PUBLICATIONS
Journal of Geophysical Research: Atmospheres
RESEARCH ARTICLE
10.1002/2017JD026803
Key Points:
• High concentrations of acidic
components in UHR indicated the
high concentration of acidic
substances in the ambient air just
before UHR
• HO had a large contribution to the
NO2 oxidative formation before the
formation of UHR
• HNO3 that was formed just before
UHR immediately reacted with NH3 to
form NH4NO3, which could
subsequently be scavenged as CCN
by UHR
Supporting Information:
• Supporting Information S1
• Table S1
Correspondence to:
R. Uchiyama,
k-u.ryunosuke@suou.waseda.jp
Citation:
Uchiyama, R., H. Okochi, N. Katsumi,
and H. Ogata (2017), The impact of air
pollutants on rainwater chemistry
during “urban-induced heavy rainfall” in
downtown Tokyo, Japan, J. Geophys.
Res. Atmos., 122, 6502–6519,
doi:10.1002/2017JD026803.
Received 17 MAR 2017
Accepted 12 JUN 2017
Accepted article online 14 JUN 2017
Published online 26 JUN 2017
©2017. American Geophysical Union.
All Rights Reserved.
UCHIYAMA ET AL.
The impact of air pollutants on rainwater chemistry
during “urban-induced heavy rainfall”
in downtown Tokyo, Japan
Ryunosuke Uchiyama1 , Hiroshi Okochi1 , Naoya Katsumi1 , and Hiroko Ogata1
1
Graduate School of Creative Science and Engineering, Waseda University, Shinjuku-ku, Tokyo, Japan
Abstract In order to clarify the impact of air pollution on the formation of sudden and locally distributed
heavy rain in urban areas (hereafter UHR = urban-induced heavy rain), we analyzed inorganic ions in
rainwater samples collected on an event basis over 5 years from October 2012 to December 2016 in Shinjuku,
Tokyo. Hourly rainfall amounts and wet deposition fluxes of acidic components (the sum of H+, NH4+, NO3
,
and nonsea-salt SO42
) in UHR were 13.1 and 17.8 times larger than those in normal rainfall, respectively,
indicating that large amount of air pollutants were scavenged and deposited by UHR with large amounts of
rainfall. The level of air pollutants, such as NO2, SO2, and potential ozone, in the ambient air increased just
before the formation of UHR and decreased sharply at the end of the UHR event. These results indicate that
NO2, which was formed secondarily by oxidants, was further oxidized by HO radicals and formed HNO3 just
before the formation of UHR, which was subsequently scavenged by UHR.
1. Introduction
Heavy rain events of more than 50 mm/h were frequently observed during the summer over two decades in
heavily urbanized area in Japan [Sato and Takahashi, 2000]. Increased heavy rain events in the summer have
been also reported for urban areas all over the world [Changnon, 1968; Atkinson, 1971; Jauregui and Romales,
1996; Mikami et al., 2005; Goswami et al., 2006; Dou et al., 2015] since the latter part of the twentieth century.
This phenomenon involves convective heavy rains that occur suddenly and locally in an urban area, which we
here call urban-induced heavy rain (UHR). An example of UHR in Japan occurred in 1999 in Nerima Ward,
Tokyo, with 91 mm of rainfall for 1 h and 131 mm/h at its peak [Mikami et al., 2005]. The UHR formed sud-
denly, and the large amount of rainfall over a short period of time caused severe flooding when the amount
of rainfall exceeded the capacity of the sewage system and draining canals [Kawasaki and Meguro, 2011]. The
UHR, with a rainfall intensity of 131 mm/h in 1999, caused urban flooding because the capacity of the sewage
system was 50 mm/h in Tokyo and killed a man who was in a storage basement [Mikami et al., 2005]. Urban
flooding has occurred not only in Japan but also in urban areas all over the world [Smith et al., 2007; IPCC2007,
2007; Huong and Pathirana, 2013; Bertoldo et al., 2016]. Elucidating the formation mechanism of UHR has
been an urgent issue in atmospheric environment.
Urban areas tend to be 2–4°C warmer than the surrounding areas because too many concrete buildings and
asphalt roads in urban areas cause the increase of heat emissions from air conditioners and automobiles
[Chandler, 1965; Bornstein, 1968; Oke and East, 1971; Kim, 1992; Wang et al., 1990; Robaa, 2003; Yamato
et al., 2011]. This phenomenon is called an urban heat island (UHI). The transportation of the ground-level
heated air by a UHI to the upper boundary layer draws the air mass from its surroundings at the ground level.
The humid air mass is drawn from the ocean into the urban area (sea-breeze convergence) and then trans-
ported to the upper layers by upwind in the urban area. The temperature of the upper atmosphere is very
low, thus water vapor condenses and continuously forms cloud droplets. Therefore, a UHI could accelerate
a UHR event [Mikami et al., 2005; Fujibe, 2004]. Numerical simulations indicate that changes in the
distribution of anthropogenic heat in urban areas greatly affect the locations and amounts of rainfall [Ito
et al., 2006; Lei et al., 2008; Miao et al., 2011]. However, there are still uncertainties as to the formation mechan-
ism of UHR, although numerical simulations, real-time monitoring, and meteorological data analysis have
been performed focusing on UHI [Shiraki et al., 2009; Mikami et al., 2005; Fowler and Hennessy, 1995;
Rajeevan et al., 2008; Bertoldo et al., 2016].
The impacts of UHI include elevated emissions of air pollutants and greenhouse gases, such as sulfur dioxide
(SO2), nitrogen oxide (NOx), volatile organic compounds (VOCs), particulate matter, carbon monoxide (CO),
HEAVY RAIN FORMATION AND AIR POLLUTANTS 6502
 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

carbon dioxide (CO2), and mercury (Hg). In addition to energy-related elevated emissions, high temperatures
can directly increase the rate of ground-level formation of ozone [Kheim et al., 2010; Ooka et al., 2011], which
is formed when NOx and VOCs react in the presence of sunlight and hot weather. Further oxidation of NO2
and SO2 by ozone and hydroxyl radicals form HNO3 and H2SO4, which react with NH3 to form hygroscopic
aerosols such as NH4NO3, NH4HSO4, and (NH4)2SO4. Hygroscopic aerosols formed in urban area are trans-
ported upward by strong updrafts, mix with moisture from the ocean, then act as cloud condensation nuclei
[Bently et al., 2010]. Mikami et al. [2005] showed that convergence of sea breezes in local areas with elevated
temperatures occurs before intensive heavy rainfall. Wang et al. [2016] simulated the impacts of anthropo-
genic aerosols on summer precipitation over the Beijing-Tianjin-Hebei urban agglomeration in China. They
reported that anthropogenic aerosols enhanced summer rainfall and ascending motion, triggering more
cloud formation and convective systems in urban areas. Lacke et al. [2009] found that instabilities were
greater and mixing layers shallower on days with high aerosol concentrations, resulting in greater precipita-
tion amounts in most parts of the region in metropolitan Atlanta from 2003 to 2004. These studies indicated
that anthropogenic aerosols formed in urban areas could contribute to the formation of convective clouds
and increase the deposition of air pollutants by rainfall in urban areas. However, the formation mechanism
of UHR is still unknown.
In order to clarify the impact of air pollution on the formation of UHR, we analyzed the inorganic ion concen-
trations in rainwater collected on an event basis in Shinjuku, which is representative of downtown of Tokyo,
Japan, over 5 years, from October 2012 to December 2016. A case study analysis of one urban heavy rain
event collected on 24 July in 2014 was performed by combining rainwater chemistry and available data of
air pollutants in detail. Our results provide useful information on the formation mechanism of UHR.

