The WRF-Chem model coupled with a single-layer urban canopy model (UCM) is
integrated for 5 years at convection-permitting scale to investigate the
individual and combined impacts of urbanization-induced changes in land cover
and pollutant emissions on regional climate in the Yangtze River Delta (YRD)
region in eastern China. Simulations with the urbanization effects reasonably
reproduced the observed features of temperature and precipitation in the YRD
region. Urbanization over the YRD induces an urban heat island (UHI) effect,
which increases the surface temperature by 0.53
Results also show that the UHI increases the frequency of extreme summer precipitation by strengthening the convergence and updrafts over urbanized areas in the afternoon, which favor the development of deep convection. In contrast, the radiative forcing of aerosols results in a surface cooling and upper-atmospheric heating, which enhances atmospheric stability and suppresses convection. The combined effects of the UHI and aerosols on precipitation depend on synoptic conditions. Two rainfall events under two typical but different synoptic weather patterns are further analyzed. It is shown that the impact of urban land cover and aerosols on precipitation is not only determined by their influence on local convergence but also modulated by large-scale weather systems. For the case with a strong synoptic forcing associated with stronger winds and larger spatial convergence, the UHI and aerosol effects are relatively weak. When the synoptic forcing is weak, however, the UHI and aerosol effects on local convergence dominate. This suggests that synoptic forcing plays a significant role in modulating the urbanization-induced land-cover and aerosol effects on individual rainfall event. Hence precipitation changes due to urbanization effects may offset each other under different synoptic conditions, resulting in little changes in mean precipitation at longer timescales.
Urbanization affects climate and hydrological cycle by changing land cover and surface albedo, which releases additional heat to the atmosphere, and by emitting air pollutants, which interact with clouds and radiation (e.g., Shepherd, 2005; Sen Roy and Yuan, 2009; Yang et al., 2011). The most discernible impact of urban land-use change is the urban heat island (UHI) effect that can result in a warmer environment over urban areas than the surrounding areas (Landsberg, 1981; Oke, 1987). In addition to the thermal perturbations, the UHI has been well documented to modify wind patterns (Hjemfelt, 1982), evaporation (Wienert and Kuttler, 2005), atmospheric circulations (Shepherd and Burian, 2003; Baik et al., 2007; Lei et al., 2008), and precipitation around urban areas (Braham, 1979; Inoue and Kimura, 2004). Previous studies have found an increase of warm-season precipitation over and downwind of major cities due to the expanded urban land cover (Huff and Changnon, 1972; Changnon, 1979; Zhong et al., 2015). Recent studies suggested that the underlying urban surface also affects the initiation and propagation of storms (Bornstein and Lin, 2000; Guo et al., 2006) and convective activities in city fringes (Baik et al., 2007; Shepherd et al., 2010).
Concurrently increases in population and anthropogenic activities over urbanized areas increase pollutant emissions and aerosol loading in the atmosphere. Atmospheric aerosols have long been recognized to affect surface and top-of-atmosphere (TOA) radiative fluxes and radiative heating profiles in the atmosphere via aerosol–radiation interactions (ARIs) (e.g., Coakley et al., 1987; Charlson et al., 1992; Hansen et al., 1997; Yu et al., 2006; Qian et al., 2006, 2007, 2015; McFarquhar and Wang, 2006), which tend to induce cooling near the surface and heating at the low and mid-troposphere (Qian et al., 2006; Bauer and Menon, 2012). Anthropogenic aerosols can also affect clouds and precipitation via aerosol–cloud interactions (ACIs) (e.g., Rosenfeld, 2000, 2008; Qian et al., 2010; Fan et al., 2013, 2015; Tao et al., 2012; Zhong et al., 2015). Localized changes in precipitation by strong aerosol perturbations can induce cold pools by evaporation, which may alter the organization of stratocumulus clouds (e.g., Wang and Feingold, 2009; Feingold et al., 2010). Aerosol impacts on deep convective clouds are complicated by the interactions among dynamical, thermodynamical, and microphysical processes. For example, deep convection could be invigorated by aerosols as more cloud water associated with the smaller cloud drops is carried to higher levels where it freezes and releases more latent heat in a polluted environment (Rosenfeld et al., 2008; Khain, 2009; Storer and van den Heever, 2013). Fan et al. (2013) revealed a microphysical effect of aerosols from reduced fall velocity of ice particles that explains the commonly observed increases in cloud top height and cloud cover in polluted environments. Therefore, urbanization may influence precipitation and circulation through multiple pathways that are more difficult to disentangle than the dominant effect on temperature.
