The Positive and Negative Effects of Green Space on PM_2.5 Concentrations: A Review

Abstract

Fine particulate matter (PM_2.5) can have considerable negative effects on human health. An increasing number of scholars are finding that green space can not only decrease PM_2.5 levels but also exacerbate PM_2.5 levels. Few scholars have provided comprehensive reviews on this subject. This study reviews research from 1995 to 2024, including 118 studies based on a search of three databases (Web of Science, Engineering Village, and ResearchGate). We found that at the macro (e.g., city-wide) and meso (e.g., high-density built-up areas) scales, most studies report that green space can play a positive role in mitigating PM_2.5 concentrations. However, at the micro-scale under specific temporal conditions, green spaces may increase PM_2.5 concentrations in some micro-environments. Whether vegetation reduces or elevates local PM_2.5 levels, these processes are influenced by various factors, including green space configuration, microclimatic conditions, built-environment characteristics, and emission source distributions. Mechanistically, vegetation can both decrease ambient PM_2.5 levels through deposition, adsorption, and absorption and block its dispersion. In the process of exploring and optimizing the effect of greening on PM_2.5, we should not only consider these factors in isolation but also account for the environmental factors that can significantly change the effect. Based on our review of a myriad of studies from different disciplinary backgrounds and scales, we propose an optimization strategy consisting of promoting ventilation through weakening sources and strengthening sinks.

1. Introduction

Particulate matter (PM) in the air can have significant negative effects on the health of humans, animals, and plants [1,2]. PM can be further divided based on particle sizes. These categories include total suspended particulate matter (TSP), with a particle size less than 100 μm; PM₁₀, with a particle size less than 10 μm; PM_2.5, with a particle size less than 2.5 μm; and PM₁, with a particle size less than 1 μm [2]. Kim et al. [3] compared the difference between PM_2.5 and PM₁₀ based on a review of numerous relevant articles and pointed out the considerable difference between the two:

(1)

the lifetime of PM₁₀ is in the range of minutes to hours, while for PM_2.5, it is in the range of days to weeks [4,5];

(2)

the travel distance of PM₁₀ is only 1–10 km, whereas this distance is always 100–1000 km for PM_2.5;

(3)

for PM_2.5, the sources can be summarized as ‘combustion of coal, oil, gasoline; transformation products of NO_x, SO₂, and organics including biogenic organics’ [6]. For PM₁₀, the sources can be summarized as ‘resuspension of soil tracked onto roads and streets; Suspension from disturbed soils’ [6];

(4)

the chemical compositions of PM_2.5 and PM₁₀ are also quite different [4,5].

(5)

some researchers have pointed out that the coarse particles in the PM₁₀ range mainly stem from natural sources (such as crustal elements), while PM_2.5 particles mainly originate from anthropogenic sources, such as sulfates, nitrates, and black carbon [5,6].

Thus, there are considerable differences between PM_2.5 and PM₁₀. More targeted research is needed for PM_2.5 control. Simply borrowing previous research results on PM₁₀ or TSP for the control of PM_2.5 pollution may lead to erroneous or even disastrous consequences. In addition, the World Health Organization (WHO) has recommended managing PM_2.5 instead of PM₁₀ because of the closer relationship between PM_2.5 and human health [7].

PM_2.5 can have significant negative effects on humans’ respiratory and cardiovascular systems. Exposure to PM_2.5 increases the risk of many diseases, such as asthma, ischemic heart disease, type 2 diabetes, lung cancer, chronic kidney disease, lower respiratory tract infections, myocardial infarction, chronic obstructive pulmonary disease, Alzheimer’s disease, and strokes [8,9,10]. PM_2.5 can also have more severe and specific effects on vulnerable groups such as children, the elderly, and pregnant women. For example, PM_2.5 exposure has been linked to adverse effects on children’s neurodevelopment and immune function [11].

Many researchers have pointed out that urban green space (i.e., areas with vegetation coverage, including parks, street trees, residential green spaces, and vertical greening systems) [12,13] can have a significant effect on PM_2.5. These green spaces vary in scale and structure, influencing PM_2.5 through deposition, blockage, adsorption, and absorption [14,15]. The deposition effect is mainly caused by atmospheric particles falling on the underlying surface through aerodynamic effects, such as gravity-induced settling and turbulent diffusion, and plants changing the conditions of the underlying surface and thus influencing the aerodynamic effects. The blocking effect is dependent on the complex structures of branches and leaves, which block the movement of particles. The adsorption effect is mainly dependent upon the surface structures of plant leaves and special secretions for dust retention. Absorption mainly refers to the ability of plant stomata to inhale PM_2.5 [14,15]. However, not all researchers have found that green space plays a significant role in reducing PM_2.5. For example, Qiu et al. [16] pointed out that larger particles can easily be retained by plants, while smaller particles are susceptible to the influence of atmospheric diffusion and diffuse over longer distances. In addition, Zhou et al. [17] used Wuhan, China, as a case study to explore the effects of urban forest spatial patterns on PM_2.5 concentrations. Their findings showed that different forest layouts had little effect, with only a modest reduction of 1–2% in PM_2.5 levels. Nevertheless, they recommend planting multiple trees on the outskirts of the urban metropolitan development zone to create a forest belt, aiming for a forest coverage rate of over 60%.

Since the 1990s, the number of studies related to the effect of green spaces on PM_2.5 concentrations has increased rapidly. With various studies conducted at different times, in different places, at different scales, and using different methods, researchers have gradually realized that green space can not only reduce PM_2.5 concentrations but also exacerbate PM_2.5 pollution, especially in some micro-environments. To the best of our knowledge, no one has comprehensively reviewed this topic from a global perspective. Thus, in this study, we aim to comprehensively review this subject and guide urban green space managers in managing green space to reduce atmospheric PM_2.5 pollution in a more effective way. To achieve this aim, the following objectives are set:

• Comprehensively review articles related to the effect of green space on PM_2.5 concentrations from different scales;
• Compare the positive and negative effects of green space on PM_2.5 based on relevant articles;
• Consider how to make full use of green space to reduce PM_2.5 concentrations based on relevant research.

2. Method

The global databases Web of Science, Engineering Village, and ResearchGate were selected as our search engines. The timeframe we set to search for articles ranged from 1995 to 2024. The types of articles selected were journal articles and reviews. In order to retrieve all the relevant articles related to the effects of green space on PM_2.5 concentrations, we did not limit the fields, scales, countries, or cities investigated. As long as an article was related to our search aim (i.e., it explored the positive and/or negative effects of green space on PM_2.5), it was selected. To comprehensively retrieve all the relevant articles, we listed all the terms synonymous with PM_2.5 and green space and combined them one by one (Figure 1 shows all the keywords used). Every time we used one combination (like fine particles and parks) in a search, we then downloaded all the relevant articles and then changed to another combination. We continued to employ this approach until we had used all the effective combinations. Ultimately, we retrieved and downloaded 118 articles.

3. Effects of Regional or Urban Green Space on PM_2.5

Many researchers have explored the effects of green spaces on PM_2.5 over broad areas, including the globe, regions, entire cities, central urban areas, core urban areas, and urban administrative districts. These areas feature many different land-use types and include residential areas, industrial areas, road and transportation facility areas, and public management and public service areas. The sources of PM_2.5 pollution in these areas are also quite various, including transportation, domestic environments, industry, dust, and agriculture.

3.1. The Amount of PM_2.5 Reduced by Regional or Urban Green Spaces

Feng et al. [18] used satellite image products and an empirical algorithm to study vegetation-related dry deposition and found that under the combined influence of the PM_2.5 mass concentration and deposition velocity, the global average dry deposition and the mean vegetation-related dry deposition from PM_2.5 sources reached 0.47 g/(m²·yr) and 0.26 g/(m²·yr), respectively. In the Amazon, Central Asia, and East Asia, levels of vegetation-related dry deposition are high because of dense vegetation or severe pollution. Zhai et al. [19] revealed that forests in the Beijing–Tianjin–Hebei (BTH) region, China, reduced 19,400 t, 19,200 t, 16,400 t, and 12,700 t of PM_2.5 in 2015, 2016, 2017, and 2018, respectively, according to the results of a mathematical model. Li et al. [20] explored the removal of PM_2.5 by green infrastructure in Shenyang, China, based on a dry deposition model. They found that between 2000 and 2019, PM_2.5 removal, PM_2.5 removal flux, and the PM_2.5 removal rate increased by 20.64 mg/year, 0.0258 g/m², and 0.377%/year, respectively. The removal of PM_2.5 by green spaces in the central urban area is relatively low. Areas with high green space coverage remove more PM_2.5. Jo et al. [21] used several South Korean cities as an example to explore the reduction in PM_2.5 levels induced by street trees in South Korea. They found that the average street tree density and street tree and shrub coverage rate were approximately 1.0 trees/100 m² and 12.5 ± 1.4%, respectively. The PM_2.5 deposition per unit area was 2.0 ± 0.3 kg/(ha·yr). Thus, from a quantitative perspective, all the above research found that green space reduces the total amount of PM_2.5 suspended in the air. However, this does not mean that green space only has a positive effect on PM_2.5 concentrations. It can precipitate the accumulation of PM_2.5 in local environments, resulting in excessively high local PM_2.5 concentrations. The later parts of this article provided a more detailed introduction to the negative effects.

3.2. The Effect of Regional and Urban Green Space Structure on PM_2.5

Many researchers have explored the effects of green space structure on PM_2.5 concentrations. Relevant studies have been conducted at different times, in different areas, on different scales, and using different methods. Even though some studies have been conducted around similar green space structural features, and there have been some clearly disparate viewpoints. For example, a study on 277 Chinese prefecture-level cities found that the higher the degree of ecological land use, the lower the PM_2.5 concentration. The order of influence, from high to low, is as follows: forests, shrubs, and grassland. The PM_2.5 distribution showed spatial clustering and decreased from the inner to the outer circles [22]. However, Chen et al. [23] indicated that the effects of green space coverage on PM_2.5 are various under different conditions. For example, the correlation between green coverage and PM_2.5 is high in the spring at the 1–2 km scale but moderate in the summer at the 2 km, 5 km, and 6 km scales. When the PM_2.5 concentration is greater than 75 µg/m³, the correlation between PM_2.5 concentration and green cover is non-significant. In addition, some researchers have noted that different structural green spaces do not cause significant differences in PM_2.5 concentrations, even if this can cause significant differences in PM₁₀ concentrations, based on a study conducted in Baoji, China [16]. In contrast, different sizes of green spaces can lead to obvious differences in PM_2.5 pollution in the surrounding environment. In addition, green spaces covering less than 2 ha do not have a significant effect on PM_2.5 in the surrounding environment [16]. One study found that the effect of greening on reducing PM_2.5 is stronger in built-up areas in which there is frequent human activity, particularly in developed areas with low vegetation coverage. However, this effect is limited in areas with high green coverage, and it is more pronounced in areas with low PM_2.5 concentrations than in areas with high concentrations [24]. In addition, Zhang et al. [25] found that on the leaf scale, forest species are quite active in accumulating PM_2.5 on the surfaces of their leaves. However, on the canopy scale, horizontal vegetation collection is the major process whereby PM_2.5 is removed, and the contribution of vertical sedimentation/emission can be ignored. At the landscape scale, Ai et al. [26] found that dry deposition is the main PM_2.5 removal process during rain-free days. One study found that PM_2.5 concentration changes are more sensitive to normalized difference vegetative index (NDVI) in the southern part of China than in the north of China. They attributed this phenomenon to the fact that vegetation in the northern part of China predominantly consist of coniferous, mixed coniferous and broad-leaved, and deciduous broad-leaved forests. In contrast, in the south, the vegetation predominantly consists of evergreen broadleaf forests. The NDVI values in the south are higher than those in the north. When similar green space changes occur in north and south of China, there is a greater influence on PM_2.5 concentration in the south of China than the north. Furthermore, it is important to note that the most effective urban greening structure for PM_2.5 reduction is not static but can vary with conditions. This is exemplified by the winter season, when PM_2.5 pollution is typically more severe and the demand for mitigation by green spaces is higher. During this period, however, the capacity of deciduous trees to remove PM_2.5 is significantly limited due to the loss of their foliage. This seasonal disparity was quantified in a 2015 study of Beijing, which found that within the city’s Sixth Ring Road, evergreen trees removed PM_2.5 at 2.7 times the efficiency of deciduous tree [27]. Niu et al. [28] have conducted research in Shenyang, but they found that canopy density has a significant positive relation with PM_2.5 concentrations and that shrub richness is negatively correlated with PM_2.5 concentrations. They attributed these findings to the existence of a certain threshold: canopy density shows a negative correlation with PM_2.5 concentrations when it is below this threshold, while a positive correlation may occur when this threshold is exceeded.

In contrast to the differences in the views held by the researchers mentioned above, there is a high degree of consistency in the findings regarding the effect of vertical greening features (green walls and green roofs) on PM_2.5. Srbinovska et al. [29] highlighted the significant role of green walls in reducing PM_2.5 pollution. Similarly, based on their studies conducted in Santiago, Chile, Viecco et al. [26] and Vera et al. [30] found that green roofs and living walls can effectively reduce PM_2.5 levels. Furthermore, Vera et al. [31] noted that the greater the biodiversity pertaining to green roofs and walls, the more effective they are at reducing PM_2.5 pollution. This phenomenon may be attributed to the fact that relevant vertical greening studies were conducted within a small research scope to exclude the influence of some irrelevant variables and only focused on one side of the vertical greening (i.e., they did not consider the PM_2.5 concentration in the narrow space between the vertical-greening area and walls or roofs).

