Science of the Total Environment 744 (2020) 140960

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Unexpected air quality impacts from implementation of green

infrastructure in urban environments: A Kansas City case study
Yuqiang Zhang a,h,⁎, Jesse O. Bash b, Shawn J. Roselle b, Angie Shatas c, Andrea Repinsky d, Rohit Mathur b,
Christian Hogrefe b, Jamie Piziali e, Tom Jacobs f, Alice Gilliland g
a
Oak Ridge Institute for Science and Education (ORISE) Fellowship Participant at U.S. Environmental Protection Agency, Research Triangle Park, NC 27711, United States of America
b
Computational Exposure Division, National Exposure Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711,
United States of America
c
Outreach Information Division, Office of Air Quality Planning and Standards, Office of Air and Radiation, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711, United States of America
d
Research Services Department, Mid-America Regional Council, Kansas City, MO 64105, United States of America
e
Water Permits Division, Office of Wastewater Management, Office of Water, U.S. Environmental Protection Agency, Washington, DC 20460, United States of America
f
Transportation and Environment, Mid-America Regional Council, Kansas City, MO 64105, United States of America
g
National Risk Management Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, Cincinnati, OH 45268, United States of America
h
Nicholas School of the Environment, Duke University, Durham, NC 27708, United States of America

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• We investigate the hypothetically city-

level green infrastructure implementa-
tion on regional changes.
• Model showed summertime PM2.5 in-
creases up to 1.1 μg m−3 in Kansas City.
• The summertime peak O3 decreased
over the downtown areas.
• The O3 decreases dominated by the in-
creased NOx titration from PBL changes
• Highlight the region-specific non-linear
process feedback from green infrastruc-
ture on regional air quality

a r t i c l e i n f o a b s t r a c t

Article history: Green infrastructure (GI) implementation can benefit an urban environment by reducing the impacts of urban
Received 31 May 2020 stormwater on aquatic ecosystems and human health. However, few studies have systematically analyzed the
Received in revised form 9 July 2020 biophysical effects on regional meteorology and air quality that are triggered by changes in the urban vegetative
Accepted 12 July 2020
coverage. In this study we use a state-of-the-art high-resolution air quality model to simulate the effects of a hy-
Available online 15 July 2020
pothetically feasible vegetation-focused GI implementation scenario in Kansas City, MO/KS on regional meteorol-
Editor: Jianmin Chen ogy and air quality. Full year simulations are conducted for both the base case and GI land use scenarios using two
different land surface models (LSMs) schemes inside the meteorological model. While the magnitudes of the
Keywords: changes in air quality due to the GI implementation differ using the two LSMs, the model outputs consistently
Green infrastructure showed increases in summertime PM2.5 (1.1 μg m−3, approximately 10% increase using NOAH LSM), which oc-
Urban air quality curred mostly during the night and arose from the primary components, due to the cooler surface temperatures
Temperature and the decreased planetary boundary layer height (PBLH). Both the maximum daily 8-hour average ozone and
Planetary boundary layer 1 h daily maximum O3 during summertime, decreased over the downtown areas (maximum decreases of 0.9 and
Particular matter
1.4 ppbv respectively). The largest ozone decreases were simulated to happen during the night, mainly caused by
Ozone
the titration effect of increased NOx concentration from the lower PBLH. These results highlight the region-

⁎ Corresponding author at: Nicholas School of the Environment, Duke University, 9 Circuit Dr, Durham, NC 27708, United States of America.
E-mail address: Yuqiang.Zhang@duke.edu (Y. Zhang).

https://doi.org/10.1016/j.scitotenv.2020.140960
0048-9697/© 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
2 Y. Zhang et al. / Science of the Total Environment 744 (2020) 140960

specific non-linear process feedback from GI on regional air quality, and further demonstrate the need for com-
prehensive coupled meteorological-air quality modeling systems and necessity of accurate land surface model for
studying these impacts.
© 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction increasing vegetation in Kansas City (KC), MO/KS on regional meteorol-