2. Materials and Methods

2.1. Sampling Site and Method
Rainwater samples were collected on the rooftop of a building (~65 m high) on the Nishi-Waseda campus
(35.7°N, 139.7°E), Waseda University, in Shinjuku, Japan (Figure 1a). Shinjuku is a centrally located ward in
Tokyo, and more than 300,000 people, including 40,000 foreigners, live there. The area around Shinjuku sta-
tion, which is the center of Shinjuku, is a huge business, commercial, and entertainment area. The sampling
site is located along Meiji Street [36,000 cars/d; Toshima City Office [2013]] and about 2 km north of Sinjuku
station, which is located at about 6 km northwest of Tokyo Bay. The Keihin industrial zone, which has devel-
oped into the largest industrial area in Japan, is also located there. Southerly winds bring various air pollu-
tants emitted from business, commercial, and industrial areas, along with humid sea air.
The 100 rainwater samples were collected on an event basis over 5 years, from October 2012 to December
2016, using a polyethylene bottle (10 L) with a polyethylene funnel (diameter: 300 mm). Prior to use, the sam-
pler was carefully washed with Milli-Q water and dried. The collector was placed manually just before each
event (approximately 30–60 min before rainfall) and collected immediately after the event (on the day or
the next day). The rainfall beginning times were checked using microwave radar charts for every 5 min
obtained from “Tokyo Amesh.” Rainfall from precipitation events of less than 0.5 mm was not collected.

2.2. Analytical Method

The rainwater samples were filtered with a membrane filter (pore size: 0.45 μm) to remove any particulate
matter. The filtrates were stored in clean polypropylene bottles in a refrigerator at 4°C until chemical and
stable isotopic analyses. The pH was measured by a pH meter (MA325, Mettler Toledo). Calibration was per-
formed before measurement using standards buffer solutions of pH 6.86 and 4.01 (Wako Pure Chemical,
Osaka, Japan). Major cations (NH4+, Na+, K+, Mg2+, and Ca2+) and anions (Cl
, NO3
, and SO42
) were mea-
sured by ion chromatography. A Dionex ICS-1000 (separation column: Dionex Ion Pac® CG12A RFIC™
4 × 250 mm; guard column: Dionex Ion Pac® CG12A RFIC™ 4 × 50 mm; eluent: 18 mM methanesulfonic acid;
suppressor: Thermo CERS) was used for analysis of major cations, while a Dionex DX-320 (separation column:
Dionex Ion Pac™ AG4A-SC RFIC™ 4 × 250 mm; guard column: Dionex Ion Pac™ AG4A-SC RFICTM 4 × 50 mm;
eluent: 1.8 mM Na2CO3 + 1.7 mM NaHCO3; suppressor: Dionex ASRS™) was used for analysis of major anions.
The standards used for calibration were prepared from stock solutions of each ion (1000 mg/L, Wako Pure
Chemical, Osaka, Japan). Dissolved inorganic carbon was measured by a TOC meter (TOC-VCSH,

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6503

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

Figure 1. The positional relationship between our sampling point and observatories. (a) Sampling point in this study,
(b) Observatory of Tokyo station (JMA), (c) Observatory of Nerima station (JMA), and (d) Observatory of Shinjuku
(Shinjuku City).

SHIMADZU). The calibration standards were prepared from stock solutions of C6H4(COOK)(COOH) (Nacalai
Tesque, Kyoto, Japan). All solutions were prepared using Milli-Q water. The ratios of the sum of anions to
the sum of cations (Σanions/Σcations) in 100 rainwater samples were 0.96 ± 0.20, and the correlation coeffi-
cient was 0.98, suggesting that almost all the major inorganic ions were measured (Figure 2).
2.3. Meteorological and Air
Pollutant Data
Meteorological data at 10 min inter-
vals (temperature, relative humidity,
air pressure, wind direction, and wind
speed) and 60 min intervals global
solar radiation amount were obtained
from the Japan Meteorological
Agency (JMA). We consulted the
nearest JMA observatory, which was
Tokyo station (35.7°N, 139.8°E, ~25 m;
Figure 1b). Rainfall amount data at
10 min intervals and air pollutant
concentration datasets with 60 min
intervals were obtained from the
observatory at Shinjuku City (35.7°N,
139.7°E, ~10 m; Figure 1d). The dis-
Figure 2. The relationship between cations and anions in the rainwater sam-
ples in Shinjuku, Tokyo during October 2012 to December 2016. The ratio of
tances from our sampling point to
anions to cations is 0.96 ± 0.2, and the correlation coefficient is 0.98, sug- the observatories were 4.16 km and
gesting that almost all major inorganic ions were measured. 1.39 km (Figure 1). In addition, to

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6504

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

clarify the rainfall amount distribution, the rainfall data at 10 min intervals were obtained from Toshima, Chuo,
Nakano, and Shibuyabashi observatory of the Bureau of Construction, Tokyo Metropolitan Government (see
Figure 10). In order to classify the heavy rainfall samples, we obtained meteorological charts for every 3 h
and infrared imagery from weather cameras every 1 h from the Japan Weather Association.

2.4. Classification of Heavy Rain

According to the JMA, heavy rain is defined as more than 30 mm of hourly rainfall. However, UHR usually
stops within 1 h; thus, we define heavy rain as a rain event with a rainfall intensity of more than
5 mm/10 min, using 10 min data of rainfall amount from JMA to classify rain events. For classification, we con-
sulted the two nearest JMA observatories: Tokyo station (35.7°N, 139.8°E, ~25 m; Figure 1b) and Nerima sta-
tion (35.7°N, 139.6°E, ~51 m; Figure 1c). We also define normal rain as a rain event with a 10 min rainfall
amount of less than 5 mm. In addition, we classify heavy rain events into three types based on their formation
process: (1) frontal heavy rain (FHR) is heavy rain caused by a front; (2) typhoon heavy rain (THR) is heavy rain
caused by a typhoon; (3) UHR is sudden and locally distributed heavy rains caused by neither a front nor a
typhoon. Heavy rains were distinguished according to 180 min meteorological charts and 60 min weather
camera imagery obtained from the Japan Weather Association. The flow of heavy rain classification is as
follows:
1. When heavy rain occurs, we consult the meteorological charts and classify heavy rain with a typhoon as
THR.
2. If the heavy rain is not caused by a typhoon, using imagery from weather cameras, we check whether the
heavy rain is caused by clouds developed at a front. We regard heavy rain caused by the front as FHR.
3. Heavy rain not caused by either a typhoon or a front is regarded as UHR.
Figure 2 shows examples of the meteorological charts and weather camera imagery for the three types of
heavy rains. Figure 3a shows the meteorological chart and weather camera image for the THR that occurred
at 15:00 JST on 30 August 2016. Cloud development by a typhoon over Japan was confirmable from Figure 3
a. Figure 3b shows the meteorological chart and weather camera image of the FHR that occurred at 09:00 JST
on 13 September 2016. Large cloud development by an autumn rain front above Japan was confirmable from
Figure 3b. Figure 3c shows the meteorological chart at 15:00 JST and the weather camera image at 14:00 JST
on 14 July 2016, the UHR formation time. According to the meteorological chart, there was no approaching
typhoon. A stationary front lay to the south of Japan, but the weather camera showed no clouds developed
by this front. Based on the method described above, we classified the 100 rainwater samples into 68
normal rain samples, 15 UHR samples, 6 THR samples, and 11 FHR samples. The date and time of heavy
rain occurrence extracted from JMA rainfall amount data were shown in Tables S1 and S2 in the
supporting information.