As one of the most developed regions in China, the Yangtze River Delta (YRD)
has been experiencing rapid economic growth and intensive urbanization
process during the past 3 decades. With the highest city density and
urbanization level in China, the YRD has become the largest adjacent
metropolitan areas in the world. It covers an area of
9.96
Land-use categories for year
The individual effects of urbanization-induced UHI and aerosol emission on local and regional climate have been examined separately in several modeling studies using short simulations of selected weather episodes at high spatial resolution or multiple-year climate simulations at coarse resolution. To more robustly quantify the urbanization-induced UHI and aerosol effects, convection-permitting simulations may reduce uncertainties in representing convection and its interactions with aerosols, which are parameterized in coarse-resolution models. Additionally, multi-year simulations are needed to understand and quantify the overall effects of land-cover change and aerosols in different large-scale environments (Oleson et al., 2008). In this study, a state-of-the-art regional model coupled with online chemistry (WRF-Chem) and a single-layer urban canopy model (UCM) is used to simulate climate features in the YRD region. The climatic effects of the separate and combined land-cover and aerosol changes induced by urbanization are investigated using a set of 5-year (2006–2010) simulations with a horizontal resolution at convection-permitting scale (3 km). The paper is organized as follows. Section 2 describes the model configuration, experiment design, and model evaluation. The urbanization effects on extreme temperature and precipitation are presented in Sect. 3, followed by a summary of the conclusions in Sect. 4.
The WRF-Chem model (Grell et al., 2005; Fast et al., 2006; Qian et al., 2010) simulates trace gases, aerosols, and meteorological fields interactively (Skamarock and Klemp, 2008; Wang et al., 2009), including ARIs (Zhao et al., 2011, 2013a) and ACIs (Gustafson et al., 2007). The coupled single-layer UCM (Kusaka et al., 2001; Chen et al., 2001) is a column model that uses a simplified geometry with two-dimensional, symmetrical street canyons to represent the momentum and energy exchanges between the urban surface and the atmosphere. The RADM2 (Regional Acid Deposition Model 2) gas chemical mechanism (Stockwell et al., 1990) and the MADE (Modal Aerosol Dynamics Model for Europe) and SORGAM (Secondary Organic Aerosol Model) aerosol module (Schell et al., 2001) are used. Detailed configuration of the above models can be found in Zhao et al. (2010). No cumulus parameterization is used at the convection-permitting resolution. The physical parameterization schemes used in our simulations are listed in Table 1.
Simulations are performed over a model domain centered at
(120.50
Configurations of the WRF physics schemes used in the present study.
Moving spatial anomalies of averaged surface skin temperature
(units:
The dominant land cover within each model grid cell is derived from the US
Geological Survey (USGS) 30 s dataset that includes 24-category land-use
type, except that the land use over urban areas is updated using the stable
nighttime light product (version 4) at 1 km spatial resolution (available at
the National Geophysical Data Center,
The anthropogenic emission fluxes of SO
Annual mean
In order to investigate the individual responses of local and regional
climate to land-cover change and increased aerosol loading, three experiments
(i.e., LU06E06, LU70E70, and LU70E06) are conducted for 5 years from 2006 to
2010. The configurations of land use and aerosol emissions for these
experiments are summarized in Table 2. All three simulations are performed
using the same initial and boundary conditions and physics schemes, but with
different land-use types and/or anthropogenic emissions. LU06E06 is the
control experiment, which represents the “present” (2006) urbanization
level for both land use and aerosol/precursor emissions. LU70E06 uses the
present aerosol emission data but with the land use of the 1970s, which is
derived from the USGS dataset without the nighttime light correction. In
LU70E70, both land use and emissions are set to the conditions of the 1970s.
The differences of LU06E06
Numerical experiments and corresponding urban land use and aerosol emissions.
Analysis strategies for the investigation of urban land-use and/or aerosol effects.