3.3. The Effect of Regional and Urban Green Space Spatial Patterns on PM_2.5 from the Perspective of Landscape Ecology and Morphology

As the landscape pattern analysis of landscape ecology and morphological spatial patterns becomes increasingly acknowledged and popular, an increasing number of researchers are using these methods to explore the effects of the spatial pattern features of green space on PM_2.5 concentrations. By comparing the relevant studies in Table 1, Table 2 and Table 3, it can be found that regional- or urban-scale studies exhibit distinct patterns between green space landscape pattern characteristics and PM_2.5 concentrations due to differences in research scope, time, scale, and other factors. Previous researchers have extensively used different landscape ecology and morphological indicators to consider the effect of green patch amount, area, shape, aggregation or dispersion, and edge characteristics on PM_2.5 [32,33,34,35,36,37,38,39,40,41,42,43]. Among the studies [32,33,34,35,36,37,38,39,40,41,42,43], the indicator related to green patch amount characteristics includes number of patch (NP), patch density (PD); the indicators related to green patch area characteristics include percentage of landscape (PLAND), Green space coverage ratio (G_Ratio), total area (TA), largest patch index (LPI), total core area (TCA), perimeter-area ratio (PARA), area-weight perimeter-area fractal dimension index (AWFRAC); the indicators related to green patch shape characteristics include landscape shape index (LSI), shape index (SI), area-weighted mean shape index (Shape_AM), area-weighted mean patch fractal dimension (FRACE_MN); the indicators related to green patch accumulation or dispersion degree include aggregation index (AI), landscape division index (LDI), contiguity index (CONTIG), aggregation index (AI), Shannon’s diversity index (SHDI), Shannon’s evenness index (SHEI) cohesion index (COHESION), mean Euclidean nearest-neighbor index (ENN_MN); the indicators related to green patch edge characteristics include total edges (TE), edge density (ED).

Table 1 shows that when studies are conducted at different times and scales, the green space spatial pattern optimization strategies considered optimal can differ. For example, Cai et al. [32] conducted a study in southeastern Fujian, China, and found that the green space LSI was positively correlated with PM_2.5 at scales of 1000 m, 2000 m, 3000 m, 4000 m, and 5000 m, but at the 500 m scale, there was no significant correlation between LSI and PM_2.5. Zhan et al. [33] pointed out that PLAND is closely related to the reduction in PM_2.5 exposure risk on a larger scale, while AI, ED, and SI are closely related to PM_2.5 exposure on a local scale. The various conclusions reached based on the above factors in regional- and urban-scale studies further demonstrate the necessity of further categorizing PM_2.5 sources (such as transportation, domestic factors, and industry) and greening types (such as road, park, and residential greening) for related research.

The studies in Table 2 show that the optimal spatial optimization strategy also differs based on location or land use. For example, Yang et al. [34] found that patch type (square, subsidiary, or other park green spaces) has a greater effect on PM_2.5 concentrations than the type of corridor (protective, ribbon-garden-in-a-park, or road subsidiary green spaces). Table 3 shows studies conducted using different methods, these studies can have some slight differences in findings. For example, in Bi et al.’s [35] study, both geographically weighted regression and the ordinary least squares method indicated that both branches and islands are positively correlated with PM_2.5. Geographically weighted regression further indicated that 20.88% and 33.62% of regional branches and islands are negatively correlated with PM_2.5. The authors’ [35] geographically weighted regression results indicated that perforations and loops can effectively reduce PM_2.5 pollution in local spaces (such as Wuchang District), while the ordinary least squares method showed that there is no correlation between perforations and loops and PM_2.5.

Therefore, different cities have some similarities and differences in terms of optimizing green space patterns to reduce PM_2.5 concentrations. In general, most studies in Table 1, Table 2 and Table 3 [32,33,34,35,36,37,38,39,40,41,42,43] have found that increasing the green space ratio and ensuring there is a more concentrated distribution of large green spaces are more effective in reducing PM_2.5 concentrations. However, there are certain differences in the optimal spatial pattern optimization measures at different times, locations, and scales. Therefore, the optimization strategy for using urban greening to reduce PM_2.5 needs to be tailored to local conditions, and there is no optimal method that is suitable for all situations.

Table 1. Studies showing, at different times and scales, that optimal green space spatial pattern optimization strategies differ.

Authors and Year	Location	Landscape Pattern	Relevant Findings	Limitations
Longyan Cai et al., (2020) [32]	Southeast Fujian province, China	TA, LSI AI	1. TA of green space is negatively correlated with PM2.5 at the 500 m, 1000 m, 2000 m, 3000 m, 4000 m, and 5000 m buffer zone scales. 2. The LSI of green space is positively correlated with PM2.5 at the 1000 m, 2000 m, 3000 m, 4000 m, and 5000 m buffer zone scales, while there is no significant correlation between LSI and PM2.5 at the 500 m buffer zone scale. 3. The AI of green space is negatively correlated with PM2.5 at the 500 m, 2000 m, 3000 m, 4000 m, and 5000 m buffer zone scales. However, there is no significant correlation between AI and PM2.5 at the 1000 m buffer zone scale.	No limitations were mentioned.
Wei Chen et al., (2022) [36]	Beijing, China	G_Ratio, PD, LPI, LDI, Shape_AM.	In the spring, in the 100 m buffer zone, LSI (representing shape complexity) has the greatest effect on PM2.5 concentrations. In the summer, SHAPE_AM has the greatest effect on PM2.5 in the 300 m to 1000 m buffer zones. In the autumn, G_Ratio has the greatest effect on PM2.5 concentrations. In the winter, G_Ratio, LPI, LSI, and Shape_AM have the greatest effects on PM2.5 at the 300 m, 500 m, 1000 m, and 1500 m buffer zone scales, respectively.	This paper did not consider extreme weather conditions or the 3D effect of vegetation on PM 2.5 dispersion. The relationship between annual precipitation and PM2.5 concentrations in the study area was uncertain. The results did not explain the mechanism of action regarding PM2.5 concentrations in the study area.
Longyan Cai et al., (2022) [37]	Fujian province, China	TA, LSI, COHESION, and AI	In the spring, the TA of green spaces is negatively correlated with PM2.5 at a scale of 4 km and above, the COHESION of green spaces is negatively correlated with PM2.5 at a scale of 5 km, and the AI of green spaces is negatively correlated with PM2.5 concentrations at a scale of 3 km and above. The LSI of green spaces is positively correlated with PM2.5 at scales of 2, 3, and 5 km. In the summer, the vast majority of landscape indicators have no significant correlation with PM2.5, and at the 1 km scale, COHESION is negatively correlated with PM2.5. In the autumn, the TA of green spaces is negatively correlated with PM2.5 at scales of 0.5 km and 4 km, while the LSI of green spaces is positively correlated with PM2.5 at scales of 2 km and 3 km. In the winter, the TA of green spaces is negatively correlated with PM2.5 at a scale of 1 km and above. The COHESION of green spaces at 4 km and above are negatively correlated with PM2.5. The LSI of green spaces is positively correlated with PM2.5 at scales of 0.5 km and 2 km.	No limitations were mentioned.
Qingming Zhan et al., (2022) [33]	Wuhan, China	PLAND, ED, TE, TCA, PD, SI, PARA, CONTIG, AI, SHDI	When the risk of PM2.5 exposure itself is relatively high, green landscape has a poor effect on reducing PM2.5 exposure risk. PLAND mitigated PM2.5 exposure risk at larger scales, while AI, ED, and SI were associated with reduced PM2.5 exposure at local scales.	The 1 km × 1 km resolution employed was not high enough. The annual interval prevented the authors from capturing short-term PM2.5 exposure risks. This study may not be applicable to other atmospheric pollutants.
Shibo Bi et al., (2022) [38]	Wuhan, China	Core, islet, edge, bridge, branch, loop, and perforation	A scale of 2 km × 2 km is optimal for analyzing the spatial patterns of green space morphology and the spatial heterogeneity of PM2.5. Edges and bridges show a negative correlation with PM2.5 in some cases and a positive correlation in others. Cores and perforations show a negative correlation with PM2.5, while islets and branches show a positive correlation with PM2.5. There was no correlation between loops and PM2.5 between 2000 and 2015, but there was a negative correlation from 2015 to 2020. The point–line–polygon combination had the best effect on reducing PM2.5 pollution.	This study only focused on Wuhan. More cities should be studied. The authors of this study could have considered the relationship between urban green space morphologies (UGSMs) and seasonal and monthly PM2.5 data in greater depth.
Yu Li et al., (2024) [39]	Shenyang, China	PD, AWMSI, FRACE_MN, AI, ENN_MN, SHAPE_AM	The correlation between PM2.5 concentrations and green landscape patterns is significantly influenced by season. For example, SHAPE_AM and FRACE_MN are negatively correlated with PM2.5 in the spring, summer, and autumn. AI and LPI are negatively correlated with PM2.5. PD is positively correlated with PM2.5 in different seasons.	The study has little consideration on the roles of meteorological parameters (e.g., air temperature, radiation, relative humidity and wind speed) and anthropogenic activities in PM2.5 concentrations. Shenyang is the only sample city of the research and the result may not be applicable to some other cities.

Table 1. Studies showing, at different times and scales, that optimal green space spatial pattern optimization strategies differ.

Authors and Year	Location	Landscape Pattern	Relevant Findings	Limitations
Longyan Cai et al., (2020) [32]	Southeast Fujian province, China	TA, LSI AI	1. TA of green space is negatively correlated with PM2.5 at the 500 m, 1000 m, 2000 m, 3000 m, 4000 m, and 5000 m buffer zone scales. 2. The LSI of green space is positively correlated with PM2.5 at the 1000 m, 2000 m, 3000 m, 4000 m, and 5000 m buffer zone scales, while there is no significant correlation between LSI and PM2.5 at the 500 m buffer zone scale. 3. The AI of green space is negatively correlated with PM2.5 at the 500 m, 2000 m, 3000 m, 4000 m, and 5000 m buffer zone scales. However, there is no significant correlation between AI and PM2.5 at the 1000 m buffer zone scale.	No limitations were mentioned.
Wei Chen et al., (2022) [36]	Beijing, China	G_Ratio, PD, LPI, LDI, Shape_AM.	In the spring, in the 100 m buffer zone, LSI (representing shape complexity) has the greatest effect on PM2.5 concentrations. In the summer, SHAPE_AM has the greatest effect on PM2.5 in the 300 m to 1000 m buffer zones. In the autumn, G_Ratio has the greatest effect on PM2.5 concentrations. In the winter, G_Ratio, LPI, LSI, and Shape_AM have the greatest effects on PM2.5 at the 300 m, 500 m, 1000 m, and 1500 m buffer zone scales, respectively.	This paper did not consider extreme weather conditions or the 3D effect of vegetation on PM 2.5 dispersion. The relationship between annual precipitation and PM2.5 concentrations in the study area was uncertain. The results did not explain the mechanism of action regarding PM2.5 concentrations in the study area.
Longyan Cai et al., (2022) [37]	Fujian province, China	TA, LSI, COHESION, and AI	In the spring, the TA of green spaces is negatively correlated with PM2.5 at a scale of 4 km and above, the COHESION of green spaces is negatively correlated with PM2.5 at a scale of 5 km, and the AI of green spaces is negatively correlated with PM2.5 concentrations at a scale of 3 km and above. The LSI of green spaces is positively correlated with PM2.5 at scales of 2, 3, and 5 km. In the summer, the vast majority of landscape indicators have no significant correlation with PM2.5, and at the 1 km scale, COHESION is negatively correlated with PM2.5. In the autumn, the TA of green spaces is negatively correlated with PM2.5 at scales of 0.5 km and 4 km, while the LSI of green spaces is positively correlated with PM2.5 at scales of 2 km and 3 km. In the winter, the TA of green spaces is negatively correlated with PM2.5 at a scale of 1 km and above. The COHESION of green spaces at 4 km and above are negatively correlated with PM2.5. The LSI of green spaces is positively correlated with PM2.5 at scales of 0.5 km and 2 km.	No limitations were mentioned.
Qingming Zhan et al., (2022) [33]	Wuhan, China	PLAND, ED, TE, TCA, PD, SI, PARA, CONTIG, AI, SHDI	When the risk of PM2.5 exposure itself is relatively high, green landscape has a poor effect on reducing PM2.5 exposure risk. PLAND mitigated PM2.5 exposure risk at larger scales, while AI, ED, and SI were associated with reduced PM2.5 exposure at local scales.	The 1 km × 1 km resolution employed was not high enough. The annual interval prevented the authors from capturing short-term PM2.5 exposure risks. This study may not be applicable to other atmospheric pollutants.
Shibo Bi et al., (2022) [38]	Wuhan, China	Core, islet, edge, bridge, branch, loop, and perforation	A scale of 2 km × 2 km is optimal for analyzing the spatial patterns of green space morphology and the spatial heterogeneity of PM2.5. Edges and bridges show a negative correlation with PM2.5 in some cases and a positive correlation in others. Cores and perforations show a negative correlation with PM2.5, while islets and branches show a positive correlation with PM2.5. There was no correlation between loops and PM2.5 between 2000 and 2015, but there was a negative correlation from 2015 to 2020. The point–line–polygon combination had the best effect on reducing PM2.5 pollution.	This study only focused on Wuhan. More cities should be studied. The authors of this study could have considered the relationship between urban green space morphologies (UGSMs) and seasonal and monthly PM2.5 data in greater depth.
Yu Li et al., (2024) [39]	Shenyang, China	PD, AWMSI, FRACE_MN, AI, ENN_MN, SHAPE_AM	The correlation between PM2.5 concentrations and green landscape patterns is significantly influenced by season. For example, SHAPE_AM and FRACE_MN are negatively correlated with PM2.5 in the spring, summer, and autumn. AI and LPI are negatively correlated with PM2.5. PD is positively correlated with PM2.5 in different seasons.	The study has little consideration on the roles of meteorological parameters (e.g., air temperature, radiation, relative humidity and wind speed) and anthropogenic activities in PM2.5 concentrations. Shenyang is the only sample city of the research and the result may not be applicable to some other cities.

Table 2. Studies showing that the optimal green space spatial pattern optimization strategy differs depending on location land use.