3. Results and Discussion

3.1. Chemical Characteristics of UHR
We have shown that UHR has lower pH, higher concentrations of NO3
and SO42
, and greater hourly rainfall
amounts compared with normal rainfall and THR [Uchiyama et al., 2017]. Adding to the data collected in 2016
and identifying additional frontal heavy rain event (FHR), the mean concentrations of major inorganic ions in
various types of rain events, collected at Shinjuku, Tokyo, on an event basis, from October 2012 to December
2016, are shown in Table 1. The HCO3
concentration was calculated by the equilibrium calculation using
rainwater pH and the concentration of dissolved total inorganic carbon. The concentration of nonsea-salt
(nss) SO42
was 30.8 μeq/L (87% of sulfate) in normal rain, 37.5 μeq/L (97% of sulfate) in UHR, 6.97 μeq/L
(74% of sulfate) in THR, and 17.8 μeq/L (91% of sulfate) in FHR, respectively, showing that most of the sulfate
in UHR originates from anthropogenic sources. The ratio of nitrate concentration to nss-SO42
concentration
in rainwater (hereafter N/S ratio) was 0.954 in normal rain, 1.09 in UHR, 0.44 in THR, and 0.69 in FHR. The high-
est N/S ratio in UHR indicates that nitric acid was the most important acid in UHR. NO3
in rainwater is origi-
nated from the dissolution of HNO3 and nitrate aerosol because of very low Henry’s law constants of nitrogen
oxides (NO and NO2, namely NOx), which are precursors of HNO3 and nitrate aerosol [Seinfeld and Pandis,
2006]. HNO3, which is formed by photochemical reaction of NO2, has a high Henry’s law constant, while
nitrate aerosol such as NH4NO3 and NaNO3 has high hygroscopic properties [Seinfeld and Pandis, 2006], so

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6505

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

Figure 3. The meteorological charts and weather camera imageries for the three types of heavy rains. (a) THR occurred time, 2016/08/30/15:00; (b) FHR occurred
time, 2016/09/13/09:00; and (c) UHR formation time, meteorological chart at 2016/07/14/15:00 and weather camera 2016/07/14/14:00.

they can both dissolve into cloud droplets and raindrops. Thus, these results imply that nitrogen oxides in the
ambient air were abundant, and the photochemical reactions formed HNO3 just before the UHR event.
The THR had lower total concentrations of chemical components than the other types of rain, but the ratios of
sea salt components (Na+ and Cl
) to the total inorganic ions were high (48%, see Figure 5). In addition, the

Table 1. The Mean Concentration of Major Inorganic Ions in Normal Rain, UHR, THR, and FHR During October, 2012, to
December, 2016
Concentration (μeq/L)
+ + + + 2+ 2+

2

2

H Na NH₄ K Mg Ca Cl No₃ So₄ HCO₃ Total Nss-SO₄

Normal (n = 68) 18.4 28.8 32.1 1.65 8.40 21.0 45.1 29.4 35.5 2.14 222 30.8
UHR (n = 15) 43.2 8.12 27.7 1.88 1.62 12.2 10.2 41.0 38.8 0.507 185 37.5
THR (n = 6) 6.93 14.5 9.96 0.683 4.19 4.77 23.7 3.04 9.33 2.40 79.5 6.97
FHR (n = 11) 15.5 10.7 14.8 0.380 2.22 9.26 22.2 12.4 19.6 2.52 110 17.8

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6506

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

Figure 4. (a) The ratio of the concentration of each major ion in UHR to that in normal rain. (b) The ratio of the concentra-
tion of each major ion in FHR to that in normal rain. The dashed line shows the mean (per event) ratio of normal rainfall
amount to that of UHR (and THR). This line means the dilution degree of chemical components by rainfall amount.

THR had the lowest N/S ratio and the highest pH, suggesting that the THR included clean air masses from the
ocean and was only weakly acidified.
Although the FHR had a similar chemical composition to normal rain (see Figure 5), the concentrations were
lower than those of normal rain. UHR could be caused by strong convergence with an urban heat island
[Mikami et al., 2005; Fujibe, 2004], while THR is caused by a typhoon, which is formed over the ocean with
clean air masses. However, the formation process of FHR is similar to that of normal rain; thus, both chemical
compositions could be similar. Because heavy rain events have larger rainfall amounts than normal rain
events, FHR would have a lower chemical concentration because of dilution [Suzuki, 1983].
Figure 4 shows the ratio of the concentration of each major ion in UHR (a) and FHR (b), which suggests the
characteristic chemical composition, to that in normal rain, respectively. The dashed lines in Figure 4 indicate
the ratios of mean rainfall amount in normal rain to that in UHR and FHR, respectively. The differences
between the ion concentrations in normal rain and heavy rain could be explained by dilution if the concen-
tration ratio of each ion was similar to the rainfall amount, indicated by the dashed line in Figure 4. Acidic
components (H+, NH4+, NO3
, and nss-SO42
) had higher concentrations in UHR than in normal rain, while
sea salt components (Na+, Cl
, and Mg2+) had lower concentrations in UHR. Meanwhile, the concentrations
of most components in FHR were as high as those in normal rain (dashed line). This indicates the same
formation process for FHR as for normal rain and the dilution effect (as mentioned above). The ratio of the
concentration of each major ion in THR to that in normal rain was reported by Uchiyama et al. [2017], and they
implied that the sea salt components (Na+, Cl
, and Mg2+) had higher concentrations in THR than in normal
rain, while the acidic components had lower concentrations in THR. These results indicated that acidic
substances were effectively scavenged by UHR.
There are two scavenging mechanisms: in-cloud scavenging and below-cloud scavenging [Schwartz and
Slinn, 1992; Seinfeld and Pandis, 2006; Levein and Cotton, 2009]. In-cloud scavenging of aerosol particles is a
formation process of cloud droplets as cloud condensation nuclei (CCN) in supersaturation conditions,
whereas below-cloud scavenging is the collection of aerosols by falling hydrometeors. In-cloud scavenging
is important for hygroscopic and water-soluble aerosol particles in the range of 0.1–2 μm, which is called
in Greenfield Gap [Greenfield, 1957; Lamb and Verlinde, 2011]. (NH4)HSO4, (NH4)2SO4, and NH4NO3 are
present in this range [Wilson and Suh, 1997], so these aerosol particles prefer to be scavenged by in-cloud
scavenging. Both SO2 and HNO3 could be scavenged by both in- and below-cloud scavenging.
3.2. Deposition Flux and Hourly Rainfall Amount of Inorganic Ions in Normal Rain, UHR, THR, and FHR
Figure 5 shows the total wet deposition flux of inorganic ions in normal rain (n = 68), UHR (n = 15), THR (n = 6),
and FHR (n = 11) with rainfall time per event and hourly mean rainfall amount from October 2012 to

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6507

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

Figure 5. The total wet deposition flux of inorganic ion in normal rain (n = 68), UHR (n = 15), THR (n = 6), and FHR (n = 11)
with rainfall time per event and hourly mean rainfall amount from October 2012 to December 2016.

December 2016. The total wet deposition flux was calculated as in Uchiyama et al. [2017]. The rainfall times of
normal rain, UHR, THR, and FHR were 15.4 h, 1.68 h, 15.9 h, and 26.3 h on average, respectively. The hourly
mean rainfall amounts of normal rain, UHR, THR, and FHR were 0.822 mm/h, 10.8 mm/h, 2.34 mm/h, and
1.65 mm/h, respectively. UHR had the shortest rainfall time, within 2 h, and the strongest hourly mean
rainfall amount, more than 10 mm. The total wet deposition flux in UHR was 11.0 times higher than in
normal rain, 10.8 times higher than in THR, and 11.1 times higher than in FHR. The wet deposition flux of
the sum of acidic components (H+, NH4+, NO3
, and nss-SO42
) in UHR was 17.8, 25.7, and 16.2 times
higher than in normal rain, THR, and FHR, and the rates of those acidic components were higher than the
rate of total wet deposition flux. Therefore, the concentration of acidic components in UHR, i.e., the
concentration of acidic gases and aerosols in ambient air just before the UHR event, was an important
controlling factor in determining the chemical composition and wet deposition flux in UHR.