The surface skin temperature simulated in LU06E06 is averaged over 2006–2010
and compared with the MODIS data. A spatial filtering method described by Wu
and Yang (2013) is applied to isolate the heterogeneous climatic forcing of
urbanization. More specifically, for each grid a spatial anomaly is defined
as the departure from the average value over a region centered at each grid.
Then, the moving spatial anomalies are calculated for all the grids with the
moving region acting as a filtering window, which has a size of
1
To further validate the model, the baseline simulation LU06E06 is evaluated against meteorological station observations for 2006–2010. Figure 3 shows the averaged near-surface temperature and precipitation from observations and LU06E06. The simulated spatial pattern of near-surface air temperature agrees well with observations, with high-temperature centers located at meteorological stations in major cities such as Shanghai and Hangzhou. The simulated temperature displays substantial spatial variability associated with heterogeneity in topography, land cover, and other regional forcings. The model captures the general north-to-south gradient of increasing precipitation in the observations. However, the model overestimates precipitation in Shanghai and central Jiangsu Province but underestimates the precipitation in the southwestern part of the domain.
Differences in mean 2 m temperature (units:
Figure 4 shows the differences in 2 m near-surface air temperature
(
Differences in net shortwave fluxes at the surface
(units: W m
The increased aerosols induced by urbanization exert a cooling effect over the entire simulation domain in both summer and winter (Fig. 4b and e). On a domain average, the temperature reduction induced by increased aerosols is less than the warming induced by the UHI effect in both seasons. Therefore, the net urbanization impact (including both land-cover change and aerosol increase) on near-surface temperature is dominated by the UHI warming effect (Fig. 4c and f) resulted from the land-cover change in the YRD.
The effects of urban land-cover change and increased aerosols on surface net
shortwave radiation are shown in Fig. 5. As the building clusters reduce
surface albedo (Oke, 1987), land-cover change increases the net shortwave
radiation over urbanized areas, with an average increase of 9.11 W m
The UHI effect can significantly increase the near-surface temperatures in
summer, thereby exacerbating extreme heat waves in urbanized areas (Stone,
2012). By definition, a heat wave occurs when the near-surface temperature
reaches or exceeds 35
Differences in column burden of PM
Differences in mean summertime
High temperature during heat wave contributes to heat exhaustion or heat
stroke, but the impact of atmospheric humidity on evaporation is also
crucial. Here we use a heat stress index to assess the combined effects of
temperature and humidity on human health due to the UHI effect, expressed as
(Masterson and Richardson, 1979):
Previous studies have provided evidence of urbanization effect on precipitation distribution in and around urban areas (e.g., Shepherd and Burian, 2003; Kaufmann et al., 2007; Miao et al., 2010). Several mechanisms have been proposed for the effects of urbanization on precipitation: (1) the UHI effect can destabilize the planetary boundary layer and trigger convection; (2) increased surface roughness may enhance atmospheric convergence that favors updrafts; (3) building obstruction tends to bifurcate rainfall systems and delays its propagation; (4) the change in land cover decreases local evaporation, (5) anthropogenic emissions increase aerosol loading in the atmosphere, with subsequent effects on precipitation through changes in radiation and cloud processes. These mechanisms contribute to positive and negative changes in precipitation, leading to more complicated effects on precipitation than temperature.
Diurnal cycles of the frequency of summertime extreme rainfall
events for LU70E70 (defined using hourly precipitation intensity above
95th percentile; black lines, right axis) and the differences between
simulations over
Differences in the frequency of summertime extreme rainfall events
(averaged from 12:00 to 20:00 LST) between simulations
In this section we analyze the results of the three 5-year simulations to examine the long-term impact of urbanization on precipitation. The results show that influences of both urban land cover and elevated aerosols on annual and seasonal mean precipitation are relatively small (not shown). This may be due to the urbanization effect for different rainfall events offsetting each other, leading to an overall weak effect on a longer timescale (see Sect. 3.2.2). Here we focus on the frequency of extreme rainfall over the YRD region. Extreme summer rainfall events are defined using hourly precipitation rate that is above 95th percentile at each grid for the period of 2006–2010. Figure 8 shows the diurnal cycles of extreme rainfall frequency and urbanization-induced changes in the areas around Nanjing, Shanghai, and Su-Xi-Chang (shown in Fig. 1b). The frequency of hourly extreme rainfall reaches its maximum at around 16:00–17:00 LST over three urban clusters. Urban land-cover change increases the occurrence of extreme precipitation in the afternoon (12:00 to 20:00 LST). The maximum increase in the frequency of extreme hourly rainfall events for Nanjing, Shanghai, and Su-Xi-Chang can reach 0.86, 1.09, and 0.79 %, respectively, with the peak increase occurring in the late afternoon. In contrast, aerosols exert an opposite impact to substantially reduce the frequency of extreme rainfall in the afternoon by up to 1.05, 0.75, and 0.72 % for Nanjing, Shanghai, and Su-Xi-Chang, respectively. These impacts are significant compared to the maximum frequency of hourly extreme rainfall of about 10 % in each area. However, opposite effects of land-cover and aerosol emission changes result in a small net urbanization effect on extreme precipitation.
