Subscriber access provided by Kansas State University Libraries
Article
Correlation between Atmospheric Boundary Layer Height and Polybrominated Diphenyl Ether Concentrations in Air Nguyen Thanh Dien, Yasuhiro Hirai, and Shin-Ichi Sakai Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.6b03004 • Publication Date (Web): 30 Nov 2016 Downloaded from http://pubs.acs.org on November 30, 2016
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 free 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 accessible to all readers and 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.
Environmental Science & Technology 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 38
Environmental Science & Technology
1
Correlation between Atmospheric Boundary Layer Height and Polybrominated
2
Diphenyl Ether Concentrations in Air
3
Nguyen Thanh Diena, Yasuhiro Hiraia,*, Shin-ichi Sakaia
4
5
a
Environment Preservation Research Center, Kyoto University, Sakyo-ku, Kyoto 606−8501, Japan
6
7 *
Corresponding Author: Yasuhiro Hirai
8
9
Environment Preservation Research Center, Kyoto University, Sakyo-ku, Kyoto 606−8501,
10
Japan
11
TEL +81-75-753-7712, FAX +81-75-753-7710
12
e-mail
[email protected] 13
14
Abstract
15
In this study, we aim to determine the correlation between the height of the atmospheric
16
boundary layer (ABL) and the concentrations of polybrominated diphenyl ether (PBDE)
17
congeners, in an effort to improve comprehension of the atmospheric behavior of PBDEs. We
18
used the PBDE data in air (n=298), which were measured by the Japan Ministry of 1
ACS Paragon Plus Environment
Environmental Science & Technology
Page 2 of 38
19
Environment (JMOE) at 50 sites across Japan during the period 2009−2012. The height of the
20
ABL, which directly affects the PBDE concentrations in the near-surface air, was estimated
21
by employing data retrieved from the Japanese global reanalysis (JRA-55) database, using the
22
parcel and Richardson number method. The ABL has shown a strong inverse relationship
23
with BDE-47 and BDE-99 (p , corresponding to an
144
unstable vertical profile, has to be fulfilled. The altitude range from start to end of this lift
145
is considered statically unstable.32
146
() = () ∙ ( / ()).
147
where T(z) is the temperature at height z (K), P(z) is the pressure at height z (Pa), P0 is the
148
reference pressure (100000 Pa), and z is the geopotential height above ground (m).
149
The Rib is a dimensionless parameter, combining the potential temperature and the vertical
150
wind shear. The Rib is used to identify regions of dynamically unstable air.32 The air
151
compartment has variable heights to make provision for the formation of the convective
152
mixed layer during the daytime and the nocturnal boundary layer and residual layer at night.32
153
During the day, as the ground heats up, the air rises and drives convective mixing. At night,
154
the air layers near the surface cool much faster, leading to a temperature inversion, which
155
prohibits convective mixing. We considered a value of 0.22 for the critical Richardson
156
number under unstable condition (convective mixed layer in daytime), and a value of 0.33
157
under stable condition (nocturnal boundary layer at night).25,33,34 The Richardson number
158
(Rib) is defined by
159
= ∙ ()
()
()
Eq 1
∙
Eq 2 9
ACS Paragon Plus Environment
Environmental Science & Technology
Page 10 of 38
160
where g is the acceleration because of gravity (9.81 m·s-2), !"# is the virtual temperature
161
near the surface, !"() is the virtual potential temperature at height z, and v(z) and u(z) are
162
the horizontal components of the wind velocity at z. Therefore, the virtual potential
163
temperature, mixing ratio, and virtual temperature near the surface were calculated as shown
164
in Eqs 3−5
165
!" () = () ∙ .∙%
166
'() = (#()/(1 − (#())
Eq 4
167
!"# = (2,) ∙ (1 + 0.61 ∙ (#(2,))
Eq 5
168
where w(z) is the mixing ratio, Hs(z) is the specific humidity, Hs(2m) is the specific humidity
169
at 2 m, and T(2m) is the temperature at 2 m (K).
170
Subsequently, the ABLs calculated by Rib and PM were subtracted from the geopotential
171
height of the surface to estimate properly the effective height of the mixing air that dilutes the
172
air pollutants. Since atmospheric PBDEs were continuously monitored for three or seven days
173
per sample, the relevant ABL values were estimated as the average of the daytime and
174
nighttime for a sampling period, and were subsequently used in the regression analysis. The
175
limitation of our study is that the ABL was calculated only at four specific times: 9 AM, 3 PM,
176
9 PM, and 3 AM. Although this could cause the biases, estimating an ABL value is preferred
177
for an air model rather than assuming a fixed ABL value. Therefore, we think that our
$() .
$()&
∙ ( / ()).
Eq 3
10
ACS Paragon Plus Environment
Page 11 of 38
Environmental Science & Technology
178
estimation method is still usable for analyzing the ABL.