ogy and air quality. The GI scenario used in this study was provided by
The term ‘green infrastructure’ means the range of measures that use the Mid-America Regional Council (MARC), and included modifications
plant or soil systems, permeable pavement or other permeable surfaces to land cover data consistent with widespread adoption of GI projects,
or substrates, stormwater harvest and reuse, or landscaping to store, in- such as urban reforestation and wetland restoration, as in Fig. 1. The
filtrate, or evapotranspirate stormwater and reduce flows to sewer sys- major land use categories changes between the hypothetical GI and cur-
tems or to surface waters. Green infrastructure (GI) implementation in rent scenario can be seen in Fig. S4.
urban areas can alleviate the impacts of urban stormwater on aquatic
ecosystems and human health, improve the quality of both surface and 2. Materials and methods
ground water (Thomas, 2001; Brattebo and Booth, 2003; Davis et al.,
2003; Ahn et al., 2005; Gilbert and Clausen, 2006; Zachary Bean et al., 2.1. The coupled WRF-CMAQ model configurations
2007; US EPA, 2009; Rowe, 2011), as well as significantly reduce surface
air temperature (Gross, 2012; Santamouris, 2014; Žuvela-Aloise et al., Air quality simulations were performed using the CMAQ version 5.2
2016; Sharma et al., 2016), and therefore reduce energy consumption on a domain covering the contiguous U.S. with 12-km horizontal resolu-
in air conditioning (Akbari et al., 2001; Donovan and Butry, 2009; He tion and inner domain with 4-km horizontal resolution and 35 vertical
et al., 2015; Wang et al., 2018). The GI implementation could also cut layers (see Fig. S1). Gas-phase chemistry was simulated with revision 3
the carbon dioxide levels, including direct effects of absorbing CO2 of the Carbon Bond 2006 mechanism (CB6r3), inorganic aerosol thermo-
though photolysis, and indirect effects of reducing CO2 emission from dynamics were based on ISORROPIA II (Nenes et al., 1998, 1999), an ex-
power plants by reducing the cooling energy demand (Georgescu et al., tendable aqueous phase chemistry option (AQCHEM – KMT, Fahey
2014). Surface modification, such as increasing surface albedos and affor- et al., 2017), primary organic aerosol was modeled as non-volatile, and
estation, could bring benefits of increases in the solar radiation reflected secondary organic aerosol was based on Pye et al. (2017). Boundary con-
by Earth, and decreases in air temperature due to shade, the role of evap- ditions were developed from a CMAQ simulation on a larger domain that
oration and transpiration in reducing sensible heat, and contribute to used boundary conditions based on a hemispheric CMAQ simulation
mitigating global climate change (Akbari et al., 2001; Zhang et al., 2016). (Mathur et al., 2017). U.S. anthropogenic emissions were based primarily
Previous studies have found benefits to air quality from GI on version 4 of the 2011 national emission inventory (last accessed March
implementations through the effects of vegetation cover changes on 3, 2019). Emissions representative of 2011 for biogenic compounds
pollutant dispersion and deposition (Currie and Bass, 2008; Nowak (Rasool et al., 2016), windblown dust (Foroutan et al., 2017) and lightning
et al., 2006, 2013, 2014, 2018; Tallis et al., 2011; Pugh et al., 2012; Wu NO emissions (Allen and Pickering, 2002) are simulated inline, and NH3
et al., 2012; Irga et al., 2015; Jeanjean et al., 2016; Selmi et al., 2016; surface-exchange was simulated with CMAQ using a bidirectional ex-
Abhijith et al., 2017; Jayasooriya et al., 2017; Yli-Pelkonen et al., change parameterization (Bash et al., 2013; Pleim et al., 2013). Meteoro-
2017a, 2017b; Sicard et al., 2018). However, few studies have consid- logical fields for CMAQ modeling were based on coupled WRF (v3.8.1)
ered the dynamic and chemical changes associated with increased model. Physics options for the WRF included the asymmetric convective
urban vegetation cover. Urban vegetation reduces surface tempera- model version 2 planetary boundary layer (Pleim, 2007), the Kain-
tures, which in turn reduces chemical reaction rates, especially for sec- Fritsch convective cumulus parameterization (Kain, 2004) with lightning
ondary air pollutants, such as ozone (O3) and secondary organic assimilation (Heath et al., 2016), and the Morrison two-moment micro-
aerosols (SOA). Cooling temperatures also decrease biogenic hydrocar- physics scheme (Morrison et al., 2008). We used two different land sur-
bon emissions (Fu et al., 2014; Ghirardo et al., 2016), which are also im- face model (LSM), Noah (Chen and Dudhia, 2001; Campbell et al., 2019)
portant precursors for O3 and fine particulate matter (PM2.5). A recent and Pleim-Xiu land-surface model (Pleim and Xiu, 2003). The default
study that utilized a coupled meteorological and chemistry model pre- and most widely used LSM inside the coupled WRF-CMAQ model is
dicted that an increase of surface albedo could decrease surface temper- Pleim-Xiu (PX). Results are focused on the Noah LSM results because
atures and the planetary boundary layer height (PBLH), but this effect the PX LSM constrains the surface fluxes by continuously adjusting soil
resulted in a negligible decrease in both 8-hour average ozone concen- moisture and temperature to minimize model errors in 2 m temperature
trations and 24-hour averaged PM2.5 during a three-day long heatwave and mixing ratio, compared with observed surface analyses with nudging
episode in Montreal, Canada (Touchaei et al., 2016). The reduced PBLH coefficients based on model parameters such as solar radiation, tempera-
from cooler temperatures could also increase air pollutant concentra- ture, leaf area, vegetation coverage, and aerodynamic resistance (Pleim
tions (Long et al., 2018), especially for the primary gaseous pollutants and Gilliam, 2009). After the nudging processes, we found the WRF-PX
(i.e., nitric oxides; NOX) that are important precursors for O3 production, LSM dampened the feedbacks from vegetation cover changes on the
as well as for primary (locally emitted) particulate pollutants regional temperature.
(i.e., primary nitrates, sulfates and ammonium). Meanwhile, tempera-
ture changes could also affect cloud cover and precipitation, which can 2.2. Model evaluation
affect the deposition of air pollutants.
Because of the competing factors and non-linear interactions of these A full year simulation in 2011 was carried out in our study at both
processes, assessment of the potential impacts of increased vegetative 12 km and 4 km resolution. We evaluated the model's performance in
cover and/or surface albedo requires the use of full prognostic meteoro- simulating PM2.5 and O3 during both summer and winter time by com-
logical and photochemical models (Taha, 2008a, 2008b, 2008c). In this paring with observations in JJA (June–July–August) and DJF (December–
study, we use a state-of-the-art coupled meteorological-chemistry January–February) at Air Quality System (AQS) datasets (Supporting
model (Weather Research and Forecasting - Community Multi-scale Air Fig. S1). Here we focused on Kansas City only, since the coupled model
Quality a.k.a., WRF-CMAQ; Wong et al., 2012) with the Noah land surface performance over the U.S. has been recently comprehensively evaluated
model (Noah LSM, Campbell et al., 2019) to simulate the effect of solely (Wong et al., 2012; Appel et al., 2017). In summertime, the mean PM2.5
 Y. Zhang et al. / Science of the Total Environment 744 (2020) 140960 3

Fig. 1. Land use maps in the greater Kansas City area for green infrastructure scenario. The differences for major land use categories can be seen in Fig. S4 in the Supporting material.