Because urbanization influences extreme precipitation primarily in the afternoon, we further analyze extreme rainfall events with a focus on the averages from 12:00 to 20:00 LST. Figure 9 shows the substantial increase in extreme precipitation frequency concentrated over the major metropolitan areas in the YRD, with some compensation in the surrounding areas in general. Aerosols, however, reduce the occurrence of extreme precipitation more uniformly in most areas of the domain. The most significant influence of aerosols is found in the northwest part of the domain where aerosol concentrations increase the most downwind of the urban centers (Fig. 6a). Similar to the effects on surface temperature and solar radiation (Figs. 4 and 5), aerosols have a substantial impact on the occurrence of extreme precipitation over a wider area than the effects of urban land-use change.
How do changes in land cover and aerosols modulate extreme rainfall frequency? Figure 10a shows the diurnal time–height cross section of the impact of urban land cover (i.e., the difference between LU06E06 and LU70E06) on temperature and divergence averaged over the three city clusters (Nanjing, Shanghai, and Su-Xi-Chang). Air temperature over the urbanized areas increases significantly in the afternoon (from 12:00 to 18:00 LST) due to the UHI effect. The warming and the increased roughness length in urban areas favor convergence in the lower atmosphere and divergence above. As a result, the mean updraft increases over the urbanized areas in the afternoon (Fig. 10b), which increases cloud water from the lower to middle troposphere in the afternoon. Shortly before noon, there is a small reduction in low clouds, which may be related to the reduced relative humidity due to warmer temperature and/or reduced evaporation from the urban land cover, the so-called urban dry island effect (e.g., Hage, 1975; Wang and Gong, 2001). The increase in cloud water in the afternoon is consistent with the enhanced updrafts. This mechanism potentially explains the increased frequency of extreme precipitation in urban areas in the afternoon (e.g., Craig and Bornstein, 2002; Rozoff et al., 2003; Wan et al., 2013; Zhong and Yang, 2015a, b).
To understand the aerosol-induced reduction in extreme rainfall events, we analyze the diurnal cycle of aerosol effect (i.e., the difference LU70E06 and LU70E70) on radiative heating, vertical velocity, and net solar radiation at the surface (Fig. 11). As BC emission rates are relatively high in the YRD region (Fig. 2d), aerosols heat the atmosphere due to absorption of solar radiation during daytime (from 08:00 to 17:00 LST). As a result of absorption and scattering of solar radiation by aerosols, less solar radiation reaches the surface. These changes at the surface and in the atmosphere stabilize the atmosphere and reduce convective intensity in the afternoon (from 14:00 to 20:00 LST), which reduces the frequency of extreme rainfall events (Koren et al., 2004; Qian et al., 2006; Zhao et al., 2006, 2011; Fan et al., 2007). Although aerosols can enhance precipitation through cloud microphysical changes that invigorate convection (e.g., Khain et al., 2009; Rosenfeld et al., 2008; Fan et al., 2013), aerosol radiative effects generally dominate in China because of the high aerosol optical depth and strong light-absorbing aerosol properties (Yang et al., 2011; Fan et al., 2015).