179
2.3. Tobit regression model for atmospheric PBDE concentrations with censored data
180
In environmental science, we take care of the left-censoring (censoring from below), since the
181
trace substances in the air, soil, and water are usually presenting the concentrations that are
182
lower than limit of detection (LOD). In Japan, the JMOE data showed the variations of LODs
183
with time for PBDE congeners (Table S2). The Tobit model, a censored regression model, is
184
designed to estimate linear relationships between variables when there is censoring in the
185
dependent variable.35 In principle, values that fall at or below some threshold are censored.
186
The Tobit model uses the censored data (non-detection) and the uncensored data (detection) in
187
a regression procedure. In this model, if the true concentration C* is larger than the LOD
188
value, then the observed concentration C is equal to C*. Otherwise, if the true concentration
189
C* is less than the LOD value, then the observed concentration C is censored to LOD as
190
shown below
191
if C* > LOD, C = C*
192
if C* ?)3,5, + 7@ ln( A + 1)3 +
199
7B C5 + D3 + E3,5,
200
where C is the concentration of PBDE congeners (pg m-3); Toutdoor is the outdoor temperature
201
(K); Ra is the rainfall rate (mm∙h-1); PD is the population density (person∙km-2); ABL is the
202
atmospheric boundary layer height (m); Y is the year (2009−2012); D is a random effect of
203
50 sampling sites across Japan; E is an error term that follows normal distribution with mean
204
0 and variance F ; a is an intercept; b1,b2,b3,b4, and b5 are regression coefficients; i is the
205
sampling location; y is the year; and s is the season.
206
The Tobit model was run by the function of panel analysis on the censored model, with the
207
left censoring limit using the Stata 13 software (StataCorp LP, Texas, USA).
208
In order to clarify the relationship between the ABL and rainfall, we filtered the data into zero
209
rainfall (sunny day) and with rainfall (rainy day). Subsequently, we conducted the Tobit
210
regression analysis (so-called filtered model) on a rainy dataset by Eq 6 and on a sunny
211
dataset by Eq 6 without the term ln(Ra+1).
212
In another analysis, we performed a quantitative test to determine the correlation strength
213
between the ABL and PBDE congeners by using a dummy variable (so-called dummy model)
Eq 6
12
ACS Paragon Plus Environment
Page 13 of 38
Environmental Science & Technology
214
for sunny/rainy day (with 0 and 1), as seen in Eq 7:
215
ln 23,5, = 6 + 7 /89:88; 3,5, + 7 ln( A + 1)3 + 7< ln(=>?)3,5, ∗ (1 − HI,,J6K) +
216
7@ ln(=>?)3,5, ∗ HI,,J6K + 7B C5 + D3 + E3,5,
217
where dummyRain with rainy day=1 and sunny day=0; b3 is a coefficient of ABL in the sunny
218
day and b4 is a coefficient of ABL in the rainy day.
219
3. Results and discussion
220
3.1. Site-specific and time-dependent ABL relevant to PBDE sampling
221
As the sampling time for the ABL data (corresponding to the PBDE sampling time) was only
222
during early autumn and winter, at specific intervals (three or seven days per sample), the
223
estimated ABLs would probably not show all the relevant information on the day/night,
224
seasonal, and interannual timescales. Consequently, at first, over a period of four years, we
225
analyzed the monthly ABLs for various representative sampling sites on land and sea to
226
determine the trends and to confirm our estimation methods of ABL by comparison with the
227
previous studies.
228
The data relevant to day/night of the urban air (Tokyo metropolitan and Osaka city) showed a
229
distinct ABL pattern, namely, this value was much higher during the daytime, as presented in
230
Figure 1 (a) and (b). On the other hand, no distinct pattern was observed in the marine ABL
231
(the sea around the Chichijima and Okinawa islands), as presented in Figure 1 (c) and (d). The
Eq 7
13
ACS Paragon Plus Environment
Environmental Science & Technology
Page 14 of 38
232
contrast between the land (urban) and the sea is attributable to there being a large daily
233
exchange of heat and mass between the ABL and the free atmosphere over the land; while,
234
over the sea, mixing occurs primarily by turbulent entrainment.36 As shown in Figure 1, the
235
seasonal changes in the nighttime ABL (nocturnal boundary layer) are weaker than are those
236
of the daytime ABL (convective mixed layer), particularly for urban sites. We therefore paid
237
attention to the convective daytime ABL for discussing the seasonal pattern. As a result, the
238
ABL height shows the differing seasonal trends for the land and marine sites. As regards the
239
urban sites in Tokyo and Osaka, the monthly mean of the ABLs during the warm season
240
(spring and late summer) was higher than was that during the cold season (winter) (Figure 1
241
(a) and (b) in main text, Figure S1 (a) and (b) in Supporting Information). In contrast, the
242
ABLs for the marine sites (Chichijima and Okinawa) were estimated as higher during the cold
243
season (winter) in comparison with the warm season (summer and autumn) (Figure 1 (c) and
244
(d), Figure S1 (c) and (d)). Furthermore, Figure S1 indicates a consistent seasonal pattern for
245
ABLs during 2009−2012 for both land and sea sites.