Authors and Year	Location	Landscape Pattern	Relevant Findings	Limitations
Ming Chen et al., (2019) [40]	Hefei, Shanghai, Wuhan, Nanjing, and Hangzhou, China	Core, islet, perforation, edge, loop, bridge, and branch	Higher proportions of cores and bridges help reduce PM2.5 concentrations. The higher the proportion of perforation, islets, and edges, the higher the PM2.5 concentration. A neighborhood green space’s capacity to attenuate PM2.5 pollution vanished when it was smaller than 200 m and was maximized when its size was within 400–500 m.	Five megacities are located in one climate zone. So future studies can consider cities from different sizes and regions. The study lacks the examination of seasonal fluctuations and spatial forms of green spaces.
Zhiyu Fan et al., (2022) [41]	Wuhan metropolitan area, China	Neighboring green space, PLAND, AWFRAC, ED, AI, COHESION, PD	Aggregation and the area and shape of green and blue spaces were the most negative attributes related to PM2.5 pollution. The effect of green and blue spaces on PM2.5 pollution varies in different spaces, and even if there is a mitigating effect in one space, another space in the same city may have a positive effect on PM2.5 concentrations.	The study was affected by the modifiable areal unit problem; the multi-scale neighboring heterogeneous effects were unknown; and some anthropogenic factors and meteorological variables were not considered.
Lijuan Yang et al., (2023) [34]	Yangtze River Delta (Zhejiang Province, Jiangsu Province, and Shanghai), and Fujian Province, China	PLAND, PD, ED, LPI, AREA_MN, LSI, COHESION, LDI, SHEI	Forestland can have a significant influence on PM2.5. The landscape pattern indices of shrubs and grasslands have weaker influences on PM2.5 concentration. Enhancing forest connectivity (COHESION) and aggregation (AI) reduces fragmentation can improve PM2.5 dust retention. Indices like PD, ED, and SHEI positively correlate with PM2.5, while LDI and SHDI show weak correlations.	1. This paper did not explore seasonal differences in regional PM2.5 concentrations. 2. This paper did not consider the effect of inter-regional transmission on regional PM2.5 concentrations.
Saiwei Luo et al., (2023) [42]	Central urban area of Nanchang city, China	NDVI, area, perimeter, LSI	The effective distance of green space affecting PM2.5 is normally less than 100 m. The effect of patches (square green space, subsidiary green space, or other form of park green space) on PM2.5 concentrations is greater than that of corridors (protective green spaces, ribbon gardens in a park, and road subsidiary green spaces). The higher the area and NDVI of green space, the lower the PM2.5 concentration.	1. The data were limited. 2. This paper did not focus on emission sources at the micro-scale.

Table 2. Studies showing that the optimal green space spatial pattern optimization strategy differs depending on location land use.

Authors and Year	Location	Landscape Pattern	Relevant Findings	Limitations
Ming Chen et al., (2019) [40]	Hefei, Shanghai, Wuhan, Nanjing, and Hangzhou, China	Core, islet, perforation, edge, loop, bridge, and branch	Higher proportions of cores and bridges help reduce PM2.5 concentrations. The higher the proportion of perforation, islets, and edges, the higher the PM2.5 concentration. A neighborhood green space’s capacity to attenuate PM2.5 pollution vanished when it was smaller than 200 m and was maximized when its size was within 400–500 m.	Five megacities are located in one climate zone. So future studies can consider cities from different sizes and regions. The study lacks the examination of seasonal fluctuations and spatial forms of green spaces.
Zhiyu Fan et al., (2022) [41]	Wuhan metropolitan area, China	Neighboring green space, PLAND, AWFRAC, ED, AI, COHESION, PD	Aggregation and the area and shape of green and blue spaces were the most negative attributes related to PM2.5 pollution. The effect of green and blue spaces on PM2.5 pollution varies in different spaces, and even if there is a mitigating effect in one space, another space in the same city may have a positive effect on PM2.5 concentrations.	The study was affected by the modifiable areal unit problem; the multi-scale neighboring heterogeneous effects were unknown; and some anthropogenic factors and meteorological variables were not considered.
Lijuan Yang et al., (2023) [34]	Yangtze River Delta (Zhejiang Province, Jiangsu Province, and Shanghai), and Fujian Province, China	PLAND, PD, ED, LPI, AREA_MN, LSI, COHESION, LDI, SHEI	Forestland can have a significant influence on PM2.5. The landscape pattern indices of shrubs and grasslands have weaker influences on PM2.5 concentration. Enhancing forest connectivity (COHESION) and aggregation (AI) reduces fragmentation can improve PM2.5 dust retention. Indices like PD, ED, and SHEI positively correlate with PM2.5, while LDI and SHDI show weak correlations.	1. This paper did not explore seasonal differences in regional PM2.5 concentrations. 2. This paper did not consider the effect of inter-regional transmission on regional PM2.5 concentrations.
Saiwei Luo et al., (2023) [42]	Central urban area of Nanchang city, China	NDVI, area, perimeter, LSI	The effective distance of green space affecting PM2.5 is normally less than 100 m. The effect of patches (square green space, subsidiary green space, or other form of park green space) on PM2.5 concentrations is greater than that of corridors (protective green spaces, ribbon gardens in a park, and road subsidiary green spaces). The higher the area and NDVI of green space, the lower the PM2.5 concentration.	1. The data were limited. 2. This paper did not focus on emission sources at the micro-scale.

Table 3. Using different study methods can also lead to differences in findings.

Authors and Year	Location	Landscape Pattern	Relevant Findings	Limitations
Shibo Bi et al., (2022) [35]	Wuhan, China	Core, islet, edge, bridge, branch, loop and perforation, TA, NP, PD, TE, ED, LSI, COHESION, SHDI, SHEI	The conclusions drawn from different spatial analysis methods are generally similar, but there are many significant differences in details. For example, both geographically weighted regression and the ordinary least squares method indicated that both branches and islands are positively correlated with PM2.5. Geographically weighted regression further indicated that 20.88% and 33.62% of regional branches and islands are negatively correlated with PM2.5. The geographically weighted regression results indicate that perforations and loops can effectively reduce PM2.5 pollution in local spaces (such as Wuchang District), while the ordinary least squares method revealed that there is no correlation between perforations and loops and PM2.5.	The seasonal and monthly effects of urban green spaces on PM2.5 need to be studied further. Differences in pollutant levels across cities suggest that this framework should be applied to more cities to acquire broader insights.
Yuanyuan Chen et al., (2022) [43]	Wuhan, China	TA, LPI, CONTAG, SHDI	According to the spatial lag model, LPI and SHDI are negatively correlated with PM2.5, while TA and CONTAG do not have a clear relationship with PM2.5. Based on ordinary linear regression, TA, SHDI, and LPI were negatively correlated with PM2.5. CONTAGE does not have a significant correlation with PM2.5.	Some PM2.5 sources and meteorology data were absent.

Table 3. Using different study methods can also lead to differences in findings.

Authors and Year	Location	Landscape Pattern	Relevant Findings	Limitations
Shibo Bi et al., (2022) [35]	Wuhan, China	Core, islet, edge, bridge, branch, loop and perforation, TA, NP, PD, TE, ED, LSI, COHESION, SHDI, SHEI	The conclusions drawn from different spatial analysis methods are generally similar, but there are many significant differences in details. For example, both geographically weighted regression and the ordinary least squares method indicated that both branches and islands are positively correlated with PM2.5. Geographically weighted regression further indicated that 20.88% and 33.62% of regional branches and islands are negatively correlated with PM2.5. The geographically weighted regression results indicate that perforations and loops can effectively reduce PM2.5 pollution in local spaces (such as Wuchang District), while the ordinary least squares method revealed that there is no correlation between perforations and loops and PM2.5.	The seasonal and monthly effects of urban green spaces on PM2.5 need to be studied further. Differences in pollutant levels across cities suggest that this framework should be applied to more cities to acquire broader insights.
Yuanyuan Chen et al., (2022) [43]	Wuhan, China	TA, LPI, CONTAG, SHDI	According to the spatial lag model, LPI and SHDI are negatively correlated with PM2.5, while TA and CONTAG do not have a clear relationship with PM2.5. Based on ordinary linear regression, TA, SHDI, and LPI were negatively correlated with PM2.5. CONTAGE does not have a significant correlation with PM2.5.	Some PM2.5 sources and meteorology data were absent.

4. The Effect of Different Types of Green Space on PM_2.5

PM_2.5 originates from different sources, such as traffic, industrial activity, domestic factors, and agriculture. PM_2.5 from different sources can have different characteristics in relation to chemical composition, physical characteristics, and emission patterns. Thus, the treatment of PM_2.5 pollution requires measures tailored to different sources. There are also different types of green spaces, such as road, residential, and industrial. Different types of green space influence different sources of PM_2.5 pollution differently for various reasons, like the spatial patterns of different green spaces and the proximity between green spaces and pollution sources. For example, traffic PM_2.5 is considered a line-source pollution; road green spaces are distributed in a strip shape on both sides of the roadway. Industrial sources are considered non-point source pollution, and green spaces in industrial areas are usually scarce and scattered around factory yards.

4.1. The Effect of Road Green Space on PM_2.5 in Street Canyons

Many researchers have explored the effects of street greening on PM_2.5 concentrations in street canyons. Table 4 and Table 5 are studies conducted based on monitoring and computational fluid dynamics (CFD) simulation respectively. Both tables include studies which found (1) road green space can reduce PM_2.5; (2) road green space can exacerbate PM_2.5 pollution level; (3) depend on specific situation, green space can either increase or reduce PM_2.5 concentration. Because studies have been conducted in different street canyon environments and used different criteria to judge the effects, there are many obvious differences in their findings. The most significant difference, as found by some researchers, relates to the positive effects of road greening on PM_2.5 concentration reduction [44,45,46,47,48,49]. However, others [50,51,52,53,54,55] have pointed out that road greening can exacerbate PM_2.5 pollution in specific micro-environments. All these researchers have provided reasonable causes. For example, Qin et al. [46] found that road greening had positive effects with respect to reducing PM_2.5 concentrations. Their evidence was mainly based on a comparison of PM_2.5 concentrations between scenarios with trees and a no-tree base-case scenario. However, Jin et al. [50], by comparing the reduction rates of PM_2.5 in the vertical direction between scenarios with and without trees, found that road greening can intercept PM_2.5 in street canyon environments. In addition, some researchers [56,57,58,59,60,61,62,63] have pointed out that depending on the situation, road greening may have both positive and negative effects on PM_2.5. For example, Chen et al. [57] conducted a study based on actual measurements and found that some vegetation barriers have a positive reduction effect, while others have a negative effect, as found by comparing the differences in PM_2.5 concentrations before and after plant barriers. However, their research was limited by two factors. Firstly, their experimental research could not completely rule out the interference of PM_2.5 sources other than traffic. Secondly, the turbulence of PM_2.5 caused by the wind itself is a very complex turbulent flow. The situation is easily affected by turbulence. In some cases, plant barriers exacerbate PM_2.5 levels, while in other cases, they reduce PM_2.5 levels. This also means that the differences mainly stem from the different criteria used in different studies to determine whether the effect is positive or negative in different situations, and all research findings are reasonable in specific research environments.

Currently, a great deal of research suffers from limitations in regard to analyzing the effect of road greening on PM_2.5 by using the difference in PM_2.5 concentrations between observation points and control points. Firstly, it is difficult to distinguish the extent to which a reduction in the PM_2.5 concentration in a green space is due to a series of complex processes, including deposition, blockage, adsorption, and absorption, and the extent to which the difference in PM_2.5 concentration before and after road greening is caused by the large increase in PM_2.5 concentrations in front of an area subjected to road greening due to a reduction in wind speed induced by plants. In addition, regardless of whether there is a high concentration of PM_2.5 in the roadway area, as long as there is a large concentration difference between the roadside (the control point) and the area subjected to road greening (the experimental point), a good reduction effect can be recorded. However, if the PM_2.5 concentration in the roadway area is excessively high, it is undoubtedly detrimental to the health of passengers who open their windows for ventilation and pedestrians crossing the road. Therefore, we call for more attention to the absolute concentration changes in different areas rather than the relative concentration differences between different areas.

In addition, currently, researchers only focus on selecting specific indicators to explore the effect of road greening on PM_2.5 in a chosen street canyon environment. Although these efforts can reasonably explain the reasons behind the phenomenon, they often fail to provide insight into some differences between the research in question and previous studies. Therefore, we suggest that in the future, researchers, when conducting research on the effect of road greening on PM_2.5, should first fully recognize the dual nature (positive and negative) of the effect of road greening on street canyon PM_2.5. New experiments should be designed based on existing research and experimental settings and findings, and several sets of experiments should be set up with reference to the research of researchers who have put forward different views in the past in order to better reveal the special circumstances under which road greening will have a positive effect on reducing PM_2.5 concentrations and when it will exacerbate PM_2.5 pollution. Contrary to the traditional view that greening is beneficial for improving air quality, more and more researchers are finding that green spaces may also exacerbate local pollution. However, there is currently a lack of insight into the conditions and mechanisms behind the occurrence of related contradictory phenomena.

Table 4. Monitoring studies that explore the effect of street green space on PM_2.5 concentrations.