Time–height cross sections of differences between LU70E06 and
LU70E70 in radiative heating profile (shade; units: K d
Rain rate (units: mm h
The time evolution of precipitation (units: mm h
The impacts of urbanization-induced UHI and aerosols on precipitation may be
highly variable under different synoptic conditions that influence the
atmospheric circulation and cloud and boundary layer processes. Precipitation
changes due to urbanization effects may offset each other under different
synoptic conditions, leading to an overall weak effect on mean precipitation
at longer timescales as discussed in Sect. 3.2.1. We select two typical
heavy late-afternoon rainfall events with different background circulations
over the YRD region. Case A occurred from 08:00 on 23 June 2006 to 08:00 on 24 June
2006 (LST) and case B occurred from 08:00 on 1 July 2006 to 08:00 on 2 July 2006 (LST).
Figure 12a and d show the mean precipitation rate and 850 hPa winds for
case A and case B, respectively. Southwesterly flow dominates the entire
region in case A (Fig. 12a), while in case B (Fig. 12d) southwesterly and
northwesterly winds dominate the southern and northern parts of precipitation
area, respectively. The averaged background wind speed in case B is much
stronger than that in case A, representing stronger synoptic forcing in
case B. The effects of urban land-cover change and aerosols on precipitation
for the case A (case B) are illustrated in Fig. 12b and c (Fig. 12e and f),
respectively. Both cases show significant precipitation responses to the
forcing of urban land cover and aerosols. We can see that urban land cover
increases the rainfall intensity in case A but aerosols decrease
precipitation over the urbanized area (Fig. 12b and c). The precipitation
response to urban land cover and aerosols is just the opposite in case B
(Fig. 12e and f). Figure 13a and d illustrate the evolution of precipitation
in region R1 (Fig. 12a) and R2 (Fig. 12d), respectively, for the two cases.
In both cases, rainfall mainly occurred between 08:00 and 20:00 LST. The
corresponding impacts of urban land cover and aerosols are shown in
Fig. 13b–c and e–f for cases A and B, respectively. In case A, the urban
land cover substantially increases the precipitation intensity in the
afternoon with a maximum increase of 6.87 mm h
Why do urban land cover and aerosols exert opposite effects on precipitation
during the two rainfall events? Here we attempt to answer this question by
examining the dynamical and thermodynamical changes induced by the UHI and
aerosols using the moisture flux convergence (MFC), which is defined as
The time–height cross sections of differences in moisture flux
convergence (shaded; units: 10
Same as Fig. 14 but for differences in the CON term (shaded;
units: 10
Figure 14a and b illustrate the time–height cross sections of changes in moisture flux convergence and cloud water mixing ratio induced by land-cover and aerosol changes over the region R1 (Fig. 12a) during the rainy period in case A. Urban land cover enhances the convergence of moisture fluxes in the lower troposphere, which results in increased precipitation (Fig. 14a). In contrast, aerosols weaken the convergence of moisture fluxes and thus reduce precipitation (Fig. 14b). These changes are consistent with those associated with extreme rainfall changes shown in Fig. 10. Interestingly for case B over R2, urban land cover weakens the convergence of moisture fluxes (Fig. 14c) and thus suppresses precipitation (Fig. 13e) from 08:00 on 1 July 2006 to 02:00 on 2 July 2006 (LST). Aerosols, however, enhance the convergence of moisture fluxes over R2 (Fig. 14d) and thus increase precipitation (Fig. 13f). These results establish obvious correspondence between moisture flux convergence changes and the precipitation response to urban land cover and aerosols in the two rainfall events and suggest different processes may dominate the moisture flux convergence changes for the two cases.
Figure 15 presents the time–height cross section of the changes in the two
terms of MFC, i.e., CON and MA, induced by
land-cover and aerosol changes averaged over R1 (Fig. 12a) for case A and
over R2 (Fig. 12d) for case B. Urban land cover enhances the wind convergence
over R1 in case A (Fig. 15a), leading to an increase in CON by up to
Same as Fig. 15 but for differences in the first term
(
The significant differences in the responses of MA between the two cases are
related to different background circulations during the two events (Fig. 12a
and d). Weaker southwesterly flow dominates the entire region in case A
(Fig. 12a), while in case B (Fig. 12d) stronger southwesterly and
northwesterly winds dominate the southern and northern parts of precipitation
area, respectively. The changes in MA could be further decomposed into three
terms, as shown below in Eq. (3):
In summary, case B represents stronger synoptic forcing than case A. The stronger winds and larger spatial coverage of clouds and precipitation associated with the larger-scale synoptic system weakens the UHI and aerosol effects through ventilation and changes in radiation, resulting in weaker CON and larger MA changes. Conversely, with weaker synoptic forcing, the stronger UHI and aerosol effects enhance the changes in CON while MA effects are smaller due to the weaker background winds. Therefore, our results highlight the distinguishing role of synoptic forcing on how urban land cover and aerosol influence the dynamical and thermodynamical environments and precipitation.