concentration in Kansas City is 11.1 μg m−3 from the AQS observation data generated by MARC were then mapped to NLCD land use catego-
sites, while the mean from the coupled WRF-CMAQ model is ries used in the WRF-CMAQ modeling system and in the estimation of
9.3 μg m−3. The coupled WRF-CMAQ model underestimates the summer- biogenic volatile carbon emissions. Table S2 details the MARC land use
time PM2.5 concentration in Kansas City with MB (mean bias) of classification scheme and the mapping to the NLCD land use categories
−1.78 μg m−3, and NMB (normalized mean bias) of −16.1% (Fig. S2a). used by WRF and CMAQ in these model simulations.
During wintertime, the coupled WRF-CMAQ overestimates PM2.5 (MB of A feasible vegetative focused green infrastructure scenario was de-
3.4 μg m−3, and NMB of 27.6%), compared with the AQS sites (Fig. S2d). veloped for Kansas City by applying the following changes to the base
During summertime the mean MDA8 O3 from the observations in Kansas land cover data:
City is 53.5 ppbv, and 60.6 ppbv from the coupled model. The coupled
WRF-CMAQ model overestimated the MDA8 O3 with MB of 7.24 ppbv 1) Change impervious or cultivated cover to forest cover within 150′ of
and NMB of 13.6% (Fig. S2b). During wintertime, however, the coupled streams. This assumes restoration of riparian forest occurs within
WRF-CMAQ model underestimates the MDA8 O3 (MB of −3.10 ppbv buffers of all streams;
and NMB of −10.6%, Fig. S2e). We observed similar patterns for 1 h 2) Change herbaceous cover to forest outside of 50 ft buffer of arterials,
daily maximum ozone with the MDA8 O3, overestimating during sum- within activity centers. MARC mapped activity centers, or areas of
mertime and underestimating during wintertime (Fig. S2c, f). We also concentration of employment and non-residential activity, by a). se-
evaluated WRF's performance by comparing the summertime daily max- lect non-residential building footprints, b). create a density raster of
imum temperature, daily minimum temperature and the daily mean the building footprint centroids, c). create a density raster of the
temperature for the 8 counties in Kansas city (Table S1), downloaded building footprint centroids, weighted by the area of the building
from the Midwestern Regional Climate Center (https://mrcc.illinois.edu/ footprints, and d). combine both rasters into one index of centerness.
CLIMATE/welcome.jsp, accessed Feb 05, 2020). In general, the WRF The original land cover data shows that most commercial areas and
model has a satisfying performance in predicting the summertime tem- other activity centers in the region contain very few trees. This step
perature in the Kansas city, with slightly overestimating compared with assumes widespread planting of trees within activity centers as a re-
observation (Fig. S3). However, notice that the bias in the evaluation placement of urban turf.
does not necessarily lead to bias in estimated changes induced by green 3) Split ‘impervious other’ class into road impervious vs. parking
infrastructure changes. impervious.
4) Change impervious other to shrub-scrub cover outside of 50 ft buffer
2.3. The GI scenario design by MARC of arterials, within activity centers. First, MARC isolated parking
areas from the generalized non-building impervious class. Then,
The Mid-America Regional Council, the association of city and MARC selected shrub-scrub cover to fill parking areas within activity
county governments for the bi-state Kansas City region, created 2.5 m centers as a way to assume that parking lots remain, but that they
land cover data using 2012 SPOT imagery. The land use and land cover are retrofitted with vegetated islands. Shrub-scrub cover is defined
4 Y. Zhang et al. / Science of the Total Environment 744 (2020) 140960

by shrubs and trees covering 10–50% of the ground, which is consis- have larger LAI than that from emergent herbaceous wetlands, especially
tent with this assumption. during the growing seasons.
5) Change freeway and interstate right-of-way, and herbaceous cover Although annual simulations were performed, our discussion fo-
within parks, golf courses, and cemeteries, from herbaceous cover cuses on summertime when conditions are more favorable for ozone
to a new grassland cover class. This step assumes that turf cover in formation. The main results for other seasons can also be seen in the
highway rights-of-way and turf within other public and semi- Supporting materials. In the following sections, we will discuss the me-
public land will be restored to native tallgrass prairie species. teorological and air quality changes in summer induced from the vege-
6) Change a 50 ft buffer of arterials to shrub-scrub cover, assuming tation cover changes from the GI implementation in the KC area.
widespread planting of street trees.
7) Change cultivated areas in floodplain to lowland herbaceous/culti- 3.1. GI implementation impacts on meteorology
vated cover. This step assumes that some of the areas that had
been converted from wetlands to agricultural use would be restored Changes in vegetation fraction affects the energy balance between
as wetlands. latent heat flux and sensible heat flux which in turn affects surface tem-
peratures. Larger changes are seen during the summer season (June–
The same mapping was applied to the GI scenario as the base sce- July-August; JJA) than in other seasons for both the 2-m surface temper-
nario (Table S2). This updated land use data then replaced the NLCD ature (T2) and PBLH (Figs. S8–S9). Fig. 3 shows that T2 decreases
data distributed with WRF version 3.8.1 and in the Biogenic Emissions throughout the KC area as a result of increases in the latent heat flux
Land use Database (BELD) data used to estimate biogenic volatile or- and decreases of sensible heat flux (Fig. S10): here, the increased vege-
ganic compound emissions (Bash et al., 2016). Updated land use data tative fraction repartitions the energy from sensible heat flux to the la-
was written to the WRF input file and BELD and then propagated into tent heat flux, which cools down the surface. The change of surface
the WRF-CMAQ modeling system. albedo could also contribute to the surface cooling which was not
shown here. The 2-m surface temperature is reduced over urban areas
3. Results and discussion by more than 0.3 °C, especially over the KC downtown areas, with a
maximum decrease of 0.7 °C. For the diurnal T2 change, the largest T2
In the GI scenario provided by MARC for KC, there are increases in the decreases occur during the daytime (Fig. S11a, b), because of the strong
vegetated natural areas, and decreases in agricultural and urban areas. evapotranspiration effects from the increased vegetation. There are also
Developed areas, including open space, and low, medium and high inten- increases in T2 over the northern and southern portion of the domain,
sity classifications in the KC downtown areas, have been transformed into resulting from repatriating the energy from latent heat flux to the sen-
forest, herbaceous and shrubland areas (Fig. S4, Table S2). Under the GI sible heat flux (Fig. S10), which warm the surface due to LAI changes
scenario, there is also a switch from cultivated crops to wetlands (includ- as a result of switches from cultivated crops to wetlands (Fig. S4). The
ing woody wetlands and emergent herbaceous wetlands; Figs. S4e, f, S5 JJA PBLH differences have similar patterns to T2, with the most signifi-
and Table S2) over the river basins. The leaf area index (LAI) and vegeta- cant decreases (90 m) over the downtown KC, and increases in PBLH
tion fraction are projected to increase across the city (Fig. 2), especially (largest increase of 81 m) over the northern and southern KC area.
during the summertime (Figs. S6–S7), except on the northeast of the do- The PBLH changes are also larger during the daytime than that in the
main over the Missouri River basins where the cultivated crops usually nighttime, following the T2 diurnal change patterns (Fig. S11c, d).