Monitoring Studies
Authors, Year, and Place	Effect of Green Space on PM2.5	Parameter(s)	Location	Main Findings	Limitations
Yingyi Zhao et al., (2018) [45]	Green space mitigates PM2.5 pollution levels	Tree species, height, and diameter at breast height	Three roads in Nantong, Jiangsu Province, China	The reduction in PM2.5 by roadside trees in April 2015 ranged from 2 g to 43 g. Tree height and diameter at breast height are important factors affecting PM2.5 removal capacity.	Street lamps and road signs, which share point clouds with street trees, may be difficult to distinguish from the trees, and improvements are needed to effectively separate the mixed point clouds. The model may need further enhancement to adapt to more complex street environments. The method for measuring breast height requires further refinement to improve accuracy. Limited data and equation used in i-tree.
Emily Riondato et al., (2020) [47]	Green space mitigates PM2.5 pollution levels	Trees	Dublin, Ireland	Eighty trees in a street canyon can remove roughly 3 kg of PM2.5 yearly. Based on monitoring data, trees lead to a 126% air quality improvement.	There were no wind data; tree effects cannot be separated from airflow. The study had a short duration, lasting only five days. It also lacks a seasonal scope. The study was conducted at a single site. Using one location limits generalization. Missing tree traits: Height and crown size were not recorded. Traffic proxy flaw: Noise used as a proxy may be inaccurate.
Sijia Jin et al., (2014) [50]	Green space exacerbates PM2.5 pollution levels	Tree species, quantity, breast height diameter, height, crown width, height under branches, lowest leaf layer height, tree spacing	Six typical street canyons in the city center of Shanghai, China	Compared with the control condition without trees, the vertical reduction rate for PM2.5 was significantly reduced when roadside trees were planted, indicating that the tree crown can intercept PM2.5 in the street canyon. Canopy density, leaf area index, and wind speed change rate are important factors affecting the PM2.5 reduction rate. The PM2.5 vertical reduction rate is positively correlated with canopy density and the leaf area index (LAI).	No limitations were mentioned.
Jiangying Xu et al., (2023) [54]	Green spaces exacerbate PM2.5 pollution levels	Green view index (GVI), NDVI	An old town in Wuhan, China	Using the three-dimensional parameter of the GVI, it was found that street greening hinders the diffusion of PM2.5 at the street level, leading to an increase in local PM2.5 concentrations. The correlation between the GVI and PM2.5 concentrations is weak in the morning and strong in the afternoon. Street greening has a significant effect on PM2.5 within a 300-m radius.	This study did not analyze more detailed vegetation information, such as vegetation morphology and tree species. The generalizability of the research results may be limited, as the study was conducted only during specific time periods and in localized areas of the old urban district of Wuhan.
Xiaoshuang Wang et al., (2020) [56]	Depending on the specific situation, green space either mitigates or exacerbates PM2.5 pollution levels	Canopy density	Street canyon in Wuhan, China	Canopy density can have a significant effect on PM2.5, with the greatest decrease in PM2.5 concentration occurring when the canopy density is within the range of 30% to 36%. A sparse canopy in high-PM2.5-concentration areas is beneficial for reducing PM2.5, while an overly dense canopy will hinder PM2.5 diffusion.	For the treeless areas in this study, it is impossible to completely exclude the influence of airflow or pollutant concentration from other adjacent areas (with trees).
Xiaoping Chen et al., (2021) [57]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	Vegetation barriers at different road sections	One east–west road and two north–south roads in Wuhan, China	The authors studied the effect of vegetation barriers on the reduction rate for PM2.5 levels (equivalent to the reduction rate from the roadway to the green belt and then to the sidewalk in the horizontal direction) and found that vegetation barriers’ ability to reduce PM2.5 levels was not consistent. Some vegetation barriers have a positive reduction effect (about half), while others have a negative effect (about the other half). The porosity of protective forests at heights of 0–2 m and 6–8 m is negatively correlated with the efficiency of PM2.5 emission reduction. The porosity of protective forests at heights of 2–4 m and 4–6 m is positively correlated with the efficiency of PM2.5 emission reduction.	This study overlooked the differences in adsorption abilities between deciduous and evergreen plants, as the research was conducted during the flourishing period from June to August. This study had less consideration on the influence of local meteorological factors.
Suyeon Kim et al., (2017) [58]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	Terrain conditions, number of green belts	Three main roads of the Songpa area in Seoul, South Korea	Among single, double, and triple rows of roadside trees, triple rows have the best effect on reducing PM2.5. Shrubs can play an important role in reducing traffic-related PM2.5. An overly dense distribution of roadside trees may lead to an increase in PM2.5 concentrations, and maintaining a certain distance between roadside trees is beneficial for PM2.5 dispersion.	This study could not include all environmental variables within urban areas. The survey was conducted only during the spring and winter periods characterized by significant changes in plant foliage, thus failing to fully account for the effects of other seasons. The study did not address key factors such as the difference in efficiency between evergreen and deciduous plants, the optimal distance between PM2.5 sources and plants, and variations in plant species.
F. H. Liang et al., (2023) [60]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	Crown density and horizontal permeability	Jinjing highway, Tianjin, China	High-canopy-density and high-transparency green spaces and low-canopy-density and low-transparency green spaces are relatively unfavorable for the diffusion of PM2.5. In contrast, low-canopy-density and high-transparency green spaces and high-canopy-density and low-permeability green spaces are relatively favorable for the diffusion of PM2.5. On sidewalks, green spaces with high canopy density and high transparency are the most unfavorable for the diffusion of PM2.5. In green spaces, green spaces with high canopy density and high transparency, as well as those with low canopy density and low transparency, are the most unfavorable for the diffusion of PM2.5.	No limitations were mentioned.
Congzhe Liu et al., (2022) [61]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	The effects of different combinations of plants at different road sections	Seven roads in Nanjing, China	The order of efficiency for the combinations in terms of the PM2.5 reduction rate is as follows: trees + shrubs + grass > trees + shrubs > trees + Grass > Trees > Lawn > Blank control. Vegetation can have both positive and negative effects on PM2.5 concentrations.	This study only focused on the reduction in PM2.5 induced by plant communities, without considering the effect of other pollutants and the relationship between other pollutants and PM2.5.

Table 4. Monitoring studies that explore the effect of street green space on PM_2.5 concentrations.

Monitoring Studies
Authors, Year, and Place	Effect of Green Space on PM2.5	Parameter(s)	Location	Main Findings	Limitations
Yingyi Zhao et al., (2018) [45]	Green space mitigates PM2.5 pollution levels	Tree species, height, and diameter at breast height	Three roads in Nantong, Jiangsu Province, China	The reduction in PM2.5 by roadside trees in April 2015 ranged from 2 g to 43 g. Tree height and diameter at breast height are important factors affecting PM2.5 removal capacity.	Street lamps and road signs, which share point clouds with street trees, may be difficult to distinguish from the trees, and improvements are needed to effectively separate the mixed point clouds. The model may need further enhancement to adapt to more complex street environments. The method for measuring breast height requires further refinement to improve accuracy. Limited data and equation used in i-tree.
Emily Riondato et al., (2020) [47]	Green space mitigates PM2.5 pollution levels	Trees	Dublin, Ireland	Eighty trees in a street canyon can remove roughly 3 kg of PM2.5 yearly. Based on monitoring data, trees lead to a 126% air quality improvement.	There were no wind data; tree effects cannot be separated from airflow. The study had a short duration, lasting only five days. It also lacks a seasonal scope. The study was conducted at a single site. Using one location limits generalization. Missing tree traits: Height and crown size were not recorded. Traffic proxy flaw: Noise used as a proxy may be inaccurate.
Sijia Jin et al., (2014) [50]	Green space exacerbates PM2.5 pollution levels	Tree species, quantity, breast height diameter, height, crown width, height under branches, lowest leaf layer height, tree spacing	Six typical street canyons in the city center of Shanghai, China	Compared with the control condition without trees, the vertical reduction rate for PM2.5 was significantly reduced when roadside trees were planted, indicating that the tree crown can intercept PM2.5 in the street canyon. Canopy density, leaf area index, and wind speed change rate are important factors affecting the PM2.5 reduction rate. The PM2.5 vertical reduction rate is positively correlated with canopy density and the leaf area index (LAI).	No limitations were mentioned.
Jiangying Xu et al., (2023) [54]	Green spaces exacerbate PM2.5 pollution levels	Green view index (GVI), NDVI	An old town in Wuhan, China	Using the three-dimensional parameter of the GVI, it was found that street greening hinders the diffusion of PM2.5 at the street level, leading to an increase in local PM2.5 concentrations. The correlation between the GVI and PM2.5 concentrations is weak in the morning and strong in the afternoon. Street greening has a significant effect on PM2.5 within a 300-m radius.	This study did not analyze more detailed vegetation information, such as vegetation morphology and tree species. The generalizability of the research results may be limited, as the study was conducted only during specific time periods and in localized areas of the old urban district of Wuhan.
Xiaoshuang Wang et al., (2020) [56]	Depending on the specific situation, green space either mitigates or exacerbates PM2.5 pollution levels	Canopy density	Street canyon in Wuhan, China	Canopy density can have a significant effect on PM2.5, with the greatest decrease in PM2.5 concentration occurring when the canopy density is within the range of 30% to 36%. A sparse canopy in high-PM2.5-concentration areas is beneficial for reducing PM2.5, while an overly dense canopy will hinder PM2.5 diffusion.	For the treeless areas in this study, it is impossible to completely exclude the influence of airflow or pollutant concentration from other adjacent areas (with trees).
Xiaoping Chen et al., (2021) [57]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	Vegetation barriers at different road sections	One east–west road and two north–south roads in Wuhan, China	The authors studied the effect of vegetation barriers on the reduction rate for PM2.5 levels (equivalent to the reduction rate from the roadway to the green belt and then to the sidewalk in the horizontal direction) and found that vegetation barriers’ ability to reduce PM2.5 levels was not consistent. Some vegetation barriers have a positive reduction effect (about half), while others have a negative effect (about the other half). The porosity of protective forests at heights of 0–2 m and 6–8 m is negatively correlated with the efficiency of PM2.5 emission reduction. The porosity of protective forests at heights of 2–4 m and 4–6 m is positively correlated with the efficiency of PM2.5 emission reduction.	This study overlooked the differences in adsorption abilities between deciduous and evergreen plants, as the research was conducted during the flourishing period from June to August. This study had less consideration on the influence of local meteorological factors.
Suyeon Kim et al., (2017) [58]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	Terrain conditions, number of green belts	Three main roads of the Songpa area in Seoul, South Korea	Among single, double, and triple rows of roadside trees, triple rows have the best effect on reducing PM2.5. Shrubs can play an important role in reducing traffic-related PM2.5. An overly dense distribution of roadside trees may lead to an increase in PM2.5 concentrations, and maintaining a certain distance between roadside trees is beneficial for PM2.5 dispersion.	This study could not include all environmental variables within urban areas. The survey was conducted only during the spring and winter periods characterized by significant changes in plant foliage, thus failing to fully account for the effects of other seasons. The study did not address key factors such as the difference in efficiency between evergreen and deciduous plants, the optimal distance between PM2.5 sources and plants, and variations in plant species.
F. H. Liang et al., (2023) [60]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	Crown density and horizontal permeability	Jinjing highway, Tianjin, China	High-canopy-density and high-transparency green spaces and low-canopy-density and low-transparency green spaces are relatively unfavorable for the diffusion of PM2.5. In contrast, low-canopy-density and high-transparency green spaces and high-canopy-density and low-permeability green spaces are relatively favorable for the diffusion of PM2.5. On sidewalks, green spaces with high canopy density and high transparency are the most unfavorable for the diffusion of PM2.5. In green spaces, green spaces with high canopy density and high transparency, as well as those with low canopy density and low transparency, are the most unfavorable for the diffusion of PM2.5.	No limitations were mentioned.
Congzhe Liu et al., (2022) [61]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	The effects of different combinations of plants at different road sections	Seven roads in Nanjing, China	The order of efficiency for the combinations in terms of the PM2.5 reduction rate is as follows: trees + shrubs + grass > trees + shrubs > trees + Grass > Trees > Lawn > Blank control. Vegetation can have both positive and negative effects on PM2.5 concentrations.	This study only focused on the reduction in PM2.5 induced by plant communities, without considering the effect of other pollutants and the relationship between other pollutants and PM2.5.

Table 5. CFD simulation studies that explore the effect of street green space on PM_2.5 concentrations.

CFD Simulation Studies
Authors, Year and Place	Effect	Parameter(s)	Location	Main Findings	Limitations
Bo Hong et al., (2017) [44]	Green space mitigates PM2.5 pollution levels	Street aspect ratio, tree crown shape, leaf area index	He qingyuan near Tsinghua university, Beijing, China	From the perspective of capturing PM2.5 (situation considered: street aspect ratios of 0.5, 1.0, and 2.0; leaf area densities of 0.5, 1.5, and 2.5 m2/m3; and tree crown shapes such as conical, spherical, and cylindrical), when the street aspect ratio is 1.0, the leaf area density (LAD) is 0.5, and the tree crown is conical, the PM2.5 capture effect is the worst. The combination of a street aspect ratio of 1.0 and an LAD of 1.5 is considered the most effective way to capture PM2.5. The order of maximum reduction rates regarding PM2.5 in different canopy shapes, from low to high, is as follows: conical, spherical, and cylindrical.	Only typical meteorological data for Beijing were used, without accounting for the recent more severe high-PM2.5 conditions. This study assumes that the deposition velocity on tree leaves is constant. The revised generalized drift flux model does not account for solar radiation or the convective heat exchange between trees and the environment.
Hongqiao Qin et al., (2020) [46]	Green space mitigates PM2.5 pollution levels	Street aspect ratio, tree species, tree height, crown diameter, crown volume, leaf area index, crown base height	Xi’an, China	The study considered three street aspect ratios (0.45, 0.90, and 1.80) and five tree species (Liriodendron chinense, Gingko biloba, Aesculus chinensis, Acer buergerianum, and Cedrus deodar) Liriodendron chinense had the strongest PM2.5 adsorption and purification ability. The height, crown diameter, and crown volume of trees in street canyons are the main factors affecting the reduction rate for pedestrian-height PM2.5, while the effect of leaf area index and crown base height is relatively small.	No limitations were mentioned.
Wei Wang et al., (2022) [48]	Green space mitigates PM2.5 pollution levels	Building height, street width, green space coverage	Street canyon in Heifei, China	Reducing building height, increasing road width, and increasing street canyon greening are beneficial for reducing PM2.5 concentrations in street canyons.	No limitations were mentioned.
Na-Ra Jeong et al., (2023) [49]	Green space mitigates PM2.5 pollution levels	Planting structure (different combinations of trees and shrubs in different areas within the urban road range), wind direction	Girin-daero region in Jeonju-si, Jeollabuk-do, South Korea	The PM2.5 concentration deviations of different planting structures increase with the increase in wind speed. In a model using a mixed planting structure consisting of trees and shrubs and a central green space zone, the PM2.5 reduction effect was found to be more significant than the other structures investigated in most locations, so this type of road greening structure should be given priority consideration.	The type of street is limited. This study did not take phenological characteristics into account. When quantifying PM2.5 concentration changes, the study had very limited consideration on the mechanisms by which plants reduce PM2.5 levels, such as dispersion and absorption.
Xinming Jin et al., (2017) [51]	Green spaces exacerbate PM2.5 pollution levels	LAI, street aspect ratio, wind speed	Street canyon environment in Beijing, China	This study considered street aspect ratios of 0.5, 1, and 2 and found that buildings and trees that reduce wind speed can easily promote the aggregation of PM2.5 in street canyons, leading to an increase in concentrations. Therefore, higher wind speeds are more beneficial for reducing PM2.5 pollution in street valleys. In this study, trees had a minor effect on PM2.5 deposition flux, and their removal efficiency for PM2.5 was limited.	Because the numerical model could not fully replicate the complex effects of tree distribution, traffic volume, human activities, and solar radiation, the simulation results show some discrepancies relative to the actual measurements.
Sasu Karttunen et al., (2020) [52]	Green spaces exacerbate PM2.5 pollution levels	Wind direction, tree species, number of green belts, planting structure (trees, shrubs, etc.)	Helsinki, Finland	This study’s main finding was as follows: with respect to the effect of roadside trees on pedestrian PM2.5 concentrations, their effect on reducing ventilation is more significant than their dry deposition effect, and roadside trees can lead to an increase in PM2.5 concentrations. However, the authors point out that if street greening can be carefully planned, PM2.5 concentrations can be significantly reduced.	This study was limited by the number of scenarios and inflow conditions examined due to the restrictions imposed by computational resources. The study focuses on the aerodynamic effects of trees on the flow of air while overlooking the shading and thermal effects of the trees. Biogenic volatile organic compounds were not considered. Due to the lack of effective vehicle-induced turbulence (VIT) parameterization for neighborhood-scale large eddy simulation (LES), VIT was neglected.
Junyou Liu et al., (2022) [53]	Green spaces exacerbate PM2.5 pollution levels	Tree height, crown width	A road in Changsha, China	Tall trees with wide crowns can hinder PM2.5 dispersion and may lead to an increase in PM2.5 concentrations in street canyons.	No limitations were mentioned.
Junyou Liu et al., [55] 2023	Green spaces exacerbate PM2.5 pollution levels	The distance between trees	A road in Changsha, China	A certain distance between street trees can promote PM2.5 dispersion and is beneficial for reducing PM2.5 concentrations in street canyons.	The mechanism through which street trees influence PM2.5 concentrations cannot be directly gleaned from the simulation results.
Riccardo Buccolieri et al., (2018) [59]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	LAD, wind direction and speed	Marylebone Road street canyon, London, the UK	In a perpendicular wind environment, planting more roadside trees can exacerbate PM2.5 concentrations in street canyons. In a parallel wind environment, planting more roadside trees is beneficial for reducing the concentration of PM2.5 in street canyons. In general, the positive deposition effects are greater for increased LAD. In addition, perpendicular winds may counterbalance the negative aerodynamic effects near places close to trees.	The study is limited to the case study investigates to cases characterized by similar geometries subject to perpendicular and parallel winds. Research confirmed that multiple variables determine trees’ impact on urban air pollutant concentrations.
Huiyu He et al., (2023) [62]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	Tree height, trunk height crown width	A block near Huacheng Avenue and Huasui Road in Guangzhou, China	A 10-m-high roadside tree can help dilute PM2.5 in the vertical direction, while a tree with a high trunk is better able to remove traffic-related pollutants at pedestrian height and promote the diffusion of pollutants. An excessively wide tree crown can make it difficult for pollutants to escape from the neighborhood, thereby increasing pedestrian exposure levels.	The study only focuses on a certain source of pollution in a limited case area of hot and humid climatic region. The study lacks mathematical analysis illustrating the PM2.5 deposition through roadside trees and particulate diffusion along building clusters during different time period.
Xiaoyu Tian et al., (2024) [63]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	Street size, building density, building height, tree height, trunk height, crown diameter, road strip type, spacing between roadside trees	A block in Guangzhou, China	Roadside trees can absorb PM2.5, but they may also cause PM2.5 to accumulate at the street level, thereby affecting the diffusion of PM2.5 in densely populated buildings. By planting trees with an appropriate height, crown width, leaf area index, and spacing within a certain range, the adsorption capacity of roadside trees can be improved.	The valid sample for this study is limited. The model needs more experiments to identify accuracy.