246
In a previous study, using a micro pulse lidar, Chen et al. (2001)37 have measured two ABL
247
peaks at the urban area of Tsukuba that occurred in the early spring and autumn during the
248
period 1999−2000. In addition, the seasonal pattern at a suburban site near Paris and at urban
249
Hong Kong indicated that the ABL was higher during the warm period (autumn, spring, and 14
ACS Paragon Plus Environment
Page 15 of 38
Environmental Science & Technology
250
summer) compared with the cold period (winter).26,28 Pal and Haeffelin (2015)26 have
251
measured diurnal (hourly) and seasonal (monthly) patterns at a suburban site near Paris
252
(Palaiseau: 2.208°E and 48.713°N, 160 m above sea level) using lidar equipment. They found
253
that the average of the daily maximum values of ABL (ABL_max) was highest during the
254
summer (maximum in July) and lowest during the winter (minimum in January). We
255
conducted a comparison between the one-year ABL measurement (year 2009) from Pal and
256
Haeffelin (2015)26 and the ABL estimation from the JRA-55 reanalysis data for Palaiseau.
257
The results pertaining to the highest ABL in summer and the lowest ABL in winter presented
258
good agreement between the measurements of Pal and Haeffelin26 and our estimation, as seen
259
in Figure S2. The slightly lower ABL_max values from our estimation compared with the
260
measurements of Pal and Haeffelin were attributed to the limited number (four) of data points
261
per day in the JRA-55 data (1 AM, 7 AM, 1 PM, and 7 PM; France time zone equal to
262
UTC+1). On the other hand, Kuribayashi et al (2011)27 observed that the ABL was high in
263
winter and low in summer over Okinawa Island and the East China Sea (marine sites), using
264
continuously measured radiosonde and lidar data from March 2008 to February 2010. This
265
result was consistent with our ABL estimation in Chichijima and Okinawa (marine sites),
266
using JRA-55 data. Therefore, these agreements with previously observed ABL data
267
enhanced the credibility of our ABL estimates, using reanalysis data. Subsequently, we 15
ACS Paragon Plus Environment
Environmental Science & Technology
Page 16 of 38
268
present the site-specific and time-dependent ABL relevant to PBDE sampling.
269
The ABL distribution plot for all the land (38) and sea (12) sites is presented in Figure 2 (a).
270
Figure 2 (a) indicates that the 25% and 75% percentiles of ABL for the warm period were at
271
537 and 846 m (median 681 m), whereas, for the cold period, they were at 400 and 1013 m
272
(median 667 m). The global planetary boundary layer was estimated by von Engeln and
273
Teixeira (2013)30 by using the reanalysis data of the European Centre for Medium Range
274
Weather Forecasts (ECMWF). This research30 has shown that over Japan, between the 27th
275
and 42nd parallels north, the boundary layer height ranged between 700 and 1400 m. This
276
result is in agreement with our estimated ABL, using the JRA-55 database. Figure 2(a)
277
indicates that the ABL values during the warm season were only slightly higher than were
278
those during the cold season. The reason for the seasonal variation being smaller is the
279
differing ABL patterns on land and sea, as discussed earlier. Accordingly, as indicated in
280
Figure 2 (b), clearly higher ABL values were found at 38 sites in urban and rural areas (land)
281
during the warm season in comparison with the cold season. On the other hand, different
282
results were produced for 12 sites at a nearby small island (sea), as shown in Figure 2 (c). The
283
day and night patterns of the ABLs for land and sea, land (only), and sea (only) are presented
284
in Figure S3 (a) and (b). In addition, the detailed estimation of site-specific and
16
ACS Paragon Plus Environment
Page 17 of 38
Environmental Science & Technology
285
time-dependent ABLs for the 50 sites is shown in Table S3. In the next section, we present an
286
examination of the correlation between the ABL and temperature.
287
3.2. Correlation between the ABL and temperature
288
We investigated the relationship between the ABL and temperature by separating three
289
datasets, including land and sea sites, land sites, and sea sites because of the different ABL
290
patterns on land and sea (as discussed in section 3.1). The histograms of the ABLs for the
291
three categorized sites were plotted, as shown in Figure S4 (a), (b), and (c) in the Supporting
292
Information. For land and sea sites, there was a weakly positive correlation (n=298,
293
coefficient correlation=133, p=0.711, R2=0.0005) between the ABL (lnABL) and temperature
294
(1/T) (Figure S5 (a)). In a similar analysis, an extremely weakly positive correlation (n=218,
295
coefficient correlation=25.28, p=0.954, R2=0.00002) was found for land sites (Figure S5 (b)).
296
On the other hand, for sea sites, there was a significantly positive correlation (n=80,
297
coefficient correlation=1665, p