Fig. 2. The summertime (June–July–August, JJA) changes for (a) leaf area index (LAI), (b) vegetation fraction, and (c) impervious fraction (unit %) between simulation with and without GI
scenario in Kansas City. The grey areas are the regions outside Kansas City.
 Y. Zhang et al. / Science of the Total Environment 744 (2020) 140960 5

Fig. 3. JJA mean of meteorological changes in (a) 2-m surface temperature (T2, unit of °C), and (b) planetary boundary layer height (PBLH, unit of m) after the feasible GI implementation
using the WRF-Noah configuration. Also see Figs. S8 and S9 in the Supporting for the T2 and PBL changes using WRF-PX LSM scheme.

3.2. GI implementation impacts on surface air quality the PM2.5 increases occur at night when the PBL shrinks. The increased
vegetative cover in the GI scenario causes the nighttime PBLH to be
For both the base and GI simulations, the configurations are the lower than the base, increasing the pollutant concentrations, as this de-
same, except for the land use changes. The air quality changes from creased volume restricts the ventilation of pollutants. PM2.5 concentra-
these two simulations are due to the land use induced chemical and tions also decrease during the daytime (Fig. S12), especially at the
physical processes. Fig. 4 shows the changes in the 24-hour average, northern and southern portions of the domain (Fig. 4b), likely caused
daytime (from 9 am to 7 pm local time in KC), and nighttime PM2.5 dur- by the slower chemical reaction rates from the relatively lower temper-
ing JJA. We can see that the three-month averaged PM2.5 concentrations atures in the GI scenario. The PBLH also increases over these same areas
increased throughout most of the domain, especially over the down- during the daytime, decreasing PM2.5 concentrations. These increases
town areas, with a maximum increase of 1.1 μg m−3 (10%). Most of are mainly caused by changes in the primary components of PM2.5,

µg m-3

Fig. 4. The changes in JJA of (a) 24-h average PM2.5, (b) daytime average (Kansas City local time, 9 am-5 pm) PM2.5, (c) nighttime average PM2.5, between the base and the GI scenario using
the WRF-Noah LSM. Unit is μg m−3.
6 Y. Zhang et al. / Science of the Total Environment 744 (2020) 140960

such as elementary carbon (EC, Fig. 5e), unspeciated components Instead the decreasing PBLH increased the NOx concentration which de-
(UNSPEC1, Fig. 5f), and primary organic carbon (POC, Fig. S13), which creases the ozone concentration by increasing the NOx titration effect.
are more sensitive to the lower PBL changes. From the mean JJA diurnal O3 changes over the KC region, the surface
The summertime maximum daily 8-hour average ozone (MDA8 O3), O3 significantly decreases in the night and early morning (Fig. S17). The
as well as 1-hour daily maximum O3 (1hrMax O3), are projected to de- domain average of JJA O3 changes in KC are 0.3 ppbv.
crease over the downtown areas (the maximum decreases are 0.9 and
1.4 ppbv for MDA8 O3 and 1hrMax O3, respectively), but increase over 3.3. Sensitivity analysis using different land surface model
the northern and southern portions of the domain (Fig. 6b, c), following
similar patterns as the T2 changes (Fig. 2a). The 24-hour average O3 has For this study, we also perform another set of sensitivity simulations
larger decreases than the MDA8 O3 and 1hrMax O3 (Fig. 6a). When sepa- using the Pleim-Xiu land surface model (PX LSM, see methods) inside
rating the surface ozone changes into daytime and nighttime, we find that the WRF-CMAQ. The PX LSM is widely used in retrospective air quality
the JJA averaged ozone has larger decreases during the night (reaching 5.2 studies (Hogrefe et al., 2015; Ran et al., 2016), and is being used here to
ppbv over the downtown regions, Fig. S14). This is likely caused by the in- test the robustness of our study results. One note concerning the PX
creased NOx titration effects on O3 during the night from the lowered LSM is that it is constrained by surface moisture and temperature with ob-
PBLH (Fig. S15). The decreased PBLH increases NOx concentrations by servations, which could dampen the signals from the land use changes.
up to 12ppbv during the night over the downtown regions (Fig. S14b). From Fig. 7, we see that the T2 and PBLH changes using the PX LSM
Maximum O3 decreases over the downtown are influenced by increased have similar patterns as those using the Noah LSM (Fig. 3), except that
O3 dry deposition from increased vegetation fraction (Fig. S16) and possi- the magnitudes of the changes are relatively smaller. The maximum de-
bly slower chemical reaction rates from reduced air temperatures crease for the JJA T2 is 0.3 °C (compared with 0.7 °C for the Noah LSM),
(Fig. 3a). Unlike PM2.5, the decreasing PBLH over the downtown area and 61 m for JJA PBLH (compared to 90 m for the Noah LSM). The mean
does not dominate surface ozone changes over the urban areas in KC. JJA air quality changes are also smaller compared to the simulation that

Fig. 5. Changes between Base and GI scenarios in major PM2.5 components for JJA: (a) sulfate (SO4), (b) nitrate (NO3), (c) ammonium (NH4), (d) organic carbon (OC), (e) elementary
carbon (EC), (f) unspecialized components (UNSPEC1), and (g) soil.
 Y. Zhang et al. / Science of the Total Environment 744 (2020) 140960 7