Table 5. CFD simulation studies that explore the effect of street green space on PM_2.5 concentrations.

CFD Simulation Studies
Authors, Year and Place	Effect	Parameter(s)	Location	Main Findings	Limitations
Bo Hong et al., (2017) [44]	Green space mitigates PM2.5 pollution levels	Street aspect ratio, tree crown shape, leaf area index	He qingyuan near Tsinghua university, Beijing, China	From the perspective of capturing PM2.5 (situation considered: street aspect ratios of 0.5, 1.0, and 2.0; leaf area densities of 0.5, 1.5, and 2.5 m2/m3; and tree crown shapes such as conical, spherical, and cylindrical), when the street aspect ratio is 1.0, the leaf area density (LAD) is 0.5, and the tree crown is conical, the PM2.5 capture effect is the worst. The combination of a street aspect ratio of 1.0 and an LAD of 1.5 is considered the most effective way to capture PM2.5. The order of maximum reduction rates regarding PM2.5 in different canopy shapes, from low to high, is as follows: conical, spherical, and cylindrical.	Only typical meteorological data for Beijing were used, without accounting for the recent more severe high-PM2.5 conditions. This study assumes that the deposition velocity on tree leaves is constant. The revised generalized drift flux model does not account for solar radiation or the convective heat exchange between trees and the environment.
Hongqiao Qin et al., (2020) [46]	Green space mitigates PM2.5 pollution levels	Street aspect ratio, tree species, tree height, crown diameter, crown volume, leaf area index, crown base height	Xi’an, China	The study considered three street aspect ratios (0.45, 0.90, and 1.80) and five tree species (Liriodendron chinense, Gingko biloba, Aesculus chinensis, Acer buergerianum, and Cedrus deodar) Liriodendron chinense had the strongest PM2.5 adsorption and purification ability. The height, crown diameter, and crown volume of trees in street canyons are the main factors affecting the reduction rate for pedestrian-height PM2.5, while the effect of leaf area index and crown base height is relatively small.	No limitations were mentioned.
Wei Wang et al., (2022) [48]	Green space mitigates PM2.5 pollution levels	Building height, street width, green space coverage	Street canyon in Heifei, China	Reducing building height, increasing road width, and increasing street canyon greening are beneficial for reducing PM2.5 concentrations in street canyons.	No limitations were mentioned.
Na-Ra Jeong et al., (2023) [49]	Green space mitigates PM2.5 pollution levels	Planting structure (different combinations of trees and shrubs in different areas within the urban road range), wind direction	Girin-daero region in Jeonju-si, Jeollabuk-do, South Korea	The PM2.5 concentration deviations of different planting structures increase with the increase in wind speed. In a model using a mixed planting structure consisting of trees and shrubs and a central green space zone, the PM2.5 reduction effect was found to be more significant than the other structures investigated in most locations, so this type of road greening structure should be given priority consideration.	The type of street is limited. This study did not take phenological characteristics into account. When quantifying PM2.5 concentration changes, the study had very limited consideration on the mechanisms by which plants reduce PM2.5 levels, such as dispersion and absorption.
Xinming Jin et al., (2017) [51]	Green spaces exacerbate PM2.5 pollution levels	LAI, street aspect ratio, wind speed	Street canyon environment in Beijing, China	This study considered street aspect ratios of 0.5, 1, and 2 and found that buildings and trees that reduce wind speed can easily promote the aggregation of PM2.5 in street canyons, leading to an increase in concentrations. Therefore, higher wind speeds are more beneficial for reducing PM2.5 pollution in street valleys. In this study, trees had a minor effect on PM2.5 deposition flux, and their removal efficiency for PM2.5 was limited.	Because the numerical model could not fully replicate the complex effects of tree distribution, traffic volume, human activities, and solar radiation, the simulation results show some discrepancies relative to the actual measurements.
Sasu Karttunen et al., (2020) [52]	Green spaces exacerbate PM2.5 pollution levels	Wind direction, tree species, number of green belts, planting structure (trees, shrubs, etc.)	Helsinki, Finland	This study’s main finding was as follows: with respect to the effect of roadside trees on pedestrian PM2.5 concentrations, their effect on reducing ventilation is more significant than their dry deposition effect, and roadside trees can lead to an increase in PM2.5 concentrations. However, the authors point out that if street greening can be carefully planned, PM2.5 concentrations can be significantly reduced.	This study was limited by the number of scenarios and inflow conditions examined due to the restrictions imposed by computational resources. The study focuses on the aerodynamic effects of trees on the flow of air while overlooking the shading and thermal effects of the trees. Biogenic volatile organic compounds were not considered. Due to the lack of effective vehicle-induced turbulence (VIT) parameterization for neighborhood-scale large eddy simulation (LES), VIT was neglected.
Junyou Liu et al., (2022) [53]	Green spaces exacerbate PM2.5 pollution levels	Tree height, crown width	A road in Changsha, China	Tall trees with wide crowns can hinder PM2.5 dispersion and may lead to an increase in PM2.5 concentrations in street canyons.	No limitations were mentioned.
Junyou Liu et al., [55] 2023	Green spaces exacerbate PM2.5 pollution levels	The distance between trees	A road in Changsha, China	A certain distance between street trees can promote PM2.5 dispersion and is beneficial for reducing PM2.5 concentrations in street canyons.	The mechanism through which street trees influence PM2.5 concentrations cannot be directly gleaned from the simulation results.
Riccardo Buccolieri et al., (2018) [59]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	LAD, wind direction and speed	Marylebone Road street canyon, London, the UK	In a perpendicular wind environment, planting more roadside trees can exacerbate PM2.5 concentrations in street canyons. In a parallel wind environment, planting more roadside trees is beneficial for reducing the concentration of PM2.5 in street canyons. In general, the positive deposition effects are greater for increased LAD. In addition, perpendicular winds may counterbalance the negative aerodynamic effects near places close to trees.	The study is limited to the case study investigates to cases characterized by similar geometries subject to perpendicular and parallel winds. Research confirmed that multiple variables determine trees’ impact on urban air pollutant concentrations.
Huiyu He et al., (2023) [62]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	Tree height, trunk height crown width	A block near Huacheng Avenue and Huasui Road in Guangzhou, China	A 10-m-high roadside tree can help dilute PM2.5 in the vertical direction, while a tree with a high trunk is better able to remove traffic-related pollutants at pedestrian height and promote the diffusion of pollutants. An excessively wide tree crown can make it difficult for pollutants to escape from the neighborhood, thereby increasing pedestrian exposure levels.	The study only focuses on a certain source of pollution in a limited case area of hot and humid climatic region. The study lacks mathematical analysis illustrating the PM2.5 deposition through roadside trees and particulate diffusion along building clusters during different time period.
Xiaoyu Tian et al., (2024) [63]	Depending on the specific situation, green spaces either mitigate or exacerbate PM2.5 pollution levels	Street size, building density, building height, tree height, trunk height, crown diameter, road strip type, spacing between roadside trees	A block in Guangzhou, China	Roadside trees can absorb PM2.5, but they may also cause PM2.5 to accumulate at the street level, thereby affecting the diffusion of PM2.5 in densely populated buildings. By planting trees with an appropriate height, crown width, leaf area index, and spacing within a certain range, the adsorption capacity of roadside trees can be improved.	The valid sample for this study is limited. The model needs more experiments to identify accuracy.

4.2. The Effect of Green Spaces on PM_2.5 Within and Around Residential Areas

Residential areas are a major aspect of people’s daily lives. The extent of PM_2.5 pollution in these areas is also a major focus. Green space in residential areas can be regarded as a unique type (residential green space) and have a considerable influence on people’s daily lives. Because green space can reduce the total amount of PM_2.5 suspended in the air, increasing the amount of green space in residential areas is generally believed to be beneficial in terms of reducing the risk of death due to several diseases, such as acute pneumonia [64,65]. Wang et al. [66] used a small town—Tao Liu, in Zibo, China—as a case and focused on green space in residential areas. They found that green space structure can have significant effects on PM_2.5 concentrations. However, the spacing between vegetation and buildings had no significant effect on PM_2.5 concentration distribution. When trees are evenly distributed, their effect on reducing PM_2.5 concentration is minimal. However, a rich distribution of trees can play quite an active role in reducing PM_2.5 concentrations. Kim et al. [58] also pointed out that the residential areas located behind planting bands can have very low PM_2.5 concentrations. In their study, it was found that the slope with the highest amount of greenery in the spring had a better effect on reducing PM_2.5 levels, while the flat type with the highest amount of shrub greenery had a better effect in the winter. Zhang et al. [67] found that large residential green spaces are typical PM_2.5 sink landscapes. The PM_2.5 concentration inside residential green spaces is significantly lower than that outside green spaces, and the area and perimeter of residential green spaces have a significant effect on PM_2.5. However, according to the results of a regression analysis, as the circumference and area increase, the PM_2.5 removal rate shows a trend of first increasing and then decreasing.

However, green space in residential areas can also increase PM_2.5 concentrations in certain micro-environments. For example, Xiong and Chen [68] explored the effect of residential green space morphology on PM_2.5 via CFD simulation analysis. They pointed out that plants in residential areas are not always conducive to reducing PM_2.5, and an inappropriate green space layout may lead to an increase in PM_2.5 concentrations in specific residential areas. Densely vegetated urban roads and residential areas may have a strong negative blocking effect on PM_2.5 dispersion. Although green spaces can positively contribute to PM_2.5 removal through deposition, the net effect in certain cases could be an increase in local PM_2.5 concentrations if the obstruction effect outweighs the removal mechanisms.

4.3. The Effect of Green Spaces on PM_2.5 Within and Around Industrial Areas

Industrial activity is also a significant source of PM_2.5. Because studies focusing on green space and industrial sources of PM_2.5 have not limited their scope to industrial land use areas, this part mainly focuses on the effect of green space on industrial-sourced PM_2.5 within industrial areas or in areas adjacent to them. Han et al. [69] used a land use regression model to study PM_2.5 during the hot season in Xi’an, China. The study included a large number of monitoring points, buffer zones, and localized emission source variables (covering information on land use, road traffic source facilities, socio-economic conditions, emission sources, and geographic spatial data). A linear model was constructed through stepwise regression, and it was found that PM_2.5 during the hot season in Xi’an, the spatial distribution of PM_2.5 is closely related to the layout of industrial land, with industrial emissions contributing more than transportation. Green spaces, especially large-scale green spaces, can effectively reduce PM_2.5 concentrations. In addition, Bikis [70] found that the green spaces near roads and in industrial areas of Addis Ababa, Ethiopia, are lacking. The correlation between the occurrence of air-quality-related health problems and the air quality index at 12 transportation stations corresponded to a value of 0.9547 for R² in terms of average concentration. In addition, the correlation between the air quality index and air-quality-related health problems with respect to manufacturing industry employees was stronger (R² = 0.9728). These results confirmed that the manufacturing industry emits more pollutants than transportation. These factors can raise PM_2.5 concentrations in these areas. Increasing the amount of green space in industrial and road areas is an effective way of solving this problem. Li et al. [71] pointed out that the expansion of industrial land is a cause of PM_2.5 pollution. However, the increase in vegetation coverage in their study did not lead to a decrease in PM_2.5 concentrations. On the contrary, the authors found that either both increase over time or there is no significant negative correlation between the two.