3.3. Behavior of Air Pollutants in UHR Formation Area

3.3.1. Case Study of UHR on 24 July 2014
Figure 6 shows the radar charts obtained from the Bureau of Sewerage, Tokyo Metropolitan Government
“Tokyo Amesh” on 24 July 2014, when we observed UHR. The red lines in Figure 6 indicate Shinjuku Ward.
UHR formed at the center of Shinjuku Ward around 17:05 JST (a), developed into a huge cumulonimbus
around 17:25 JST (e), and disappeared around 17:45 JST (i) (hereafter, first UHR). This shows that this UHR
developed rapidly within 20 min and disappeared within 40 min of formation. We identified the center of
the first formed region of UHR as the formation point of the UHR and measured the distance from our sam-
pling point using Google Map. Figure 7 shows the six UHR formation points (only samples that could be cor-
rectly identified as the UHR formation point) and the distances from the UHR formation area to the sampling
point. UHR is a locally distributed phenomenon and the UHR case of (2) was the nearest UHR event at our
sample point (1.29 km), so we analyzed the meteorological factors and air pollutant concentrations for the
UHR event on 24 July 2014.
Figure 8 shows the temporal variation of the meteorological factors (air temperature, relative humidity, wind
direction, and wind speed at Tokyo Observatory), 10 min rainfall amount, and concentrations of air pollutants
(SO2, NOx, NO2, NO, Ox, nonmethane hydrocarbon (NMHC), methane, total carbon, suspended particulate
matter (SPM), and CO at Shinjuku Observatory) on 24 July 2014. UHR formed in Shinjuku Ward around 17:00
(Figure 6). Figure 9 shows the meteorological chart at 18:00 JST on 24 July 2014 and the weather camera ima-
gery at 17:00 on 24 July 2014. According to the meteorological chart, there was no approaching typhoon. A
stationary front lay to the north of Tokyo, but the weather camera imagery showed that no cloud was devel-
oped by this front. Although Figure 6 showed that UHR with rainfall intensity of more than 30 mm/h occurred
from 17:15 JST to 17:35 JST, Figure 8 showed only 3 mm/10 min (17:21–17:30 JST). Figure 10 shows the dis-
tribution of rainfall amount in Shinjuku and its surrounding area from 17:00 JST to 17:50 JST on 24 July 2014

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6508

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

Figure 6. The radar charts of an UHR-observed day on 24 July 2014 obtained from Bureau of sewerage Tokyo Metropolitan Government “Tokyo Amesh.” The red lines
locate Shinjuku. (a) UHR formed at Shinjuku around 17:05 JST, (e) developed to the huge cumulonimbus around 17:25 JST, and (i) disappeared around 17:45 JST.
These showed that this UHR developed rapidly in 20 min and disappeared within 40 min from formation.

(first UHR initiation time). According to Figure 10, the surrounding area of Shinjuku Ward did not observe
rainfall, indicating that UHR was distributed so locally that the observational network did not detect the
UHR event. In addition, it is possible that the center of the UHR event had the highest rainfall intensity,
and the center of UHR could be located elsewhere from the Shinjuku observatory.
According to Figure 8, this day also had high rainfall intensity of 5 mm/10 min around 19:30 JST. This was the
contribution of another UHR event (hereafter, second UHR), which formed in Suginami Ward (Figure 10).
Figure 11 shows the radar charts of the second UHR formation and the subsequent development. The red
lines in Figure 11 indicate Shinjuku Ward, and the red circle shows the second UHR. The explanatory notes
are the same as for Figure 6. According to Figure 11, the second UHR was formed in Suginami Ward (red circle
in Figure 11a) at around 17:25 JST and reached Shinjuku around 18:30 JST. The second UHR was caused by a
very large cumulonimbus. The first UHR formed nearest our sampling point (Shinjuku); thus, we focused on
the first UHR.

Figure 7. The six UHR formation point (only samples which were correctly identified the UHR formation point) and the
distances from UHR formation area to the sampling point. (1) Sampling point in this study, (2) UHR in 24 July 2014
(1.29 km, first UHR), (3) UHR in 25 June 2013 (2.58 km), (4) UHR in 14 July 2013 (2.90 km), (5) UHR in 25 June 2014 (6.80 km),
(6) UHR in 20 July 2014 (8.06 km), and (7) UHR in 24 July 2014 (9.07 km, second UHR).

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6509

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

Figure 8. The temporal variation of the meteorological factors (temperature, relative humidity, wind direction, and wind
speed; at Tokyo observatory), precipitation amount every 10 min (Nerima observatory), and the air pollutants concentra-
tions (SO2, NOx, NO2, NO, Ox, nonmethane hydrocarbon, methane, total carbon, SPM, and CO; at Shinjuku observatory)
in 24 July 2014.

Before the formation of the first UHR (14:00–17:00), concentrations of SO2 and NOx were high; in particular,
the NO2 concentration increased. The concentration of CO, which is an indicator of anthropogenic burning
[Liang et al., 2004], also increased; thus, the concentrations of acidic gases could have included the
contribution of primary emissions from anthropogenic burning, such as automobiles and factories. The
high concentrations of acidic components (H+, NH4+, NO3
, and nss-SO42
) in UHR could reflect these high

Figure 9. The meteorological chart at 2014/07/24/18:00 JST and the weather camera imagery at 2014/07/24/17:00 JST, the
UHR formation time. According to the meteorological chart, there is no approaching typhoon. A stationary front lied to
the north of Tokyo; however, weather camera shows that no cloud was developed by this front.

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6510

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

concentrations of air pollutants in

ambient air before the formation of
UHR. In addition, the air temperature
and NMHC concentration, which
contribute to the formation of oxi-
dants [Saiki et al., 1986], were high,
and the Ox concentration generally
rises or falls during this time
(14:00–17:00). These results imply
that the oxidation of NO to NO2
and further oxidation of NO2 to
HNO3 could occur.
3.3.2. Primary Emission NO2 and
Secondary Formation of NO2
There are two sources of NO2 in
ambient air: (1) NO2 which is emitted
directly from anthropogenic burning,
Figure 10. The distribution of rainfall amount in Shinjuku and its surround- such as automobiles and factories
ing area from 2014/07/24/17:00 JST to 2014/07/24/17:50 JST (first UHR (hereafter; NO2(e)); and (2) NO2 which
initiation time). is secondarily formed by oxidation of
NO in the atmosphere (hereafter,
NO2(r)) [Uno et al., 1989]. The oxidation of NO is expressed by the following [Okochi, 2014]:

NO þ O3 → NO2 ðrÞ þ O2 (1)

Equation (1) indicates that NO is oxidized by O3 and forms NO2(r). If the O3 concentration is zero, the NO/NOx
ratio indicates the ratio of NO in NOx emitted from primary sources [Uno et al., 1989]. Therefore, we calculated
the NO/NOx ratio when the O3 concentration was zero in July 2014. There was no available observational O3
data for Shinjuku Ward, but we can use the observational Ox data which is close to O3 [Ohara, 2010]. However,
there was no time period when the Ox concentration was zero, so we used the NO/NOx ratio when the Ox
concentration was 2 ppb, the minimum value of Ox in July 2014. The average NO/NOx ratio was 0.7476 when
the Ox concentration was 2 ppb (n = 4). Thus, 75% of NOx could have been emitted as NO and 25% as NO2 in
July 2014. The concentrations of NO and NO2 in Shinjuku in July 2014 are expressed by the following:
½NO ¼ 0:7476½NOx  (2)
½NO2 ðeÞ¼ 0:2524NOx  (3)
½NO2  ¼ ½NO2 ðeÞþNO2 ðrÞ (4)
Thus, the NO2(r) concentration can be expressed as follows:

½NO2 ðrÞ¼NO2
NO2 ðeÞ¼NO2
0:2524NOx  (5)

Naemura and Watanabe [2016] implied that the mean NO/NOx ratio from primary emission in Japan is 0.1,
suggesting that the NO/NOx ratio in Shinjuku in July 2014 was very high. Ohara [2010] indicated that the
NO/NOx ratio from primary emission changes with season and time of day and pointed out that calculation
of the NO/NOx ratio from primary emissions over short time intervals is important to obtain more accurate
values. Figure 12 shows the temporal variation in the NO2(r) and NO2(e) concentrations calculated using
equations (5) and (3), with the amount of solar radiation. The blue zones in Figure 12 indicate the rainfall time
of the day (see Figure 8). Before the UHR event (14:00–17:00), when the NO2 concentration increased, both
NO2(r) and NO2(e) concentrations were high. The oxidant of NO is O3 (equation (1)), but the hydroxyl radical
(HO), which is formed by photodecomposition of O3, is also an important oxidant for NO oxidation. At
11:00–14:00, before the NO2(r) concentration reached a peak, the amount of solar radiation reached its peak.
Therefore, both O3 and HO could contribute largely to the increase in NO2(r) before the first UHR
(at 11:00–14:00).

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6511

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

Figure 11. The radar charts of the second UHR formation and development. The red lines locate Shinjuku City, and the red
circle shows the second UHR. The second UHR was formed at Suginami City (red circle in Figure 11a) at around 17:25
and reached Shinjuku around 18:30. The explanatory note is the same as Figure 6.

3.3.3. Contribution of O3 and HO to NO Oxidation

We used “potential ozone” (PO) [Ohara, 2010] to evaluate the HO contribution to NO oxidation. NO in
ambient is oxidized and forms NO2(r) (equation (1)), and the formed NO2(r) is decomposed by solar insolation
as follows [Ohara, 2010]:
þO2
NO2 þ hν → NO þ O3 (6)

Equations (1) and (6) are in a relationship of inverse reaction; thus, they are at equilibrium, as expressed by the
following [Ohara, 2010]:

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6512

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

Figure 12. The temporal variation of NO2(r) and NO2(e) concentration calculated by equations (6) and (8), with the amount
of global solar radiation. The blue zone means the rainfall time (first UHR and second UHR) of this day.

NO þ O3 ↔NO2 ðrÞ (7)

Equation (7) indicates that atmospheric O3 runs through the formation and the decomposition. This equili-
brium masks the cause of the variation in O3 concentration. For example, even if the O3 concentration
increased, we could not determine whether the cause was increased O3 formation or decreased O3 consump-
tion by NO. PO is used for the estimation of the cause of the variation in O3 concentration. PO is expressed as
follows [Itano et al., 2007]:
½PO ¼ ½NO2 ðrÞ þ ½O3  (8)

PO does not change with the consumption of O3 by NO because the consumption of O3 produces the same
amount of NO2(r) but changes with the reaction of HO with NO and NO2, as well as advective flow of NO2 and
O3. Using the Ox concentration variation (ΔOx) and PO concentration variation (ΔPO), we could discern a
change in the variation factors of the concentration of O3 and NO2(r):
1. ΔOx > ΔPO: Contribution of O3 consumption by NO is large;
2. |ΔOx
ΔPO| is small: Substantive Ox concentration is increased by the formation of Ox locally and/or
inflow of Ox from the surrounding area [Ohara, 2010].
As just described, PO is used for discussion of the O3 variation factor. However, not only the O3 but also the
NO2(r) contributes to the variation in PO; thus, we use PO as an indicator of the NO2(r) variation.
Figure 13 shows the temporal variation in PO and Ox on 24 July 2014. The blue zones in Figure 13 indicate the
rainfall time for this day (see Figure 8). The PO concentration increased from around 10:00 JST, reached a peak
around 16:00 JST, then decreased gradually. Before formation of the first UHR (14:00–17:00 JST), the concen-
tration of air pollutants was high, and the PO concentration increased. Thus, in this time period, inflow of NO2
or Ox from the surrounding area and/or the formation of NO2 by HO could occur. According to Figure 8, the
NO2(r) concentration increased, and the Ox concentration decreased, although the Ox concentration rose or
fell during this time. This indicates that the inflow of NO2 and/or the formation of NO2 with oxidation of NO by
HO contributed to the increase in PO. During this time, the wind speed was low (average wind speed:
2.6–3.7 m/s, 3.2 m/s average over 10 min; maximum wind speed: 6.1 m/s at 16:50 JST), indicating that the con-
tribution of NO2 inflow was low, but the formation of NO2 with the oxidation of NO by HO was high. These
results suggest that HO had a large contribution to the NO2 oxidative formation before the formation of
UHR (14:00–17:00).

Figure 13. The temporal variation of PO and Ox in 24 July 2014. The blue zone means the rainfall time (first UHR and second
UHR) of this day.

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6513

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

Figure 14. The temporal variation of NO2, both the calculated and observed value. The blue zone means the rainfall time
(first UHR and second UHR) of this day.

3.3.4. Formation of HNO3 Just Before UHR Formation

We discussed the NO2 (precursor of HNO3) formation 4 h to 5 min before the first UHR (the first UHR formed at
17:05 JST, Figure 6). In this section we discuss the behavior of the air pollutants, especially HNO3 (a precursor
of hygroscopic particles, which act as cloud condensation nuclei) 5 min before formation of the first UHR, to
the first UHR, to the end (17:00–18:00 JST). Figure 8 shows that at 17:00–18:00, the NO2 concentration
decreased sharply (52.4% of NO2 at 17:00). According to Figure 13, the PO concentration decreased. The
outflow of NO2 or Ox from Shinjuku to the surrounding area and/or the consumption of NO2 by HO (HNO3
formation) may have occurred. According to Figure 8, the NO2(r) concentration decreased, and the Ox
concentration increased sharply. This indicates that the outflow of NO2 to the surrounding area and/or the
consumption of NO2 by HO contributed to the decrease in PO. The wind speed was high (average wind
speed: 1.9–5.0 m/s, 3.4 m/s average over 10 min; maximum wind speed: 9.0 m/s at 18:00 JST), indicating
the possibility of NO2 outflow from Shinjuku to the surrounding area. On the other hands HO always exist
in the atmosphere in the daytime; thus, the possibility of NO2 elimination by HO (HNO3 formation) could
be also high:
NO2 þ HO→ HNO3 (9)

This indicates that both outflow of NO2 from Shinjuku to the surrounding area and the consumption of NO2
by HO (HNO3 formation) could affect the sharply decreasing NO2 concentration at 17:00–18:00.
In order to estimate the contribution of both processes, we calculate the NO2 concentration, assuming that
the NO, NO2, and Ox are in a steady-state condition (no advective flow and oxidation by HO). The formation of
NO2 is calculated as follows:
½NO2 formation ¼ k 1 ½NO½Ox  (10)

where k1 is the reaction rate of equation (1) and [NO], [NO2], and [Ox] are concentrations of NO, NO2, and Ox,
respectively, in ppmv.
The decomposition of NO2 is calculated as follows:

½NO2 degradation ¼ k 2 ½NO2  (11)

where k2 is the reaction rate of equation (6). The value of k1 and k2 are 3.3 × 103exp(
1430/T) (ppm
1 min
1)
[Baulch et al., 1982] and 0.45 min
1 [Darnall et al., 1981], respectively. Thus, the values calculated from
equations (10) and (11) are the formation and decomposition of NO2 per minute. The formation of NO and
Ox by NO2 decomposition is the same as the decomposition of NO2 in mole number. We calculated the
NO2 concentration at a given time using each observed value of NO2 concentration before 1 h according
to equations (10) and (11) for 60 min. Figure 14 shows the temporal variation of the calculated and observed
values of NO2. The blue zones in Figure 14 indicate the rainfall time on this day (see Figure 8). The upper panel
shows the temporal variation in NO2, while the lower shows the difference between calculated and observed
values of NO2 concentration (the observed value minus the calculated value). At 18:00, the calculated value
was higher than the observed value, and the difference was 65 ppb (see the below chart of Figure 14). This
indicates that 65 ppb of NO2 in the atmosphere were eliminated by other means than photochemical

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6514

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

Figure 15. The temporal variation of ΔOx
ΔPO. The ΔOx and ΔPO is the difference of the concentration in “(X) o’clock”
and “(X
1) o’clock” implied the variation for 1 h. The blue zone in Figure 15 means the rainfall time (first UHR and second
UHR) of this day.

decomposition, briefly, by outflow of NO2 from Shinjuku to the surrounding area and/or NO2 elimination by
HO (HNO3 formation) in the vicinity of the first UHR.
We estimated how much of the 65 ppb of NO2 was eliminated by HO oxidation. The formation of HNO3 is
calculated as follows:
½HNO3  ¼ k 3 ½NO2 ½HO (12)
where k3 is the reaction rate of equation (9) and is 1.8 × 104 ppm
1 min
1 [Sander et al., 2011]. Assuming the
atmospheric HO concentration is 1.63 × 10
7 ppm in steady state [Kanaya et al., 2007], we calculated the
HNO3 concentration (calculated value) at a given time using each observed value of NO2 concentration
before 1 h according to equation (12) for 60 min. Oxidation by HO formed 21.8 ppb of HNO3 from 17:00 to
18:00 in the vicinity of the first UHR, and 15.9 ppb of NO2 was eliminated (24.5% of the NO2 concentration,
which was eliminated by means other than photochemical decomposition). Thus, the remaining 49.1 ppb
of NO2 (75.5% of the NO2 concentration) could have been eliminated by means other than photochemical
decomposition and HO oxidation. 21.8 ppb of formed HNO3 could be scavenged by two processes: (1)
Scavenged by both in- and below-cloud scavenging; (2) and HNO3 reacts with NH3 to form NH4NO3 and is
scavenged by in-cloud process as CCN, because of the upward transportation of NH4NO3 by the strong
upwind before UHR.
As mentioned above, strong upwind with convection just before UHR has been pointed out [Mikami et al.,
2005; Fujibe, 2004]. Figure 8 showed the increase in wind speed from just before formation of the first
UHR, suggesting that not only outflow of NO2 but also upward transportation of NO2 could contribute to
the elimination of the 49.1 ppb of NO2, which could have been eliminated by means other than photoche-
mical decomposition and HO oxidation. Solar insolation is stronger in the upper atmosphere than near the
Earth’s surface [IPCC1995, 1996], so there could exist more HO in the upper atmosphere than near the
Earth’s surface. Thus, the nitric acid and nitrate could be formed by the increased HO.
Figure 15 shows the temporal variation of ΔOx
ΔPO on 24 July 2014. The ΔOx and ΔPO indicate the differ-
ence for 1 h. As mentioned in section 3.3.3, the larger ΔOx than ΔPO indicates that the contribution of O3 con-
sumption by NO is large; i.e., the larger value of ΔOx
ΔPO value means that more NO is consumed by
oxidation to form NO2(r). The value of ΔOx
ΔPO had the highest value at 17:00 to 18:00 in the vicinity of
the first UHR and was high at 18:00 to 19:00 (1 h after the end of the first UHR). The period 17:00 to 19:00
is the time when the NOx concentration decreased sharply. After 19:00, the value of ΔOx
ΔPO was almost
zero, and the NOx concentration was constant (approximately 20 ppb). This suggests that most atmospheric
NO was oxidized and consumed in the vicinity of the UHR event. As mentioned above, NO2, which was
formed by NO consumption, was eliminated in the atmosphere by the outflow, upwind, and
hydroxyl radicals.
For comparison, we show the temporal variation in the meteorological factors and air pollutant concentra-
tions on a normal rain day on 2 July 2015 in Figure 16. The explanatory note is the same as for Figures 8,
12, and 13. It rained around 9:00, and the rainfall intensity was 0.5 mm/10 min. Compared to UHR, the con-
centrations of air pollutants were lower than those shown in Figure 8, and no increase in air pollutants was
observed just before the event. The NO2(r), NO2(e), and PO were monotonous, indicating that the oxidation
of NO and NO2 was not promoted on this day. Strong winds (maximum: 6.4 m/s) occurred at 20:50 JST, but
there was no variation with the rainfall occurrence.

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6515

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

Figure 16. The temporal variation of the meteorological factors (temperature, relative humidity, wind direction, and wind
speed; at Tokyo observatory), precipitation amount every 10 min (Nerima observatory), and the air pollutants concen-
trations (SO2, NOx, NO2, NO, Ox, nonmethane hydrocarbon, methane, total carbon, SPM, and CO; at Shinjuku observatory)
in 2 July 2015. The explanatory note is the same as Figures 8, 12, and 13.

3.3.5. Impact of Air Pollutants Just Before Rainfall Event on Chemical Composition of Rainwater
We have discussed the oxidation and transportation of NO and NO2, which are the precursors of nitric
acid/nitrate in ambient air before, during, and after the UHR event observed on 24 July 2014 in detail. In this
section, we compare the chemical composition of rainwater in UHR and normal rain. Figure 17 shows the che-
mical composition of UHR on 24 July 2014 (a) and normal rain on 2 July 2015 (b). The sample collected on 24
July 2014 includes two UHR events: the first UHR and second UHR (as described above). Both UHR events
could have been formed by similar processes (discussed in sections 3.3.1–3.3.5) and could have had similar
chemical compositions. Figure 17 indicates that the UHR had a much higher NO3
concentration than normal
rain. As described above, the SO2 and NO2 concentrations in ambient air increased just before the UHR event,

Figure 17. (a) The chemical composition of UHR in 24 July 2014. (b) The chemical composition of normal rain in 2 July 2015.