Fig. 6. The changes in JJA mean of (a) 24-h average O3, (b) MDA8 O3, and (c) 1 h daily maximum O3, after the feasible GI implementation using the WRF-Noah LSM. Unit is ppbv.

used the Noah LSM, with maximum increases of 0.6 μg m−3 for surface shading and vegetation transpiration effects from an aggressive GI im-
PM2.5, and maximum decrease of 1.4 ppbv for 24-h O3. plementation scenario may have confounding ecologic and economic
This study uses a state-of-the-art regional air quality model to investi- benefits from such as a strategy. Additional research is needed to under-
gate the effects of a possible green infrastructure (GI) scenario on regional stand these effects on anthropogenic and biogenic emissions to fully
meteorology and air quality in Kansas City (KC). Vegetative fractions and capture air quality impacts, which would be especially important for
leaf area indexes increase in the GI scenario, especially over the KC down- areas exceeding the PM2.5 National Ambient Air Quality Standard.
town area, as developed areas are replaced with more vegetated natural Other meteorological effects, such as changes in wind direction, wind
areas. The impervious surfaces also decrease in accordance with the speed, the relative humidity, surface moisture, terrestrial erosion, and
land use type transition. The regional temperature and planetary bound- surface roughness, which could also directly or indirectly influence the
ary height (PBLH) over this area are projected to decrease as the evapo- air quality. Further follow-up research should consider to develop algo-
transpiration from increased vegetation reduces sensible heat fluxes. rithms to quantify the relative contribution of these effects.
The surface ozone concentrations are projected to decrease as much as
2.0 ppbv during the daytime, and 5.2 ppbv during the nighttime. The JJA CRediT authorship contribution statement
daytime O3 decreases reflect the competing effects from increased
ozone dry deposition and slower chemical reaction rates, while the Yuqiang Zhang: Conceptualization, Methodology, Formal analysis,
much larger ozone decrease noted during the nighttime is due to NOx Investigation, Data curation, Writing - original draft, Writing - review
titration effects attributable to the lower PBLH. The increased vegetation & editing, Visualization. Jesse O. Bash: Supervision, Methodology, Data
cover also affects surface PM2.5; summertime PM2.5 concentrations over curation, Resources, Writing - review & editing. Shawn J. Roselle:
the KC downtown area are projected to increase as high as 10% (about Supervision, Methodology, Resources. Angie Shatas: Resources.
1.1 μg m−3), especially during the night. The lower PBLH is responsible Andrea Repinsky: Resources. Rohit Mathur: Conceptualization,
for the PM2.5 concentration increases, especially for the primary PM2.5 Methodology, Formal analysis, Investigation, Writing - review &
components (i.e., elementary carbon, primary organic carbon and editing. Christian Hogrefe: Conceptualization, Methodology, Formal
unspeciated portion of PM2.5 emissions). analysis, Investigation, Writing - review & editing. Jamie Piziali:
When referring to the conclusions, we should keep in mind that Resources. Tom Jacobs: Data curation. Alice Gilliland: Data curation.
these results are for a specific GI scenario which was developed as
part of a storm water management plan for the urban area of KC, and Declaration of competing interest
not as a plan for urban heat island mitigation. The results may vary de-
pending on the GI scenario introduced, as well as the topography of the The authors declare that they have no known competing financial
region. In particular, electricity generating units emissions change to ac- interests or personal relationships that could have appeared to influ-
count for changes in electricity demand due to changes in air tempera- ence the work reported in this paper.
tures. Biogenic emissions may increase due to additional biomass in the
region but may decrease regionally due to lower temperatures, which Acknowledgment
are not quantified in this study. The results from this study can inform
other similar studies, such as examining the impact of urban heat island This research was supported in part by an appointment to the Re-
mitigation strategies (e.g., green roofs/white roofs). This study indicates search Participation Program at the U.S. Environmental Protection
air quality changes in O3 and PM2.5 associated with the additional Agency (EPA), Office of Research and Development, administered
8 Y. Zhang et al. / Science of the Total Environment 744 (2020) 140960

Fig. 7. The changes of JJA average (a) 24-h PM2.5 (μg m−3), (b) 24-h O3 (ppbv), (c) T2 (°C), and (d) PBLH (m) after the feasible GI implementation using the WRF-PX LSM.

by the Oak Ridge Institute for Science and Education (ORISE) through References
an interagency agreement between the U.S. Department of Energy
Abhijith, K.V., Kumar, P., Gallagher, J., McNabola, A., Baldauf, R., Pilla, F., Broderick, B., Di
and the U.S. EPA. We would like to thank Kirk Baker and Val Garcia Sabatino, S., Pulvirenti, B., 2017. Air pollution abatement performances of green infra-
from U.S. EPA for their insightful suggestions for improving the structure in open road and built-up street canyon environments – a review. Atmos.
initial version of this manuscript. The views expressed in this article Environ. 162, 71–86. https://doi.org/10.1016/j.atmosenv.2017.05.014.
are those of the authors and do not necessarily reflect the views or Ahn, J.H., Grant, S.B., Surbeck, C.Q., Digiacomo, P.M., Nezlin, N.P., Jiang, S., 2005. Coastal
water quality impact of stormwater runoff from an urban watershed in Southern Cal-
policies of the U.S. EPA. ifornia. Environ. Sci. Technol. 39, 5940–5953. https://doi.org/10.1021/es0501464.
Akbari, H., Pomerantz, M., Taha, H., 2001. Cool surfaces and shade trees to reduce energy
Appendix A. Supplementary data use and improve air quality in urban areas. Sol. Energy 70, 295–310. https://doi.org/
10.1016/S0038-092X(00)00089-X.
Allen, D.J., Pickering, K.E., 2002. Evaluation of lightning flash rate parameterizations for
Supplementary data to this article can be found online at https://doi. use in a global chemical transport model. J. Geophys. Res. Atmos. 107, 1–21.
org/10.1016/j.scitotenv.2020.140960. https://doi.org/10.1029/2002JD002066.
 Y. Zhang et al. / Science of the Total Environment 744 (2020) 140960 9