4.4. The Effect of Green Spaces on PM_2.5 Within and Around School Campuses

School campuses are important places of study from kindergarten to university. A healthy school campus environment is quite important for the student population. Some studies have explored how green spaces within and around school campuses affect the PM_2.5 levels within the campuses. Campus green spaces can significantly reduce PM_2.5 pollution that spreads from car lanes to classrooms, playing an important role in providing a healthier learning environment for young people. Matthaios et al. [72] found that the degree of greenery within a 270-m buffer zone around a school is associated with indoor traffic, with the following PM_2.5 correlation: (−0.068, 95% confidence interval: −0.124, −0.013). For greenery further away from the school (within a 1230-m buffer zone), this association is stronger, with the following correlation regarding PM_2.5: (−0.101, 95% confidence interval: −0.156, −0.046). Wang et al. [73], in their field-monitoring research conducted around university campuses, found that the PM_2.5 concentration in lakes and green leisure areas on campus was significantly lower than the average level. Therefore, increasing the distribution of centralized green spaces on campus can provoke people to relax and provide a healthier environment in these areas. Jiang et al. [74] pointed out that air temperature is negatively correlated with PM_2.5 (r = −0.280, p < 0.01), while wind speed and relative humidity are positively correlated with PM_2.5 (relative humidity and PM_2.5: r = 0.476, p < 0.01; wind speed: r = 0.355, p < 0.01). Adjusting campus green spaces (plants communities, underlying surfaces, and the surrounding environment) can influence meteorological factors (temperature, humidity, wind speed, and illumination) and reduce PM_2.5 concentrations.

4.5. The Effect of Parks on PM_2.5

The abundant vegetation coverage in urban parks undoubtedly has a significant effect on PM_2.5. In addition, the parks themselves generate less pollution due to their distance from areas where PM_2.5 pollution is generated, such as roadways, factories, and residential areas. Many researchers have explored the effect of urban parks on PM_2.5. These efforts include the experimental research conducted by Junior et al. [75], who obtained PM_2.5-concentration-related information from PM_2.5 collectors at two locations. One was located in a botanical garden with high traffic flow in the surrounding area, and the others were located near tunnel entrances and exits with heavy traffic. Through box plot comparison, it was found that the PM_2.5 concentration at the observation points in the botanical garden was 33% lower, leading to the conclusion that botanical gardens have a significant effect on reducing PM_2.5 levels. By setting up sampling points inside and outside a park for comparison and using CFD to simulate the differences in PM_2.5 concentrations under multi-scenario conditions, Heshani et al. [76] found that the vegetation inside the park had a good barrier effect, effectively reducing the inflow of PM_2.5 from outside the park, thereby making the PM_2.5 concentration inside the park lower than outside. LAD has a significant effect on the removal efficiency of PM_2.5 at the height of human respiration; that is, the higher the LAD value, the greater the PM_2.5 removal efficiency. In addition, hedgerows with high LAD values are more effective in reducing PM_2.5 concentrations than trees with deeper crowns. Liu et al. [77] compared the difference between the vertical PM_2.5 concentration in a city center and a forest park by monitoring data collected using DustTrak II 8532 aerosol detector (TSI, Saint Paul, MN, USA). Their collection area included a 33-floor-high building (100 m) and a national forest park located in the same city. They collected PM_2.5 concentrations at heights of 1.5 m, 10 m, 25 m, 35 m, 50 m, 65 m, 80 m, and 100 m at the building collection point and the forest park. They found that in the forest park, the PM_2.5 concentration throughout the year showed an overall decreasing trend with an increase in height. At a height of 100 m, the average PM_2.5 concentration decreased to 66.8%, 61.9%, 74.1%, and 80.8% of the maximum value at about 1.5 m, respectively. However, in the city center, the peak concentrations were not found at ground level but at higher altitudes. The PM_2.5 concentration in the city center showed an upward trend from the ground to a height of about 25 m and then decreased with an increase in height. At a distance of 100 m, the proportions of the decrease during the four periods were only 7.0%, 12.9%, 18.1%, and 19.7% of the PM_2.5 concentration, respectively. The authors believe that the increasing trend regarding the vertical distribution at low altitudes (0 to 25 m) in the city center is related to roadside trees’ interception effect on PM_2.5. Roadside trees can reduce the vertical diffusion of PM_2.5, resulting in higher concentrations of PM_2.5 at low altitudes. Yan et al. [78] conducted an observation study on near-ground PM_2.5 in a forest park in Beijing using a grassland sampling point and a bare-ground control sampling point, with a distance of about 30 m between the two and sampling conducted from 7:00 to 18:00 on the first 4 days of each month from September 2014 to September 2015. It was found that the annual concentration of PM_2.5 in the forest is highest in the winter, followed by the spring and autumn, and it is lowest in the summer. The average concentration of PM_2.5 near the ground reached a peak from 07:00 to 09:30, and its lowest value occurred from 12:00 to 15:00. The concentration of PM_2.5 in forests is significantly positively correlated with relative humidity and significantly negatively correlated with temperature. In the forest examined, the highest removal efficiency in the summer occurs between 7:00 and 9:30, while the lowest removal efficiency in the winter occurs between 9:30 and 12:00. Based on observations of PM_2.5 concentrations made in a forest park in Guangzhou, China, Xiao et al. [79] found that the PM_2.5 concentration in the dry season was higher than that in the wet season. Industrial activities, transportation emissions, and soil dust are the main sources of metals in PM_2.5 in the forest park.

Different degrees of vegetation coverage in an urban park can have different effects on PM_2.5. In Su et al.’s [80] study, the rate of reduction in PM_2.5 by parks ranged from 2.51% to 35.57%. Through an analysis of heavy metals deposited in leaves in urban parks, the authors found that the particles in the parks were less toxic, indicating that the air in the parks was cleaner. In addition, parks with dense vegetation showed stable PM_2.5 emission reduction effects during the year monitored and exhibited stronger performance in providing cleaner air in severe PM_2.5 pollution situations. In contrast, parks with sparse vegetation had unstable emission reduction effects, leading to rapid dissipation or accumulation of pollutants within the park.

In urban parks, different vegetation configurations can have significant effects on PM_2.5. Yin et al. [81] used i-tree software to conduct a simulation and found significant differences in PM_2.5 interception among different tree species in parks in Beijing. Tree coverage, tree species richness, average tree height, average crown width, and number of trees have a significant effect on the PM_2.5 retention in green spaces, with average tree height having the greatest influence. The estimated degree of annual PM_2.5 retention in urban parks and green spaces within the Fifth Ring Road in Beijing is about 6380 tons. The heterogeneity of park plant communities and its effect on PM_2.5 have also been a focus for researchers. A study was conducted in forest park in Wuhan, with 44 sample plots selected. The authors found that the composition and location of plant communities are significant factors affecting PM_2.5 [82]. Wetlands in urban parks have also been identified as playing quite an active role in reducing PM_2.5 concentrations. They are highly efficient in removing PM through dry deposition and vegetation collection. Therefore, wetlands should be protected and constructed in future urban development projects. More regular wetland patches accelerate deposition processes. Long and dense wetland patches at the edges can lead to faster deposition. In contrast, fragmented landscapes are not conducive to dry deposition [82].

5. The Effect of the Structural Characteristics of Vegetation on PM_2.5

The effect of plants on PM_2.5 is extremely complex. Firstly, trees differ significantly in terms of their effects on PM_2.5 concentrations at different heights. Ji and Zhao [83] pointed out that trees (specifically cypress trees) around buildings can effectively reduce concentrations of PM_2.5. In their study, compared with a treeless environment, trees increased the PM_2.5 removal rate by about 20%. The effect varies depending on the height of the tree, and the bottom of the tree (including pedestrian height) has a more significant effect on reducing PM_2.5 than other heights. At maximum tree height (crown height, below building height), trees maintain their influence. Above the treetops, the influence of trees is almost negligible. Yang et al. [84] used a literature search and simple weighting methods to explore the capacity of 100 ordinary trees in 328 cities across 60 countries to reduce PM_2.5 levels. They pointed out that trees undoubtedly have a positive effect on reducing PM_2.5 levels, but some common trees are not the most effective in reducing PM_2.5. A mixed distribution of evergreen and deciduous trees is very effective in reducing PM_2.5. However, Shen et al. [85] pointed out that although the presence of leaves can increase the surface area for dry deposition, an excessively high LAD can hinder natural ventilation and the rate of exchange between air and the surrounding environment, leading to an increase in PM_2.5 concentrations in the microenvironment. Tree species with rough leaf microstructures and high PM_2.5 deposition rates constitute a better choice for improving air quality. Ciro et al. [86] found that even though an increase in tree coverage is beneficial for PM_2.5 removal, trees with lower LAIs are better at reducing PM_2.5 pollution.

There are significant differences in the effects of different tree species, and even different parts of the same tree, on PM_2.5 [87,88]. Chen et al. [89] focused on the differences in PM_2.5 capture efficiency among different plant leaves and pointed out that the needle-shaped leaves of coniferous trees make them more effective in pollutant interception and post-rain recovery than broad-leaved trees. The shapes and leaf veins of broad-leaved trees seem to have no effect on pollutant interception, but the number of grooves and trichomes on broad-leaved leaves is positively correlated with pollutant interception. Liang et al. [90] found a strong relationship between the proportion of grooves, stomatal size, and the ability of plants to capture PM_2.5 based on a study of 25 plant species in Beijing and Chongqing. There is no significant correlation between stomatal density, leaf hair, and PM_2.5 capture efficiency. The leaf morphology of broad-leaved tree species is more complex, and, compared to coniferous trees, they have a stronger ability to capture PM_2.5 per unit leaf area. However, due to the larger total leaf area of coniferous tree species, the PM_2.5 capture capacity of each tree may be stronger. However, Kim et al. [91] conducted a study on the PM_2.5 reduction ability of 13 major greening tree species and found that the PM_2.5 reduction of broad-leaved tree species was about 8.6 times that of coniferous tree species. Tree species with a larger leaf area were found to have greater PM_2.5 reduction ability, while trees with larger leaf mass were found to have higher PM_2.5 reduction capacity. Different tree species will also affect particles of different sizes, such as Cinnamomum camphora (L.) J. Presl and Osmanthus fragrans Lour., which are better at capturing particles smaller than 1 µm, while Ginkgo biloba L. and Platycodon grandiflorus tend to capture larger particles, up to a maximum of 1.6 µm. From the perspective of leaf wax composition, there are significant species differences with respect to the chemical composition of particles captured by leaves. The viscous hydrophobic wax in leaf wax is conducive to the adhesion of particles, while fatty acids and alkanes are conducive to the adhesion of particles rich in organic carbon and heavy metals. The physiological characteristics of leaves, such as stomatal conductance and water potential, significantly enhance the capture of larger particles [92]. Chaparro et al. [93] found that tree bark can also capture a certain amount of PM_2.5, especially iron-rich particles. Therefore, the influence of tree bark on PM_2.5 concentrations cannot be ignored.

Under different microclimate conditions, the effect of plants on PM_2.5 will also differ significantly. Xie et al. [94] pointed out that the plant morphological and structural characteristics that have the greatest effect on PM_2.5 vary under different wind speed conditions. Under three wind speed conditions of 1 m/s, 3.5 m/s, and 8 m/s, the surface area/weight density of branches and leaves as well as the overall characteristics of the branches are the most important factors affecting retention, and the efficiency is negative. At wind speeds of 1 m/second and 3.5 m/s, leaf size and surface roughness are the other two key positive factors, while at wind speeds of 8 m/s, the numerical density of branches and leaves is the key positive factor. As wind speed increases, the influence of leaf size and surface characteristics on retention decreases, while the characteristics of tree crown structure become more significant. Compared with leaf morphological characteristics, the variation in crown morphological structure variables among different tree species is greater, and it also has a greater effect on the retention effect of PM_2.5. As rainfall can affect plants’ ability to retain dust, there are significant differences in the dust retention ability of different plants after rain. Luo et al. [95] pointed out that after rainfall, different tree species’ adsorption capacity per unit leaf area decreases to varying degrees. On the second day after rainfall, the adsorption capacity of various tree species rapidly increases. The increase in adsorption capacity of broad-leaved tree species after rainfall is greater than that of coniferous tree species.

6. The Regulatory Effect of Plants on PM_2.5

PM_2.5 has a slow sedimentation rate due to its small size, and it can remain in the air for quite a long time and spread far and wide. Plants can influence PM_2.5 through deposition, blockage, adsorption, and absorption [14,15]. Table 6, Table 7, Table 8 and Table 9 reviewed the deposition, blockage, adsorption and absorption effects of plants on PM_2.5 respectively [96,97,98,99,100,101,102,103,104,105,106]. The deposition effect mainly refers to PM_2.5 falling on plants’ surfaces through aerodynamic effects such as gravity settling and turbulent diffusion (the process by which vegetation directly reduces PM in the atmosphere is called dry deposition). For instance, coniferous trees can have a higher deposition efficiency than broad-leaved trees due to leaf roughness and their high LAI. The blockage effect is achieved through the complex structures of branches and leaves, which obstruct the movement of PM_2.5. An excessively high vegetation density may impede air circulation, leading to the accumulation of PM_2.5 within street canyons. The adsorption effect mainly relies on the surface structures of plant leaves and special secretions to capture and hold PM_2.5. Absorption mainly refers to the ability of plant stomata to take in PM_2.5. [14,15].

Table 6, Table 7, Table 8 and Table 9 reveals that vegetation can have obvious deposition, blockage, adsorption, absorption effects on PM_2.5 concentrations. These play an important role in the mitigation of PM_2.5 pollution. Different leaf surfaces can adsorb certain amounts of PM_2.5 and help decrease PM_2.5 concentrations. Some researchers [105,106] who have explored the absorption effect of vegetation even believe it is as weak when compared with plants’ deposition and adsorption capacities. These properties positively contribute to decreasing PM_2.5 concentrations. However, vegetation can block the dispersion of PM_2.5. This may increase PM_2.5 concentrations in the micro-environment. Due to the complex mechanism whereby vegetation effects PM_2.5 concentrations, the net effect is highly dependent on the specific situation. More studies can be conducted to explore ways of increasing the positive effect and decreasing the negative effect and determine how to more effectively reduce PM_2.5 concentrations through vegetation.