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6516

 Journal of Geophysical Research: Atmospheres
Acknowledgments
The meteorological data with a 10 min
interval (temperature, relative humidity,
air pressure, wind direction, and wind
speed) and that with a 60 min interval
(global solar radiation amount) are
available at the JMA website: http://
www.data.jma.go.jp/obd/stats/etrn/
index.php. Air pollutants concentration
data sets with a 60 min interval are
available upon request from the
Shinjuku City Office: http://www.city.
shinjuku.lg.jp/seikatsu/file10_05_00002.
html. The rainfall amount data with
10 min interval obtained from Shinjuku,
Toshima, Chuo, Nakano, and
Shibuyabashi observatories are avail-
able upon request from the Bureau of
Construction, Tokyo Metropolitan
Government: http://www.kensetsu.
metro.tokyo.jp/. The rainwater chemical
components data used in this study are
available in Table S3. This research was
partly supported by the Nippon Life
Insurance Foundation.
UCHIYAMA ET AL.
10.1002/2017JD026803
while these trends were not observed on the normal rain day. SO2 dissolves in water, but NO2 does not. Thus,
most of NO2 could not be scavenged by rainfall. However, HNO3, which is formed by NO2 oxidation by HO,
dissolves quickly and is easily scavenged by rainfall (NH4NO3 is also) [Okochi, 2014]. The difference between
the NO3
concentration in UHR and normal rain could be the difference in the HNO3 concentration in the
ambient air before each rain event. Thus, the inorganic ion composition in rainwater also reveals the promo-
tion of NO2 oxidation just before UHR.
4. Conclusions
We analyzed inorganic ions in rainwater samples collected on an event basis over 5 years from October 2012
to December 2016 in Shinjuku, Tokyo, and available data of air pollutants in several points of Tokyo, to clarify
the impact of air pollutants on the formation of sudden and locally distributed heavy rain in urban areas
(UHR). The following three points were made clear in this study:
1. The total wet deposition flux, hourly mean rainfall amount, and mean rainfall time per UHR event were
11.0, 13.1, and 0.11 times higher, respectively, than that of normal rain, suggesting that UHR could effec-
tively scavenge air pollutants with a large amount of rainfall in a very short time (within 2 h).
2. The UHR had especially high concentrations of acidic components (H+, NO3
, and nss-SO42
) and a high
N/S ratio, indicating that the concentration of acidic substances in ambient air was very high just before
the UHR event.
3. The concentration of secondarily formed NO2 and PO in ambient air increased just before the formation of
UHR and decreased sharply toward the end of the UHR event. These results indicate that NO2, which was
formed by O3 and/or HO radical oxidation of NO, was further oxidized by HO radicals. The HNO3 that was
formed just before the formation of UHR and NH4NO3, formed by HNO3 reaction with NH3, could subse-
quently be scavenged by both in- and below-cloud scavenging.
To our knowledge, this is the first report which discussed the impact of air pollutants on the formation of UHR,
based on the linkage between rainwater chemistry and air pollutants just before UHR. We estimate that UHR
could occur in several large cities with the strong updrafts by UHI, enough moisture to form cloud, and air
pollutants which could form CCN, even in inland area, as the same mechanism proposed in this study. The
increased frequency of UHR could be likely an environmental and ecological risk because UHR locally pro-
vides large amounts of rainfall and air pollutants such as acidic substances on the ground during a very
short time.
Further field study of rainfall collected at short time intervals during UHR events along with the observation of
air pollutants such as HNO3, ammonium, nitrate, and sulfate aerosol, which are not observed by ambient air
pollution monitoring station operated by national and local governments, is needed. In addition, regional-
scale simulation model analyses of meteorological factors, especially the wind convergence, are needed
for getting advanced solutions for the UHR formation mechanism. We also need to analyze the 3D structure
of the convective clouds, which cause UHR in the linkage of rainwater chemistry and stable isotope of water
and chemicals in UHR collected on the ground.
References
Atkinson, B. W. (1971), The effect of an urban area on the precipitation from a moving thunderstorm, J. Appl. Meteorol., 10, 47–55,
doi:10.1175/1520-0450.
Baulch, D. L., R. A. Cox, P. J. Crutzen, R. F. Hampson, F. A. Kerr, and R. P. Watson (1982), Evaluated kinetic and photochemical data for
atmospheric chemistry: Supplement 1 codata task group on chemical kinetics, J. Phys. Chem. Ref. Data Monogr., 11, 327–496, doi:10.1063/
1.555664.
Bently, M., T. Stallins, and W. Ashley (2010), The Atlanta thunderstorm effect, Weatherwise, 63, 24–29.
Bertoldo, S., C. Lucianaz, M. Allegretti, and G. Perona (2016), Real time monitoring of extreme rainfall events with simple X-band mini weather
radar, Atmos. Clim. Sci., 6, 285–299, doi:10.4236/acs.2016.62024.
Bornstein, R. D. (1968), Observations of the urban heat island effect in New York City, J. Appl. Meteorol., 7, 575–582, doi:10.1175/1520-0450.
Chandler, T. J. (1965), Temperature, in The Climate of London, pp. 126–184, Hutchinson of London, London.
Changnon, S. A., Jr. (1968), The La Porte weather anomaly—Fact or fiction?, B. Am. Meteorol. Soc., 49, 4–11, doi:10.4236/acs.2014.42027.
Darnall, K. R., R. Atkinson, A. M. Winer, and J. N. Pitts Jr. (1981), Effects of constant versus diurnally-varying light intensity on ozone formation,
Japca, J. Air. Waste. Ma., 31, 262–264, doi:10.1080/00022470.1981.10465221.
Dou, J., Y. Wang, R. Bornstein, and S. Miao (2015), Observed spatial characteristics of Beijing urban climate impacts on summer thunder-
storms, J. Appl. Meteorol. Clim., 54, 94–105, doi:10.1175/JAMC-D-13-0355.1.
Fowler, A. M., and K. J. Hennessy (1995), Potential impacts of global warming on the frequency and magnitude of heavy precipitation,
Nat. Hazards, 11, 283–303, doi:10.1007/BF00613411.
HEAVY RAIN FORMATION AND AIR POLLUTANTS 6517
 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

