Urban-Scale Quantification of Rainfall Interception Drivers in Tree Communities: Implications for Sponge City Planning

Abstract

Urban trees play a crucial role in regulating hydrological processes within urban ecosystems by intercepting rainfall to effectively reduce surface runoff and mitigate urban flooding. Current research lacks a systematic quantification of rainfall interception capacity and its community-level impacts at the urban scale. This study adopts a city-scale perspective, integrating field survey data with the i-Tree Eco model to systematically explore the contributions of 20 factors to the average annual rainfall interception of tree species and the average annual rainfall interception efficiency of communities. The study revealed that Deciduous broadleaf trees (1.28 m³ year⁻¹) and Pure coniferous forests (90.7 mm year⁻¹) exhibited substantial rainfall interception capacity. Relative Height, Average Tree Height, Average Crown Width, and Planting Density of trees significantly influence interception capacity. Urban planning can optimize the selection of tree species (e.g., Paulownia, Populus tomentosa, etc.) and community structure (e.g., mixed planting of conifers and deciduous broadleaf trees) to improve rainfall interception capacity, thereby effectively reducing stormwater runoff, mitigating the risk of urban flooding. These findings provide a scientific basis for designing urban vegetation to mitigate flooding, support water management, and advance sponge city development.

1. Introduction

Urbanization is progressing at an unprecedented pace [1]. The United Nations predicts that 62% (5.4 billion) of the global population will live in urban areas by 2036 [2]. Urbanization driven land use and cover changes have reduced flood regulation capacity by 13.39% in Chinese cities [3], thereby increasing flood risks and exacerbating water scarcity. Urban rainstorms and floods cause average annual direct losses of US$45.83 billion [4], while water ecosystem issues increasingly impact urban environments and residents’ lives [5,6]. Urban trees, as a crucial green resource, maintain the sustainable development of cities through a range of ecosystem services. Surface runoff is diminished by tree canopy via rainfall interception and transpiration [7]. The hydrological function of urban trees in stormwater management is a critical hydrological process that influences rainfall redistribution, nutrient cycling, as well as soil and water conservation in urban ecosystems [8,9]. Studies show that the rainfall interception rate of trees ranges from 36.7% in Ningxia to 29.45% in Inner Mongolia [10]. The role of urban trees in intercepting rainfall has become a crucial component of urban water management [11]. Therefore, investigating the rainfall interception capacity of urban trees is essential for developing sponge city ecosystems that can withstand the pressure of urbanization.

The traditional approach for studying canopy interception primarily utilizes in situ observation facilities to measure precipitation, throughfall, and runoff in forests [12]. These studies often focus on natural forests, specific tree species on plantation or small-scale accurate data [12,13,14]. During a single rainfall event, rainfall intercepted by urban tree canopies ranges from approximately 0.6–24.3 mm, with interception rates of the total rainfall ranging from 1.2 to 76.5% [15,16,17]. In natural forests, broadleaf forests intercepted approximately 18–29% of total precipitation, while coniferous forests intercepted 18–45% [18]. Coniferous forests intercepted larger raindrops than broadleaf forests and typically exhibit a higher rainfall interception capacity [19,20,21]. The four common deciduous tree species in Hamilton, Canada, can intercept and evaporate approximately 6.5% to 27% of the total rainfall during their leaf expansion period [22]. Field measurements and event-scale models, such as the Gash model and SCS-CN model, provide higher accuracy in capturing hydrological processes, but they do not account for interannual variations in the rainfall interception capacity of urban trees [23,24]. Moreover, scaling up field measurements to the spatial scale of the city remains challenging. In comparison, modeling canopy rainfall by integrating measurement data with parameter fitting enables extrapolation of field observations to different spatial and temporal scales [13]. The i-Tree Eco model, developed by the US Forest Service as a professional ecological function assessment model [25,26], is able to assess the structure of urban trees and quantify the eco-efficiency of trees [25,27], in which the water regulation of trees includes rainfall interception, surface runoff, evaporation and so on, which provides a realistic representation of how forest structure and underlying surface properties influence interception and runoff reduction [28]. The model integrates allometric growth equations for species and local meteorological data. Due to its operability and high degree of data standardization, it has become an important tool for urban green infrastructure research, and is now widely used in countries and regions such as the United States, Japan, and Europe. In Kyoto and New York City, researchers used the i-Tree model to estimate that street tree canopies reduce annual stormwater runoff by approximately 1699 m³ and 3.37 million m³, respectively [29,30]. Although studies have demonstrated the capability of i-Tree Eco in assessing rainfall interception by trees, there is a relative lack of research on the differences in rainfall interception and its drivers across community structures (e.g., pure forests, mixed forests) at the urban scale and at the inter-annual scale. There is an urgent need to use the model to quantify the interannual rainfall interception capacity of urban tree communities, which can make up for the lack of research in urban-scale spaces.