Appel, K.W., Napelenok, S.L., Foley, K.M., Pye, H.O.T., Foley, K.M., Spero, T.L., Appel, K.W., Pleim, Morrison, H., Thompson, G., Tatarskii, V., 2008. Impact of cloud microphysics on the de-
J.E., Gilliam, R.C., Fahey, K.M., Kang, D., Luecken, D.J., Pouliot, G.A., Wong, D.C., Roselle, S.J., velopment of trailing stratiform precipitation in a simulated squall line: comparison
Foroutan, H., Sarwar, G., Napelenok, S.L., Young, J.O., Bash, J.O., Hutzell, W.T., Hogrefe, C., of one- and two-moment schemes. Mon. Weather Rev. 137, 991–1007. https://doi.
Heath, N.K., Mathur, R., Schwede, D.B., Gantt, B., 2017. Description and evaluation of the org/10.1175/2008mwr2556.1.
Community Multiscale Air Quality (CMAQ) modeling system version 5.1. Geosci. Model Nenes, A., Pandis, S.N., Pilinis, C., 1999. Continued development and testing of a new ther-
Dev. 10, 1703–1732. https://doi.org/10.5194/gmd-10-1703-2017. modynamic aerosol module for urban and regional air quality models. Atmos. Envi-
Bash, J.O., Cooter, E.J., Dennis, R.L., Walker, J.T., Pleim, J.E., 2013. Evaluation of a regional ron. 33, 1553–1560. https://doi.org/10.1016/S1352-2310(98)00352-5.
air-quality model with bidirectional NH3 exchange coupled to an agroecosystem Nowak, D.J., Crane, D.E., Stevens, J.C., 2006. Air pollution removal by urban trees and
model. Biogeosciences 10, 1635–1645. https://doi.org/10.5194/bg-10-1635-2013. shrubs in the United States. Urban For. Urban Green. 4, 115–123. https://doi.org/
Bash, J.O., Baker, K.R., Beaver, M.R., 2016. Evaluation of improved land use and canopy 10.1016/j.ufug.2006.01.007.
representation in BEIS v3.61 with biogenic VOC measurements in California. Geosci. Nowak, D.J., Hirabayashi, S., Bodine, A., Hoehn, R., 2013. Modeled PM2.5 removal by trees
Model Dev. 9, 2191–2207. https://doi.org/10.5194/gmd-9-2191-2016. in ten U.S. cities and associated health effects. Environ. Pollut. 178, 395–402. https://
Brattebo, B.O., Booth, D.B., 2003. Long-term stormwater quantity and quality performance doi.org/10.1016/j.envpol.2013.03.050.
of permeable pavement systems. Water Res. 37, 4369–4376. https://doi.org/10.1016/ Nowak, D.J., Hirabayashi, S., Bodine, A., Greenfield, E., 2014. Tree and forest effects on air
S0043-1354(03)00410-X. quality and human health in the United States. Environ. Pollut. 193, 119–129. https://
Campbell, P.C., Bash, J.O., Spero, T.L., 2019. Updates to the Noah Land Surface Model in doi.org/10.1016/j.envpol.2014.05.028.
WRF-CMAQ to improve simulated meteorology, air quality, and deposition. J. Adv. Nowak, D.J., Hirabayashi, S., Doyle, M., McGovern, M., Pasher, J., 2018. Air pollution re-
Model. Earth Syst. https://doi.org/10.1029/2018MS001422. moval by urban forests in Canada and its effect on air quality and human health.
Chen, F., Dudhia, J., 2001. Coupling an advanced land surface–hydrology model with the Urban For. Urban Green. 29, 40–48. https://doi.org/10.1016/j.ufug.2017.10.019.
Penn State–NCAR MM5 modeling system. Part I: model implementation and sensitiv- Pleim, J.E., 2007. A combined local and nonlocal closure model for the atmospheric
ity. Mon. Weather Rev. 129, 569–585. https://doi.org/10.1175/1520-0493(2001) boundary layer. Part I: model description and testing. J. Appl. Meteorol. Climatol.
129b0569:CAALSHN2.0.CO;2. 46, 1383–1395. https://doi.org/10.1175/jam2539.1.
Currie, B.A., Bass, B., 2008. Estimates of air pollution mitigation with green plants and Pleim, J.E., Gilliam, R., 2009. An indirect data assimilation scheme for deep soil tempera-
green roofs using the UFORE model. Urban Ecosyst. 11, 409–422. https://doi.org/ ture in the Pleim-Xiu land surface model. J. Appl. Meteorol. Climatol. 48,
10.1007/s11252-008-0054-y. 1362–1376. https://doi.org/10.1175/2009JAMC2053.1.
Davis, A.P., Shokouhian, M., Sharma, H., Minami, C., Winogradoff, D., 2003. Water quality Pleim, J.E., Xiu, A., 2003. Development of a land surface model. Part II: data assimilation.
improvement through bioretention: lead, copper, and zinc removal. Water Environ. J. Appl. Meteorol. 42, 1811–1822. https://doi.org/10.1175/1520-0450(2003)
Res. 75, 73–82. https://doi.org/10.2175/106143003X140854. 042b1811:doalsmN2.0.co;2.
Donovan, G.H., Butry, D.T., 2009. The value of shade: estimating the effect of urban trees Pleim, J.E., Bash, E.O., Walker, J.T., Cooter, E.J., 2013. Development and evaluation of an
on summertime electricity use. Energy Build 41, 662–668. https://doi.org/10.1016/j. ammonia bidirectional flux parameterization for air quality models. J. Geophys. Res.
enbuild.2009.01.002. Atmos. 118, 3794–3806. https://doi.org/10.1002/jgrd.502622013.
Fahey, K.M., Carlton, A.G., Pye, H.O.T., Baek, J., Hutzell, W.T., Stanier, C.O., Baker, K.R., Wyat Pugh, T.A.M., MacKenzie, A.R., Whyatt, J.D., Hewitt, C.N., 2012. Effectiveness of green infra-
Appel, K., Jaoui, M., Offenberg, J.H., 2017. A framework for expanding aqueous chem- structure for improvement of air quality in urban street canyons. Environ. Sci.
istry in the Community Multiscale Air Quality (CMAQ) model version 5.1. Geosci. Technol. 46, 7692–7699. https://doi.org/10.1021/es300826w.
Model Dev. 10, 1587–1605. https://doi.org/10.5194/gmd-10-1587-2017. Pye, H.O.T., Murphy, B.N., Xu, L., Ng, N.L., Carlton, A.G., Guo, H., Weber, R., Vasilakos, P.,
Wyat Appel, K., Hapsari Budisulistiorini, S., Surratt, J.D., Nenes, A., Hu, W., Jimenez,
Foroutan, H., Young, J.O., Napelenok, S.L., Ran, L., Appel, K.W., Gilliam, R.C., Pleim, J.E.,
J.L., Isaacman-Vanwertz, G., Misztal, P.K., Goldstein, A.H., 2017. On the implications
2017. Development and evaluation of a physics-based windblown dust emission
of aerosol liquid water and phase separation for organic aerosol mass. Atmos.
scheme implemented in the CMAQ modeling system. J. Adv. Model. Earth Syst. 9,
Chem. Phys. 17, 343–369. https://doi.org/10.5194/acp-17-343-2017.
1248–1269. https://doi.org/10.1002/2016MS000823.
Ran, L., Pleim, J., Gilliam, R., Binkowski, F.S., Hogrefe, C., Band, L., 2016. Improved meteorology
Fu, Y., Liao, H., Change, C., 2014. Impacts of land use and land cover changes on biogenic
from an updated WRF/CMAQ modeling system with MODIS vegetation and albedo.
emissions of volatile organic compounds in China from the late 1980s to the mid-
J. Geophys. Res. Atmos. Res. 121, 2393–2415. https://doi.org/10.1002/2015JD024406.
2000s: implications for tropospheric ozone and secondary organic aerosol. Tellus B
Rasool, Q.Z., Zhang, R., Lash, B., Cohan, D.S., Cooter, E.J., Bash, J.O., Lamsal, L.N., 2016. En-