In this study, the state-of-the-art WRF-Chem model coupled with a single-layer UCM is run at convection-permitting scale to investigate the influences of urbanization-induced land-cover change and elevated aerosol concentrations on local and regional climate in the YRD in China. A 5-year period (2006–2010) is selected for multi-year simulations to investigate urbanization effects on extreme events and the role of synoptic forcing. Three experiments were conducted with different configurations of land cover and aerosol emissions: (1) urban land and emissions in 2006, (2) urban land in the 1970s and emissions in 2006, and (3) urban land and emissions in the 1970s. The experiment with the 2006 land-use type and anthropogenic emissions reproduces the observed spatial patterns of near-surface air temperature and precipitation fairly well.
The expanded urban land cover and increased aerosols have opposite impacts on
the near-surface air temperature. The urban land-use change increases 2 m
air temperature due to the UHI effect in commercial areas with a
domain-averaged increase of 1.49
The urban land-cover change and increased aerosols have opposite effects on the frequency of extreme rainfall during summer. The UHI effect leads to more frequent extreme precipitation over the urbanized area in the afternoon because of an enhanced near-surface convergence and vertical motion. In contrast, aerosol tends to decrease the frequency of extreme precipitation because of its cooling effect near the surface and heating effect (by light-absorbing particles) above, leading to an increased atmospheric stability and weakened updrafts. Additional aerosols can also induce decreases in the frequency of extreme precipitation over non-urban areas, particularly in the downwind area of the city clusters.
The effects of both urban land cover and increased aerosols on summertime rainfall vary with synoptic weather systems and environmental conditions. Two late-afternoon rainfall events are selected for in-depth analysis. For the two cases, urbanization exerts similar impacts on local-scale convergence and mean wind speed, which modify the strength of moisture transport. More specifically, the effect of urban land cover increases local-scale convergence due to the UHI-induced circulation and reduces low-level wind speed, while aerosols have an opposite effect due to the cooling near the surface. We found that the impacts of urban land cover and aerosol on precipitation are not only determined by their effect on local-scale convergence but also modulated by the large-scale weather systems. Our analyses suggest that synoptic forcing plays a significant role in how urbanization-induced land cover and aerosols influence individual rainfall events. Although the two rainfall events selected for the analysis do not represent all types of precipitation events in the YRD region, they demonstrate how the effect of urbanization on precipitation may vary and offset each other under different synoptic conditions, leading to an overall weak effect on mean precipitation at longer timescales. To further quantify urbanization effects, uncertainties in anthropogenic emissions and heating, unresolved urban building and streets structure, and representation in aerosol–cloud interactions and cloud microphysics in the model should be investigated in future studies. Further investigation is also needed to have a better and more comprehensive understanding of the complicated mechanisms through which urbanization influences heavy rainfall under a full range of weather conditions.
All model results are archived on a PNNL cluster and available upon request. Please contact Yun Qian (yun.qian@pnnl.gov).
The authors declare that they have no conflict of interest.
The contributions of PNNL authors are supported by the US Department of Energy's Office of Science as part of the Regional and Global Climate Modeling Program and Atmospheric System Research (ASR) program. The contribution of Shi Zhong and Xiu-Qun Yang is supported by the National Basic Research Program of China (2010CB428504), Jiangsu Collaborative Innovation Center for Climate Change, and the Scholarship Award for Excellent Doctoral Student granted by China Scholarship Council. The work of Ben Yang is supported by the National Natural Science Foundation of China (41305084). Computations were performed using resources of the National Energy Research Scientific Computing Center (NERSC) at Lawrence Berkeley National Laboratory and PNNL Institutional Computing. The Pacific Northwest National Laboratory is operated for DOE by Battelle Memorial Institute under contract DE-AC05-76RL01830. Edited by: L. Zhang Reviewed by: two anonymous referees