The rainfall interception capacity of tree canopies is influenced by rainfall characteristics (such as intensity, duration, and droplet size) [31,32], tree-specific traits (including canopy structure, leaf area index, and bark texture) [33,34], and prevailing climate conditions [35]. Previous studies have focused on quantifying the canopy interception effects of tree characteristics and rainfall events on canopy interception in forest ecosystems. Studies have shown that the interaction between forest litter mass and rainfall characteristics exerts the strongest influence on maximum interception storage capacity [36]. Rainfall amount is positively correlated with canopy rainfall interception in a stepwise manner [37]. Moreover, leaves with greater hydrophobicity capture more precipitation than leaves that are more hydrophilic [38]. These studies provide valuable insights into the mechanisms of canopy rainfall interception in forests. However, due to differences in growth space, canopy structure, soil, and microclimate between urban trees and forest plants [39], the findings of natural forest rainfall interception studies cannot be directly applied to the highly heterogeneous urban environment. Currently, there is still a lack of comprehensive databases detailing the combined effects of urban tree communities in terms of structural complexity, habitat conditions and species diversity. Therefore, in this study, 20 multi-dimensional indicators covering tree morphological characteristics, ecological attributes and community diversity indices were systematically selected from the perspective of urban social-ecological system coupling to comprehensively assess the role of tree species and communities in urban stormwater management.

To further address the challenges of urban flooding and fill knowledge gaps in city-scale studies, this study assesses the rainfall interception capacity of trees in urban park green spaces by integrating field survey data with the i-Tree Eco model. By comprehensively considering tree morphology, ecological interactions, and interspecies relationships, this study aims to identify key drivers affecting rainfall interception by urban trees, thereby providing practical guidance and scientific support for flood mitigation in green space management. Specifically, this study aims to: (1) quantify and compare the rainfall interception capacity of different tree species and community canopies in Zhengzhou, China; (2) identify key factors affecting rainfall interception by urban trees and their community canopies; and (3) propose measures to regulate rainfall interception by urban tree communities, thereby providing decision support for sustainable urban development.

2. Data and Methods

2.1. Study Area

Zhengzhou (E 112°42′–114°14′, N 34°16′–34°58′), is the capital of Henan Province in the Central Plains region of China [40] (Figure 1a). Zhengzhou has a temperate semi-arid continental monsoon climate, characterized by distinct seasons and significant seasonal variations in vegetation [41]. The city’s annual average rainfall is approximately 542.2 mm, with the most rainfall occurring from June to September, accounting for 65.7% of the total annual rainfall. Zhengzhou has experienced an intense urbanization process in the past 30 years [42]. On 20 July 2021 (the “7.20 event”), an extreme rainstorm triggered severe urban flooding, causing extensive infrastructure damage, significant casualties, and substantial economic losses [43,44]. This event highlighted the growing challenges of urban flood management in the context of climate change and rapid urbanization [45]. The study area is located in The core development area of Zhengzhou, including Jinshui, Erqi, Huiji, Zhongyuan, and Guanchenghuizu (Figure 1b,c). According to the Zhengzhou Statistics Bureau (https://tjj.zhengzhou.gov.cn/ (accessed on 14 August 2025), as of 2021, the main urban area had a permanent population of 5.0285 million, covering 105.3 km², with a population density of 4857 people/km² and a green coverage rate of 41.63%. The core development area contains 123 parks and green spaces, totaling 37.23 km², with a green coverage rate of 53.08%. The sample plots in the study area were evenly distributed within 123 parks (Figure 1d–f).