Table 6. Studies focusing on the deposition effects of trees.

The Regulatory Mechanism Studied	Authors and Year	Method	Main Findings	Limitations
Deposition	Jeanjean et al., (2016) [96]	CFD model and i-tree model	Tree aerodynamics have a strong effect on PM2.5, reducing PM2.5 levels by 9% during the summer in Leicester, while the deposition effects of trees and grasslands are relatively weak, only reducing PM2.5 levels by 2.8% and 0.6%, respectively.	No dynamic factors: Wind, heat, and chemical effects were ignored. Low accuracy: The accuracy of the CFD model was only 30–40%.
	Yu Zhang et al., (2021) [97]	Experimental measurement	PM2.5 deposition flux in forests increases with an increase in air pollution levels. When the wind speed is high, atmospheric turbulence is strong, air resistance decreases, and the deposition velocity of PM2.5 increases accordingly. Meteorological factors affect the deposition of PM2.5 through wind and water wash-off.	Limited near-surface studies: Research on near-surface PM2.5, which is closely linked to human activity, remains insufficient. Statistical insignificance: Although forest deposition fluxes are generally higher than those of wetlands, the differences are not statistically significant.
	Mattias Gaglio et al., (2022) [98]	Experimental measurement and i-Tree Eco model	Leaf characteristics play an important role in the dry deposition of PM2.5 via vegetation. Tree species with sticky substances such as wax or resin can accumulate more PM2.5 than other tree species, but, at the same time, they are not conducive to thorough cleaning of leaves, thereby reducing the amount of PM2.5 removed.	Leaf age variation: Older leaves accumulate more PM2.5, affecting accuracy. Seasonal dynamics were ignored: Leaf trait changes in spring/autumn were not modeled. Incomplete leaf washing: PM2.5 may remain on waxy surfaces. Soluble PM2.5 loss: Some particles may dissolve and escape detection. Needle-area uncertainty: The complex shape of the needle hinders precise measurement. Subjective trait scoring: Some leaf traits lack objective quantification. Low sampling frequency: There are insufficient data for assessing year-round dynamics. Pollution source variation: Differences across sampling sites may skew results.
	Jiaqi Yao et al., (2023) [99]	Mathematical model	Broad-leaf forests are the main source of dry deposition in urban spaces (accounting for 89.22% in the study). For PM2.5, the explanatory power of each driving factor is ranked as follows: monthly cumulative precipitation > monthly average surface temperature > monthly average wind speed > monthly maximum comprehensive normalized vegetation index.	The estimation model simplified the calculation of the resuspension rate and did not account for precipitation frequency, potentially affecting annual deposition accuracy. The lack of sufficient field observations limits the precision and validity of the model results. Multivariable analysis shows limited explanatory power for dry deposition variations, with some underlying mechanisms remaining unclear. The national-scale model may not be directly applicable to local-level analysis, and its generalizability requires further testing.
	Jiansheng Wu et al., (2015) [100]	CFD simulation	Wind speed is a major factor affecting PM2.5 concentrations. The ranking of the dry deposition rate and PM2.5 removal capacity by tree species is as follows: Ficus macrocarpa > Ficus altissima > Codiaeum variegatum > Fagraea ceilanica. The number of grooves on the leaf surface and the size of stomata are significantly positively correlated with the ability to remove PM2.5. Among the three green belt configurations (trees, shrubs, and a combination of trees and shrubs), the shrub-dominated green belt exhibited PM2.5 concentrations at pedestrian breathing height that were 15–20% lower than those observed in the other configurations.	The study only focussed on one specific season. This study only focussed on one specific wind direction. The study overlooked the resuspension process. The study only considered traffic source without the consideration of background PM2.5 concentration.
	Xuyi Zhang et al., (2020) [101]	Experimental measurement and statistical model	The dry deposition velocity of coniferous trees is higher than that of broad-leaved trees. Leaf surface free energy and specific leaf area have a significant effect on the deposition velocity of coniferous plants. Broadleaved trees have a more flexible leaf structure, leading to erratic flutter, which can hinder deposition and cause PM2.5 to resuspend.	Partial explanation: Leaf traits explain only 72.77% of Vd variation. Trait gaps: Wax structure/composition was not analyzed. Single particle type: Only nonpolar tracers were used. Limited samples: Examination of only a few conifer species reduced representativeness.

Table 6. Studies focusing on the deposition effects of trees.

The Regulatory Mechanism Studied	Authors and Year	Method	Main Findings	Limitations
Deposition	Jeanjean et al., (2016) [96]	CFD model and i-tree model	Tree aerodynamics have a strong effect on PM2.5, reducing PM2.5 levels by 9% during the summer in Leicester, while the deposition effects of trees and grasslands are relatively weak, only reducing PM2.5 levels by 2.8% and 0.6%, respectively.	No dynamic factors: Wind, heat, and chemical effects were ignored. Low accuracy: The accuracy of the CFD model was only 30–40%.
	Yu Zhang et al., (2021) [97]	Experimental measurement	PM2.5 deposition flux in forests increases with an increase in air pollution levels. When the wind speed is high, atmospheric turbulence is strong, air resistance decreases, and the deposition velocity of PM2.5 increases accordingly. Meteorological factors affect the deposition of PM2.5 through wind and water wash-off.	Limited near-surface studies: Research on near-surface PM2.5, which is closely linked to human activity, remains insufficient. Statistical insignificance: Although forest deposition fluxes are generally higher than those of wetlands, the differences are not statistically significant.
	Mattias Gaglio et al., (2022) [98]	Experimental measurement and i-Tree Eco model	Leaf characteristics play an important role in the dry deposition of PM2.5 via vegetation. Tree species with sticky substances such as wax or resin can accumulate more PM2.5 than other tree species, but, at the same time, they are not conducive to thorough cleaning of leaves, thereby reducing the amount of PM2.5 removed.	Leaf age variation: Older leaves accumulate more PM2.5, affecting accuracy. Seasonal dynamics were ignored: Leaf trait changes in spring/autumn were not modeled. Incomplete leaf washing: PM2.5 may remain on waxy surfaces. Soluble PM2.5 loss: Some particles may dissolve and escape detection. Needle-area uncertainty: The complex shape of the needle hinders precise measurement. Subjective trait scoring: Some leaf traits lack objective quantification. Low sampling frequency: There are insufficient data for assessing year-round dynamics. Pollution source variation: Differences across sampling sites may skew results.
	Jiaqi Yao et al., (2023) [99]	Mathematical model	Broad-leaf forests are the main source of dry deposition in urban spaces (accounting for 89.22% in the study). For PM2.5, the explanatory power of each driving factor is ranked as follows: monthly cumulative precipitation > monthly average surface temperature > monthly average wind speed > monthly maximum comprehensive normalized vegetation index.	The estimation model simplified the calculation of the resuspension rate and did not account for precipitation frequency, potentially affecting annual deposition accuracy. The lack of sufficient field observations limits the precision and validity of the model results. Multivariable analysis shows limited explanatory power for dry deposition variations, with some underlying mechanisms remaining unclear. The national-scale model may not be directly applicable to local-level analysis, and its generalizability requires further testing.
	Jiansheng Wu et al., (2015) [100]	CFD simulation	Wind speed is a major factor affecting PM2.5 concentrations. The ranking of the dry deposition rate and PM2.5 removal capacity by tree species is as follows: Ficus macrocarpa > Ficus altissima > Codiaeum variegatum > Fagraea ceilanica. The number of grooves on the leaf surface and the size of stomata are significantly positively correlated with the ability to remove PM2.5. Among the three green belt configurations (trees, shrubs, and a combination of trees and shrubs), the shrub-dominated green belt exhibited PM2.5 concentrations at pedestrian breathing height that were 15–20% lower than those observed in the other configurations.	The study only focussed on one specific season. This study only focussed on one specific wind direction. The study overlooked the resuspension process. The study only considered traffic source without the consideration of background PM2.5 concentration.
	Xuyi Zhang et al., (2020) [101]	Experimental measurement and statistical model	The dry deposition velocity of coniferous trees is higher than that of broad-leaved trees. Leaf surface free energy and specific leaf area have a significant effect on the deposition velocity of coniferous plants. Broadleaved trees have a more flexible leaf structure, leading to erratic flutter, which can hinder deposition and cause PM2.5 to resuspend.	Partial explanation: Leaf traits explain only 72.77% of Vd variation. Trait gaps: Wax structure/composition was not analyzed. Single particle type: Only nonpolar tracers were used. Limited samples: Examination of only a few conifer species reduced representativeness.

Table 7. Studies mainly focusing on trees’ blockage effect.

The Regulatory Mechanism Studied	Authors and Year	Method	Main Findings	Limitations
Blockage	Zheming Tong et al., 2016 [102]	CFD model	Roadside vegetation barriers have the potential to reduce near-road PM2.5 concentrations. Increasing the LAD and the width of and vegetation barriers can significantly lower the PM2.5 concentration behind the barrier compared to the baseline situation. However, increasing the width of vegetation barriers may also increase roadside PM2.5 concentrations due to weakened roadside diffusion. The presence of solid obstacles on the roadside will cause the incoming airflow to deflect upwards and form a backflow zone behind the obstacles, which will increase the concentration of PM2.5 on the road, but the concentration on both sides of the obstacles will decrease. The cumulative effect of vegetation covering solid obstacles on PM2.5 is minimal. The boundary layer formed on the surface of obstacles such as vegetation may inhibit airflow through the vegetation.	The model did not consider factors such as structure and terrain, which may affect near-field diffusion. The research results are based on coniferous evergreen tree species and may not be applicable to broad-leaved trees or shrub plants. Parameters such as LAD distribution, dry deposition, resistance coefficient, and meteorological conditions induce uncertainty.
	Fan Li et al., (2022) [103]	CFD mode ls	An unfavorable layout of buildings and trees can hinder wind flow, resulting in excessive PM2.5 pollution. The aerodynamic effect of trees on the windward side of buildings tends to reduce wind speed, hindering the diffusion of PM2.5. On the contrary, trees on the leeward side of buildings tend to enhance ventilation, thereby reducing PM2.5 accumulation. Low PM2.5 concentrations are usually found in areas with high airflow velocity, and PM2.5 substances are mainly concentrated in the windward areas of downstream buildings and densely populated areas with trees.	The study has very limited consideration about dipositive and absorptive effects. The current study simplified the wind characteristics as a constant value The study has limited consideration about wind direction.

Table 7. Studies mainly focusing on trees’ blockage effect.

The Regulatory Mechanism Studied	Authors and Year	Method	Main Findings	Limitations
Blockage	Zheming Tong et al., 2016 [102]	CFD model	Roadside vegetation barriers have the potential to reduce near-road PM2.5 concentrations. Increasing the LAD and the width of and vegetation barriers can significantly lower the PM2.5 concentration behind the barrier compared to the baseline situation. However, increasing the width of vegetation barriers may also increase roadside PM2.5 concentrations due to weakened roadside diffusion. The presence of solid obstacles on the roadside will cause the incoming airflow to deflect upwards and form a backflow zone behind the obstacles, which will increase the concentration of PM2.5 on the road, but the concentration on both sides of the obstacles will decrease. The cumulative effect of vegetation covering solid obstacles on PM2.5 is minimal. The boundary layer formed on the surface of obstacles such as vegetation may inhibit airflow through the vegetation.	The model did not consider factors such as structure and terrain, which may affect near-field diffusion. The research results are based on coniferous evergreen tree species and may not be applicable to broad-leaved trees or shrub plants. Parameters such as LAD distribution, dry deposition, resistance coefficient, and meteorological conditions induce uncertainty.
	Fan Li et al., (2022) [103]	CFD mode ls	An unfavorable layout of buildings and trees can hinder wind flow, resulting in excessive PM2.5 pollution. The aerodynamic effect of trees on the windward side of buildings tends to reduce wind speed, hindering the diffusion of PM2.5. On the contrary, trees on the leeward side of buildings tend to enhance ventilation, thereby reducing PM2.5 accumulation. Low PM2.5 concentrations are usually found in areas with high airflow velocity, and PM2.5 substances are mainly concentrated in the windward areas of downstream buildings and densely populated areas with trees.	The study has very limited consideration about dipositive and absorptive effects. The current study simplified the wind characteristics as a constant value The study has limited consideration about wind direction.

Table 8. Study mainly focusing on trees’ adsorption effect.

The Regulatory Mechanism Studied	Authors and Year	Method	Main Findings	Limitations
Adsorption	Xinxin Zhao et al., (2019) [104]	Experimental measurement	Wrinkled leaves and leaves covered with fine hairs have superior PM2.5 adsorption capacity, as these leaves have rough surfaces with many protrusions and depressions. Tree species with smooth leaves, low stomatal density, and small stomatal openings are less capable of adsorbing PM2.5.	No limitations were mentioned.
	Kunhyo Kim et al., (2022) [91]	Experimental measurement	The adsorption capacity of plant leaf surfaces is positively correlated with pollution levels. In different polluted areas, the adsorption capacity per unit leaf area is highest in the winter, followed by the spring, autumn, and, finally, summer. The PM2.5 adsorption capacity per unit leaf area of coniferous trees is greater than that of broad-leaved trees.	The model is idealized and does not account for real-world urban complexity. Only neutral atmospheric conditions are considered; thermal effects are ignored. Wind and turbulence settings are limited, reducing applicability. The particle size range is narrow, possibly leading to the underestimation of deposition.

Table 8. Study mainly focusing on trees’ adsorption effect.

The Regulatory Mechanism Studied	Authors and Year	Method	Main Findings	Limitations
Adsorption	Xinxin Zhao et al., (2019) [104]	Experimental measurement	Wrinkled leaves and leaves covered with fine hairs have superior PM2.5 adsorption capacity, as these leaves have rough surfaces with many protrusions and depressions. Tree species with smooth leaves, low stomatal density, and small stomatal openings are less capable of adsorbing PM2.5.	No limitations were mentioned.
	Kunhyo Kim et al., (2022) [91]	Experimental measurement	The adsorption capacity of plant leaf surfaces is positively correlated with pollution levels. In different polluted areas, the adsorption capacity per unit leaf area is highest in the winter, followed by the spring, autumn, and, finally, summer. The PM2.5 adsorption capacity per unit leaf area of coniferous trees is greater than that of broad-leaved trees.	The model is idealized and does not account for real-world urban complexity. Only neutral atmospheric conditions are considered; thermal effects are ignored. Wind and turbulence settings are limited, reducing applicability. The particle size range is narrow, possibly leading to the underestimation of deposition.