66, 1–17. https://doi.org/10.3402/tellusb.v66.24987.
hanced representation of soil NO emissions in the Community Multiscale Air Quality
Georgescu, M., Morefield, P.E., Bierwagen, B.G., Weaver, C.P., 2014. Urban adaptation can
(CMAQ) model version 5.0.2. Geosci. Model Dev. 9, 3177–3197. https://doi.org/
roll back warming of emerging megapolitan regions. Proc. Natl. Acad. Sci. 111,
10.5194/gmd-9-3177-2016.
2909–2914. https://doi.org/10.1073/pnas.1322280111.
Rowe, D.B., 2011. Green roofs as a means of pollution abatement. Environ. Pollut. 159,
Ghirardo, A., Xie, J., Zheng, X., Wang, Y., Grote, R., Block, K., Wildt, J., Mentel, T., Kiendler-
2100–2110. https://doi.org/10.1016/j.envpol.2010.10.029.
Scharr, A., Hallquist, M., Butterbach-Bahl, K., Schnitzler, J.P., 2016. Urban stress-
Santamouris, M., 2014. Cooling the cities - a review of reflective and green roof mitigation
induced biogenic VOC emissions and SOA-forming potentials in Beijing. Atmos.
technologies to fight heat island and improve comfort in urban environments. Sol.
Chem. Phys. 16, 2901–2920. https://doi.org/10.5194/acp-16-2901-2016.
Energy 103, 682–703. https://doi.org/10.1016/j.solener.2012.07.003.
Gilbert, J.K., Clausen, J.C., 2006. Stormwater runoff quality and quantity from asphalt,
Selmi, W., Weber, C., Rivière, E., Blond, N., Mehdi, L., Nowak, D., 2016. Air pollution re-
paver, and crushed stone driveways in Connecticut. Water Res. 40, 826–832.
moval by trees in public green spaces in Strasbourg city, France. Urban For. Urban
https://doi.org/10.1016/j.watres.2005.12.006.
Green. 17, 192–201. https://doi.org/10.1016/j.ufug.2016.04.010.
Gross, G., 2012. Effects of different vegetation on temperature in an urban building envi- Sharma, A., Conry, H.J.S., Hamket, A.F., Hellmann, J.J., Chen, F., 2016. Green and cool roofs
ronment. Micro-scale numerical experiments. Meteorol. Zeitschrift 21, 399–412. to mitigate urban heat island effect in the Chicago metropolitan evaluation with a re-
https://doi.org/10.1127/0941-2948/2012/0363. gional climate model. Environ. Res. Lett. 11, 1–15. https://doi.org/10.1088/1748-
He, Y., Yu, H., Zhao, M., 2015. Thermal performance study of extensive green roof in 9326/11/6/064004.
Shanghai District: a case study of lightweight building in winter. Procedia Eng 121, Sicard, P., Agathokleous, E., Araminiene, V., Carrari, E., Hoshika, Y., De Marco, A., Paoletti,
1597–1604. https://doi.org/10.1016/j.proeng.2015.09.186. E., 2018. Should we see urban trees as effective solutions to reduce increasing
Heath, N.K., Pleim, J.E., Gilliam, R.C., Kang, D., 2016. A simple lightning assimilation tech- ozone levels in cities? Environ. Pollut. 243, 163–176. https://doi.org/10.1016/j.
nique for improving retrospective WRF simulations. J. Adv. Model. Earth Syst. 8, envpol.2018.08.049.
1248–1269. https://doi.org/10.1002/2016MS000689. Taha, H., 2008a. Episodic performance and sensitivity of the urbanized MM5 (uMM5) to
Hogrefe, C., Pouliot, G., Wong, D., Torian, A., Roselle, S., Pleim, J., Mathur, R., 2015. Annual perturbations in surface properties in Houston Texas. Boundary-Layer Meteorol 127,
application and evaluation of the online coupled WRF-CMAQ system over North 193–218. https://doi.org/10.1007/s10546-007-9258-6.
America under AQMEII phase 2. Atmos. Environ. 115, 683–694. https://doi.org/ Taha, H., 2008b. Meso-urban meteorological and photochemical modeling of heat island mit-
10.1016/j.atmosenv.2014.12.034. igation. Atmos. Environ. 42, 8795–8809. https://doi.org/10.1016/j.atmosenv.2008.06.036.
Irga, P.J., Burchett, M.D., Torpy, F.R., 2015. Does urban forestry have a quantitative effect Taha, H., 2008c. Urban surface modification as a potential ozone air-quality improvement
on ambient air quality in an urban environment? Atmos. Environ. 120, 173–181. strategy in California: a mesoscale modelling study. Boundary-Layer Meteorol 127,
https://doi.org/10.1016/j.atmosenv.2015.08.050. 219–239. https://doi.org/10.1007/s10546-007-9259-5.
Jayasooriya, V.M., Ng, A.W.M., Muthukumaran, S., Perera, B.J.C., 2017. Green infrastructure Tallis, M., Taylor, G., Sinnett, D., Freer-Smith, P., 2011. Estimating the removal of atmo-
practices for improvement of urban air quality. Urban For. Urban Green. 21, 34–47. spheric particulate pollution by the urban tree canopy of London, under current
https://doi.org/10.1016/j.ufug.2016.11.007. and future environments. Landsc. Urban Plan. 103, 129–138. https://doi.org/
Jeanjean, A.P.R., Monks, P.S., Leigh, R.J., 2016. Modelling the effectiveness of urban trees 10.1016/j.landurbplan.2011.07.003.
and grass on PM2.5reduction via dispersion and deposition at a city scale. Atmos. En- Thomas, M.A., 2001. The effect of residential development on ground-water quality near
viron. 147, 1–10. https://doi.org/10.1016/j.atmosenv.2016.09.033. Detroit, Michigan. J. Am. Water Resour. Assoc. 36, 1023–1038.
Kain, J.S., 2004. The Kain–Fritsch convective parameterization: an update. J. Appl. Meteorol. Touchaei, A.G., Akbari, H., Tessum, C.W., 2016. Effect of increasing urban albedo on meteorol-
43, 170–181. https://doi.org/10.1175/1520-0450(2004)043b0170:TKCPAUN2.0.CO;2. ogy and air quality of Montreal (Canada) - episodic simulation of heat wave in 2005.
Long, X., Wu, J., Li, X., Feng, T., Xing, L., Zhao, S., Cao, J., Tie, X., An, Z., Li, G., Bei, N., 2018. Atmos. Environ. 132, 188–206. https://doi.org/10.1016/j.atmosenv.2016.02.033.
Does afforestation deteriorate haze pollution in Beijing-Tianjin-Hebei (BTH), China? US EPA, 2009. Green Roofs for Stormwater Runoff Control (PDF). Publication No. EPA/
Atmos. Chem. Phys. 18, 10869–10879. https://doi.org/10.5194/acp-18-10869-2018. 600/R-09/026|Abstract. U. S. Environmental Protection Agency, Washington, DC.
Mathur, R., Xing, J., Gilliam, R., Sarwar, G., Hogrefe, C., Pleim, J., Pouliot, G., Roselle, S., Wang, C., Wang, Z.H., Yang, J., 2018. Cooling effect of urban trees on the built environment
Spero, T.L., Wong, D.C., Young, J., 2017. Extending the Community Multiscale Air of contiguous United States. Earth’s Futur 6, 1066–1081. https://doi.org/10.1029/
Quality (CMAQ) modeling system to hemispheric scales: overview of process consid- 2018EF000891.
erations and initial applications. Atmos. Chem. Phys. 17, 12449–12474. https://doi. Wong, D.C., Pleim, J., Mathur, R., Binkowski, F., Otte, T., Gilliam, R., Pouliot, G., Xiu, A.,
org/10.5194/acp-17-12449-2017. Young, J.O., Kang, D., 2012. WRF-CMAQ two-way coupled system with aerosol
10 Y. Zhang et al. / Science of the Total Environment 744 (2020) 140960