Fujibe, F. (2004), Effect of heat island to precipitation: Convective precipitation in summer, TENKI, 51, 109–115.
Goswami, B. N., V. Venugopal, D. Sengupta, M. S. Madhusoodanan, and P. K. Xavier (2006), Increasing trend of extreme rain events over India
in a warming environment, Science, 314, 1442–1445, doi:10.1126/science.1132027.
Greenfield, S. M. (1957), Rain scavenging of radioactive particulate matter from the atmosphere, J. Meteorol., 14, 115–125, doi:10.1175/
1520-0469.
Huong, H. T. L., and A. Pathirana (2013), Urbanization and climate change impacts on future urban flooding in Can Tho city, Vietnam, Atmos.
Chem. Phys., 17, 379–394, doi:10.5194/hess-17-379-2013.
IPCC1995 (1996), The Climate System: an Overview, in Climate Change 1995, the Science of Climate Change, pp. 51–64, Cambridge Univ. Press.,
Cambridge, U. K.
IPCC2007 (2007), Synthesis Report, in Climate Change 2007, Synthesis Report—An Assessment of the Intergovernment Panel on Climate Change,
pp. 23–73, Cambridge Univ. Press, Cambridge, U. K.
Itano, Y., H. Bandow, N. Takenaka, Y. Saitoh, A. Asayama, and J. Fukuyama (2007), Impact of NOx reduction on long-term ozone trends in an
urban atmosphere, Sci. Total Environ., 379, 46–55, doi:10.1016/j.scitotenv.2007.01.079.
Ito, Y., Q. Moteki, K. Souma, K. Yorozu, K. Tanaka, and S. Ikebuchi (2006), Effects of urban heating on development of the Nerima heavy rainfall
by a cloud resolving model coupled with a precise land surface model, Proc. Hydraul. Eng., 50, 385–390, doi:10.2208/prohe.50.385.
Jauregui, E., and E. Romales (1996), Urban effects on convective precipitation in Mexico City, Atmos. Environ., 30, 3383–3389, doi:10.1016/
1352-2310.
Kanaya, Y., R. Cao, H. Akimoto, M. Fukuda, Y. Komazaki, Y. Yokouchi, M. Koike, H. Tanimoto, Y. Takegawa, and Y. Kondo (2007), Urban
photochemistry in central Tokyo: 1. Observed and modeled OH and HO2 radical concentrations during the winter and summer of 2004,
J. Geophys. Res., 112, D21312, doi:10.1029/2007JD008670.
Kawasaki, A., and K. Meguro (2011), A preliminary investigation on the introduction of housing water tanks to store rainwater for improving
urban environment and disaster risk reduction, SEISANKENKYU, 63, 451–456, doi:10.11188/seisankenkyu.63.451.
Kheim, M., R. Ooka, H. Hong, H. Hayami, H. Yoshikado, and Y. Kawamoto (2010), Analysis of the relationship between changes in meteoro-
logical condition and the variation in summer ozone levels over the central Kanto area, Adv. Meteorol., 2010, 1–13, doi:10.1155/2010/
349248.
Kim, H. H. (1992), Urban heat island, Int. J. Remote. Sens., 13, 2319–2336, doi:10.1080/01431169208904271.
Lacke, M. C., T. L. Mote, and J. M. Shepherd (2009), Aerosols and associated precipitation patterns in Atlanta, Atmos. Environ., 43, 4359–4373,
doi:10.1016/j.atmosenv.2009.04.022.
Lamb, D., and J. Verlinde (2011), Cloud-scale and population effects, in Physics and Chemistry of Clouds, pp. 415–547, Cambridge Univ. Press,
Cambridge, U. K.
Lei, M., D. Niyogi, C. Kishtawal, R. A. Pielke Sr., A. Beltran-Przekurat, T. E. Nobis, and S. S. Vaidya (2008), Effect of explicit urban land surface
representation on the simulation of the 26 July 2005 heavy rain event over Mumbai, India, Atmos. Chem. Phys., 8, 5975–5995, doi:10.5194/
acp-8-5975-2008.
Levein, Z., and W. R. Cotton (2009), Effects of Pollution and Biomass Aerosols on Clouds and Precipitation: Observational Studies, in Aerosol
Pollution Impact on Precipitation, pp. 205–242, Springer, Dordrecht, Netherlands.
Liang, Q., L. Jaeglé, D. A. Jaffe, P. Weiss-Penzias, A. Heckman, and J. A. Snow (2004), Long-range transport of Asian pollution to the northeast
Pacific: Seasonal variations and transport pathways of carbon monoxide, J. Geophys. Res., 109, D23S07, doi:10.1029/2003JD004402.
Miao, S., F. Chen, Q. Li, and S. Fan (2011), Impacts of urban processes and urbanization on summer precipitation: A case study of heavy rainfall
in Beijing on 1 August 2006, J. Appl. Meteorol. Clim., 50, 806–825, doi:10.1175/2010JAMC2513.1.
Mikami, T., H. Yamato, H. Ando, H. Yokoyama, T. Yamaguchi, M. Ichino, Y. Akiyama, and K. Ishii (2005), Climatological study on the summer
intensive heavy rainfall in Tokyo, Annual Report of the Tokyo Metropolitan Research Institute for Environmental Protection, pp. 33–42.
Naemura, A., and Y. Watanabe (2016), Relations with high NO2 concentration at night-time and potential ozone concentration at the Tokyo
Tower, Jpn. J. Biometeorology, 53, 39–44, doi:10.11227/seikisho.53.39.
Ohara, T (2010), Study on characteristics of photochemical oxidants and particulate matter, Research Report from the National Institute for
Environmental Studies, 203, 1–219.
Oke, T. R., and C. East (1971), The urban boundary layer in Montreal, Boundary Layer Meteorol., 1, 411–437, doi:10.1007/BF00184781.
Okochi, H. (2014), Chemistry of clouds, in The Physics and Chemistry of Trans-Boundary Air Pollution [in Japanese], edited by S. Fujita et al.,
pp. 146–167, Seizando-Shoten, Tokyo, 2014.
Ooka, R., M. Khiem, H. Hayami, H. Yoshikado, H. Hong, and Y. Kawamoto (2011), Influence of meteorological condition on summer ozone
levels in the central Kanto area of Japan, Prog. Environ. Sci., 4, 138–150, doi:10.1016/j.proenv.2011.03.017.
Rajeevan, M., J. Bhate, and A. K. Jaswal (2008), Analysis of variability and trends of extreme rainfall events over India using 104 years of
gridded daily rainfall data, Geophys. Res. Lett., 35, L18707, doi:10.1029/2008GL035143.
Robaa, S. M. (2003), Urban-suburban/rural differences over Greater Cairo, Egypt, Atmosfera, 16, 157–171.
Saiki, Y., H. Kashimura, Y. Suyama, M. Nakazawa, and H. Makino (1986), Compositions and photochemical reactivities of non-methane
hydrocarbons in various sources and environments, J. Jpn. Soc. Air Pollut., 21, 123–131, doi:10.11298/taiki1978.21.123.
Sander, S. P., et al. (2011), Chemical kinetics and photochemical data for use in atmospheric studies, Evaluation Number 17, JPL Publ. 10–6.
Sato, N., and M. Takahashi (2000), Long-term changes in the properties of summer precipitation in the Tokyo area, TENKI, 47, 29–34.
Seinfeld, J. H., and S. N. Pandis (2006), Atmospheric Trace Constituents, in Atmospheric Chemistry and Physics, Second ed., pp. 21–74,
Wiley-Interscience, New York.
Schwartz, S. E., and W. G. N. Slinn (1992), Cloud and Aerosol Microphysics, in Precipitation Scavenging and Atmosphere-Surface Exchange,
vol. 1, pp. 5–202, CRC Press, Boca Raton, Fla.
Shiraki, Y., A. Higuchi, and A. Kondo (2009), The effect of an urban environment on the precipitation in areas around Tokyo, Environ. Sci., 22,
187–195, doi:10.11353/sesj.22.187.
Smith, J. A., M. L. Baeck, K. L. Meierdiercks, A. J. Miller, and W. F. Krajewski (2007), Radar rainfall estimation for flash flood forecasting in small
urban watersheds, Adv. Water Resour., 30, 2087–2097, doi:10.1016/j.advwatres.2006.09.007.
Suzuki, K. (1983), A chemical study of snow in Sapporo, Geogr. Rev. Jpn, 56, 171–184.
Toshima city office (2013), Fiscal year 2013 road traffic row and vibration survey result. [Available at http://www.city.toshima.lg.jp/153/
machizukuri/sumai/kogai/kotsu/033120.html, accessed 6 December 2016.]
Uchiyama, R., H. Okochi, H. Ogata, N. Katsumi, D. Asai, and T. Nakano (2017), Geochemical and stable isotope characteristics of urban heavy
rain in the downtown of Tokyo, Japan, Atmos. Res., 194, 109–118, doi:10.1016/j.atmosres.2017.04.029.
Uno, I., H. Ueda, and S. Wakamatsu (1989), Formatian mechanism for high concentrations of NO2 over a nocturnal urban area, J. Jpn. Soc. Air
Pollut., 24, 130–143, doi:10.11298/taiki1978.24.130.

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6518

 Journal of Geophysical Research: Atmospheres 10.1002/2017JD026803

Wang, J., J. Feng, Q. Wu, and Z. Yan (2016), Impact of anthropogenic aerosols on summer precipitation in the Beijing–Tianjin–Hebei urban
agglomeration in China: Regional climate modeling using WRF-Chem, Adv. Atmos. Sci., 33, 753–766, doi:10.1007/s00376-015-5103-x.
Wang, W. C., Z. Zeng, and T. R. Karl (1990), Urban heat islands in China, Geophys. Res. Lett., 17, 2377–2380, doi:10.1029/GL017i013p02377.
Wilson, W. E., and H. H. Suh (1997), Fine particles and coarse particles: Concentration relationships relevant to epidemiologic studies, J. Air
Waste Manage. Assoc., 47, 1238–1249, doi:10.1080/10473289.
Yamato, H., T. Mikami, and H. Takahashi (2011), Influence of sea breeze on the daytime urban heat island in summer in the Tokyo metro-
politan area, J. Geography, 120, 325–340, doi:10.5026/jgeography.120.325.

UCHIYAMA ET AL. HEAVY RAIN FORMATION AND AIR POLLUTANTS 6519