Table 9. Studies mainly focusing on trees’ absorption effect.

The Regulatory Mechanism Studied	Authors and Year	Method	Main Findings	Limitations
Absorption	Qi Yang et al., (2018) [105]	Experimental measurement	The SO42- absorbed by plants can be assimilated into sulfur-containing products and stored in cellular vacuoles or be transported between different tissues and organs. PM2.5 pollution leads to stomatal closure in poplar leaves, and exposure to high concentrations of PM2.5 causes more severe stomatal closure than exposure to low concentrations of PM2.5.	No limitations were mentioned.
	Yifan Li et al., (2019) [106]	Experimental measurement	As the concentration of PM2.5 increases, the net photosynthetic rate and stomatal conductance decrease over time, and the higher the concentration of PM2.5, the faster the decrease. The higher proportion of grooves in plants and the presence of hairy bodies on the surfaces of leaves may buffer the effect of PM2.5 on stomata, slowing down the rate of changes in stomatal conductance.	No limitations were mentioned.

Table 9. Studies mainly focusing on trees’ absorption effect.

The Regulatory Mechanism Studied	Authors and Year	Method	Main Findings	Limitations
Absorption	Qi Yang et al., (2018) [105]	Experimental measurement	The SO42- absorbed by plants can be assimilated into sulfur-containing products and stored in cellular vacuoles or be transported between different tissues and organs. PM2.5 pollution leads to stomatal closure in poplar leaves, and exposure to high concentrations of PM2.5 causes more severe stomatal closure than exposure to low concentrations of PM2.5.	No limitations were mentioned.
	Yifan Li et al., (2019) [106]	Experimental measurement	As the concentration of PM2.5 increases, the net photosynthetic rate and stomatal conductance decrease over time, and the higher the concentration of PM2.5, the faster the decrease. The higher proportion of grooves in plants and the presence of hairy bodies on the surfaces of leaves may buffer the effect of PM2.5 on stomata, slowing down the rate of changes in stomatal conductance.	No limitations were mentioned.

7. Greening Interacts with Other Environmental Factors to Mitigate PM_2.5

Alongside the abovementioned deposition, blockage, adsorption, and absorption effects, green spaces can also change microclimates through cooling and humidification, and also affect the wind environment, thereby affecting PM_2.5 concentrations [107]. Feng et al. [107] found significant differences in the correlation between temperature, relative humidity, and PM_2.5 concentrations in different seasons. The correlation between temperature and PM_2.5 concentrations varies throughout the four seasons, with a positive correlation in the spring (r = 0.163, p < 0.01) and a negative correlation in the summer (r = −0.367, p < 0.01), autumn, and winter, with the strongest correlation being in the summer, with the rest having overall correlation coefficients < 0.4, indicating a weak correlation. In the summer and winter, PM_2.5 is significantly positively correlated with relative humidity, with an r = 0.475 (p < 0.01) in the summer and an r = 0.368 (p < 0.01) in the winter. The correlation is weak in the spring and autumn (r < 0.2). In addition, plants can not only significantly affect the dry deposition process but also influence the wet deposition process on rainy days. Xie et al. [108] found that the retention of PM_2.5 by trees during rainfall is the result of a combination of multiple factors, with the key factor being the characteristics of the tree species. During rainfall, broad-leaved trees may have a higher efficiency regarding rainwater wash-off than coniferous trees. For broad-leaved trees, the wash-off process usually has two peaks, while coniferous trees have only one peak or no peaks. This means that the particle wash-off process for coniferous trees may be more stable than that for broad-leaved trees. The effect of greenery in different seasons on PM_2.5 also shows significant differences. In the study by Hong et al. [109], the average reduction efficiency regarding PM_2.5 within the forest was approximately 18.2%; the seasonal average reduction efficiency regarding PM_2.5 was highest in the winter (about 28.1%) and lowest in the summer (about 9.6%). Aside from the influence of the seasons, Shao et al. [110] found that there were significant differences in the effect of green spaces on PM_2.5 at different times of the day. By comparing the PM_2.5 concentration differences between observation points 45 m away from the road, it was found that although the afternoon value was positive, the reduction rate on a spring morning was negative, especially in March, before 9:30 (local time), when the reduction rate was the lowest. In the winter, the decrease rate is always negative, but it is significantly higher in the afternoon than in the morning. In addition, the authors of [110] pointed out that when the leaves are fully mature, the PM_2.5 concentration is strongly positively correlated with tree crown density and negatively correlated with porosity. There is no significant correlation between PM_2.5 concentration and tree crown density and porosity during the leaf-shedding and leaf-spreading periods.

Determining how to optimize an urban green space pattern to maximize its value is an urban planning issue. Lee et al. (2015) [111] pointed out that urban planners need to match the health benefits pursued with the needs of the community and the functions that urban green spaces will provide. In the process of planning green spaces, the community’s needs extend far beyond just reducing PM_2.5 pollution; they include factors like promoting physical exercise and mental health. The functions of green spaces are also influenced by microclimate conditions and other built-environment factors. This also indicates that the effect of green spaces on PM_2.5 cannot be viewed in isolation, and the interaction between greenery and other environmental factors should also be considered. In Chi et al.’s [112] study, livability was interpreted from six aspects: environment, health, accessibility, housing, economy, and education. These aspects complement each other. Chen et al. [113] pointed out that the health promotion value of urban green spaces can be extended to brand elements, with a focus on the quality and landscape aesthetic functions of urban green spaces. Guo et al. [114] pointed out that urban spatial structure, land use, spatial form, transportation, and green spaces are all related to exposure to PM_2.5 pollution. Reasonably arranging green spaces in the downwind direction of urban pollution sources may play an important role in reducing pollution exposure. It would be prudent to reduce plot ratios and building density in city centers. Improving the spatial continuity of green spaces in urban centers may have a significant effect on reducing PM_2.5. Eisenman et al. [115] reviewed relevant research and pointed out that there is currently no scientific consensus that urban trees can reduce asthma by improving air quality. In some cases, urban trees may reduce air quality and increase the incidence of asthma. Therefore, an interdisciplinary and comprehensive understanding of the effect of green spaces on PM_2.5 and human health is also crucial.

8. Discussion

The phenomenon of green spaces potentially reducing PM_2.5 pollution and exacerbating PM_2.5 in local environments is increasingly being studied by scholars, but there is a lack of comprehensive reviews on this phenomenon. This article represents an attempt to integrate the research conducted by scholars from various disciplines, such as urban and rural planning, landscape architecture, geography, forestry, environmental science, and atmospheric science, in order to gain a more comprehensive understanding of the dual effects of green spaces on PM_2.5 and promote better research in the future.

Through relevant reviews conducted on large scales, such as across regional or urban areas, it has been found that green spaces can reduce a significant amount of PM_2.5 pollution every year, and this is believed to be beneficial for human health [116,117]. From a macro perspective, afforestation within regions or cities is beneficial for reducing PM_2.5 concentrations and creating a healthier environment. In addition, the selection of different research units during a study may also lead to certain differences in research results. Most studies conducted at the macro scale has found that green spaces are beneficial for reducing PM_2.5 levels. However, in some studies, specific areas such as waterfront green coverage and areas with high green coverage have also been found to have relatively high PM_2.5 concentrations. There are some differences in how different studies optimize the spatial patterns of green spaces, such as patch size, shape characteristics, and the distance between patches, to maximize their effectiveness in reducing PM_2.5 concentrations [117,118,119]. This may be due to differences in the sizes of the research units selected and differences in where (in which cities) and when the studies were conducted.

From the perspective of different types of green spaces, there are significant differences in the findings related to road greening in road areas. Some researchers have found that road greening generally has a positive effect on PM_2.5 levels, while others have focused on negative effects. There are also some studies revealing that there may be both positive and negative effects under different conditions. The daily traffic-sourced PM_2.5 generated in road areas accounts for a large share, and many roads are urban air ducts that are greatly affected by other sources. Some researchers have used CFD or wind tunnels to simulate the ideal conditions of traffic-only sources, while others have not completely ruled out other sources of PM_2.5 in road areas as a reason for the differences. Studies are conducted in different regions, and there are significant differences in the characteristics of roads and street buildings. For example, in some cities, street buildings are mainly low-rise and multi-story, with narrow roads, while in some cities, street buildings are mainly high-rise or even super-high-rise, with wide roads. In addition, there are significant differences in the greening conditions of different roads, including with respect to tree species, green coverage, and tree height characteristics. Research on residential areas and surrounding green spaces has also yielded similar findings. Some studies have found that greening mainly has a positive effect on reducing PM_2.5 concentrations, while others have found a negative effect. The research conducted around urban industrial zones mostly focuses on the effect of green conditions in and around the entire industrial zone on PM_2.5. Currently, there is a lack of research on the spatial distribution of PM_2.5 at the meso and micro scales in industrial areas, so the relevant findings are relatively simple. The main advantage is that greening is conducive to reducing complex urban PM_2.5 pollution, including industrial-source PM_2.5. Research on campus greening and parks has found that greening is beneficial for reducing PM_2.5 pollution. These areas themselves produce very little or even no PM_2.5 pollution. Large-scale greening can not only block some foreign PM_2.5 outside an area but also fully exploit large green spaces’ ability to settle, block, adsorb, and absorb PM_2.5, thus reducing PM_2.5 levels [120,121,122]. Therefore, the optimization of greening at the meso and micro scales requires refined management tailored to local conditions. Overall, measures such as employing sparse sources and strong sinks, as well as promoting ventilation, can be considered. In areas producing high levels of PM_2.5, excessive planting should be avoided. Ventilation and other methods can be used to promote the flow of PM_2.5 into areas with large green coverage and low PM_2.5 production.

From the perspective of the mechanisms of action, many researchers have focused on the process whereby plants reduce PM_2.5 concentrations through dry deposition. They have found that this process has a significant reducing effect. The notion that green spaces can lead to an increase in local PM_2.5 pollution is more closely related to research focusing on the blockage effect of plants on PM_2.5. Plants’ adsorption of PM_2.5 and their effectiveness in reducing its levels in other ways are significantly influenced by the structural characteristics of the plants themselves, such as leaf surface roughness. There is relatively little research on absorption effects, with a focus on the components of PM_2.5 that are small and can be absorbed by stomas. Different tree species have significant differences in their ability to reduce PM_2.5 due to differences in leaf surface roughness, leaf size, leaf area index, and other characteristics. Therefore, selecting tree species of green spaces to reduce PM_2.5 is also an important consideration in optimizing the structure.

There are significant differences in the effect of green spaces on PM_2.5 under different environmental conditions, such as seasons, meteorological conditions (temperature, humidity, rainfall, etc.), and land-use characteristics. The numerous economic, social, and environmental benefits of green spaces are closely related. This also indicates that deep exploration of the connection between green spaces and their surrounding environment is necessary to maximize their value.

9. Conclusions

From the macro-regional and city scales, green spaces are mostly negatively correlated with PM_2.5 concentrations, but they may not have a significant correlation with PM_2.5 in some periods and locations and at certain research scales. At the aforementioned scales, a high green space ratio, large green space patches, and increased concentrated green space distributions are more likely to have a better effect on reducing PM_2.5 levels. However, this does not hold true for all times, places, and scales. For example, in one study [32], the AI of green space was found to be negatively correlated with PM_2.5 across most buffer zone scales, but there was no significant correlation at the 1000 m buffer zone scale. From the perspective of other indicators, such as urban green space shape complexity, different cities may exhibit significantly different regularity due to differences in factors like urban spatial morphology, pollution source distribution characteristics, and climate conditions, as shown by one study using LSI which found predominantly positive correlations with PM_2.5 concentration [32]. Moreover, this was shown by another study [39], which found predominantly negative correlations between SHAPE_AM and PM_2.5 concentration. In addition, there is a scale effect, with an optimal scale optimizing the macro-scale green space pattern for reducing PM_2.5 levels. For example, Bi et al. [38] pointed out that 2 km × 2 km is the optimal scale for analyzing the spatial pattern of green space morphology and the spatial heterogeneity of PM_2.5. Therefore, from the perspective of regional and urban development, increasing the green space ratio is conducive to reducing urban PM_2.5 pollution, especially when increasing large patches of green space in cities such as parks and squares. Promoting a more concentrated distribution of green spaces can lead to a better reduction in PM_2.5 pollution in most situations relative to a dispersed distribution.

Due to their proximity to pollution sources such as transportation and daily life, urban road areas and residential areas have a particularly pronounced dual effect (through greening) on PM_2.5, which can lead to an increase or decrease in local PM_2.5 concentrations. These green spaces undoubtedly play a certain role in reducing PM_2.5 pollution, but in narrow environments, excessively wide tree crowns or dense leaves may lead to an accumulation of PM_2.5 [61,62]. Areas with dense greenery and low PM_2.5 production, such as parks and campuses, are widely regarded as PM_2.5 sinks, and green spaces are widely believed to be beneficial for reducing PM_2.5 pollution. Therefore, we suggest that practitioners in urban planning and design, as well as landscape architecture, avoid planting too many dense plants in narrow areas close to pollution sources. However, promoting ventilation may facilitate the flow of PM_2.5 into high-concentration areas such as campuses and parks, thereby efficiently reducing PM_2.5 pollution.

The selection of plants is an important factor to consider in the rational planning and greening process. Overall, the rougher the plant leave surface, the more favorable it is for PM_2.5 adsorption [91,104]. A dense canopy with a large number of leaves provides a surface allowing for more dry deposition to occur. However, it may also lead to the accumulation of a large amount of PM_2.5 in the local environment due to strong blocking effects and reduced wind speed [97,98,99]. Therefore, the selection of plants needs to be tailored to local conditions, and there may not be an absolutely good solution.

There are significant differences in the effect of greening on PM_2.5 under different urban spatial characteristics, climate, pollution source locations, and intensities. For example, there are significant differences in the adsorption capacity of plants for PM_2.5 before and after rainfall [95,108]. The optimization of greening to reduce PM_2.5 pollution cannot be viewed in isolation with respect to greening and PM_2.5 pollution. It also needs to be combined with the environment in order to better understand the complexity of the mechanism of greening on PM_2.5 and effectively use greening to reduce PM_2.5 pollution.