feedback: software development and preliminary results. Geosci. Model Dev. 5, Zachary Bean, E., Frederick Hunt, W., Alan Bidelspach, D., 2007. Evaluation of four perme-
299–312. https://doi.org/10.5194/gmd-5-299-2012. able pavement sites in eastern North Carolina for runoff reduction and water quality
Wu, S., Mickley, L.J., Kaplan, J.O., Jacob, D.J., 2012. Impacts of changes in land use and land impacts. J. Irrig. Drain. Eng. 133, 583–592. https://doi.org/10.1061/(ASCE)0733-9437
cover on atmospheric chemistry and air quality over the 21st century. Atmos. Chem. (2007)133:6(583).
Phys. 12, 1597–1609. https://doi.org/10.5194/acp-12-1597-2012. Zhang, J., Zhang, K., Liu, J., Ban-Weiss, G., 2016. Revisiting the climate impacts of cool roofs
Yli-Pelkonen, V., Scott, A.A., Viippola, V., Setälä, H., 2017a. Trees in urban parks and forests around the globe using an Earth system model. Environ. Res. Lett. 11, 1–11. https://
reduce O3, but not NO2 concentrations in Baltimore, MD, USA. Atmos. Environ. 167, doi.org/10.1088/1748-9326/11/8/084014.
73–80. https://doi.org/10.1016/j.atmosenv.2017.08.020. Žuvela-Aloise, M., Koch, R., Buchholz, S., Früh, B., 2016. Modelling the potential of green
Yli-Pelkonen, V., Setälä, H., Viippola, V., 2017b. Urban forests near roads do not reduce and blue infrastructure to reduce urban heat load in the city of Vienna. Clim.
gaseous air pollutant concentrations but have an impact on particles levels. Landsc. Chang. 135, 425–438. https://doi.org/10.1007/s10584-016-1596-2.
Urban Plan. 158, 39–47. https://doi.org/10.1016/j.landurbplan.2016.09.014.