly, K. E.; Whitaker, J.; Petty, A.; Widmer, C.; Dybwad, A.; Sleeth, D.; Martin, R.;
309
Butterfield, A. Ambient and Laboratory Evaluation of a Low-cost Particulate Matter Sensor.
310
Environmental Pollution 2017, 221, 491-500.
311 312
(23) Pummakarnchana, O.; Tripathi, N..; Dutta, J. Air Pollution Monitoring and GIS Modeling: A
313
New Use of Nanotechnology Based Solid State Gas Sensors. Science and Technology of
314
Advanced Materials 2005, 6, 251-255.
315 316
(24) Steinle, S.; Reis, S.; Sabel, C. E. Quantifying Human Exposure to Air Pollution – Moving
317
from Static Monitoring to Spatio-temporally Resolved Personal Exposure Assessment.
318
Science of the Total Environment 2013, 443, 184-193.
319 320
(25) Steinle, S.; Reis, S.; Sabel, C. E.; Semple, S.; Twigg, M. M.; Braban, C. F.; Leeson, S. R.;
321
Heal, M. R.; Harrison, D.; Lin, C., Wu, H. Personal Exposure Monitoring of PM2.5 in Indoor
322
and Outdoor Microenvironments. Science of the Total Environment 2015, 508, 383-394.
323
11 ACS Paragon Plus Environment
Environmental Science & Technology Letters
324 325
Figure 1(a) and (b)
326 327 328 329 330 331
Figure 1(a) Mobile air monitoring setup with a pair of collocated sensors (research-grade SidePak and low-cost PTQS). PTQS is connected with a Diymore development board that wirelessly transmitted real-time PM2.5 data to a computer server; and (b) Indoor positioning setup with 12 stationary and 1 mobile Marvelmind ultrasound beacons deployed in an one-bedroom rental apartment in Santa Clara, CA. Location data were wirelessly transmitted in real-time, from the beacons to the computer server.
12 ACS Paragon Plus Environment
Page 12 of 14
Page 13 of 14
332
Environmental Science & Technology Letters
Figure 2(a)-(c)
333 334 335 336 337
Figure 2 (a) 1-s concentration time series of PM2.5 measured by the collocated SidePak and PTQS (Figure 1(a)) carried by the occupant. Occupant location history shown as circles on the floor map with PM2.5 levels on a color scale measured by (b) SidePak and (c) PTQS. PTQS data were rescaled using the source-specific scaling factors in Table S2.
338 339
13 ACS Paragon Plus Environment
Environmental Science & Technology Letters
340
Figure 3(a)-(c)
341 Figure 3 (a) Moving trajectories of the occupant who performed random walk in the presence of 3 hidden burning incense sources (red stars outside the open living room window, inside the cabinet under the kitchen sink, and inside the storage room in the bedroom). Spatial concentration fields of PM2.5 on a color scale created by 2-D interpolations of (b) SidePak and (c) PTQS measurements collected at 751 positions (black dots in Figure 3(a)).
14 ACS Paragon Plus Environment
Page 14 of 14