2.2. Research Methods

The research flow of this study is shown in Figure 2. Remote sensing imagery was used, and field survey were conducted on 805 sample plots across 123 parks in the core development area of Zhengzhou. The rainfall interception capacity of trees was quantified by integrating field data with the i-Tree Eco model. The differences in rainfall interception capacity and influencing factors were compared and analyzed among different tree species and communities through the ANOVA. Pearson correlation analysis was conducted to identify key factors influencing rainfall interception across various tree species and plant communities. Stepwise regression was applied to evaluate the relative importance of the driving factors. Finally, targeted strategies were proposed to address urban hydrological challenges.

2.2.1. Data Collection

This study incorporates both remote sensing and field sampling data. Remote sensing data were derived from GF-2 satellite images with a spatial resolution of 0.8 m, captured on 25 May 2017, and 16 April 2018, which were used to generate random points. These were manually calibrated using high-resolution Google Earth imagery of the study area from 2022, and the final accuracy of the classification results was evaluated by the confusion matrix, with an overall accuracy of 89%. Following the data collection guidelines in the i-Tree Eco User Manual, sampling plot locations were determined using a random point generation method. The number of sample plots per park was set according to the size of the park, based on a standard of one plot per 2.5 ha in the park, with one plot created for parks smaller than 2.5 ha and up to 25 plots measured for larger parks. Each sample plot was defined a circular area of 0.04 ha, centered on a randomly generated point. Plot locations were adjusted as necessary to account for on-site conditions, and the final spatial coordinates of each plot were recorded. Field data collection includes species composition, tree growth types, diameter at breast height, total tree height, crown width (east–west & north–south), branch height, crown dieback, percent crown missing, Crown light exposure, percent impervious surface under the tree. Field data were collected in July 2021, resulting in 805 valid sample plots. These data were essential for estimating rainfall interception using the i-Tree Eco model and served as a critical foundation for the analysis.

Foliar characteristics are important factors influencing rainfall interception in trees [46]. The trees are classified into three categories, Evergreen Broadleaf Tree, Deciduous Broadleaf Tree and Coniferous Tree, respectively, based on leave characteristics. Then the tree communities are divided into 7 types based on various tree species compositions, as shown in

Table 1. Tree species classification vs. tree community classification.

Classification Scale	Type Name	Abbr.	Composition Description
Tree Species	Evergreen Broadleaf Tree Species	EVET	Evergreen and Broadleaf
	Deciduous Broadleaf Tree Species	DECT	Deciduous and Broadleaf
	Coniferous Tree Species	CONT	Evergreen and Coniferous (Note: Deciduous conifers are not considered due to the limited representation by Metasequoia only.)
Tree Community	Pure Evergreen Broadleaf Forest	EVE	Composed exclusively of evergreen broadleaf tree species
	Pure Deciduous Broadleaf Forest	DEC	Composed exclusively of deciduous broadleaf tree species
	Pure Coniferous Forest	CON	Composed exclusively of coniferous tree species
	Mixed Evergreen and Deciduous Broadleaf Forest	E-D	Composed of evergreen broadleaf and deciduous broadleaf tree species
	Mixed Evergreen Broadleaf and Coniferous Forest	E-C	Composed of evergreen broadleaf and coniferous tree species
	Mixed Deciduous Broadleaf and Coniferous Forest	D-C	Composed of deciduous broadleaf and coniferous tree species
	Mixed Forest of Evergreen Broadleaf, Deciduous Broadleaf, and Coniferous Species	E-D-C	Composed of evergreen broadleaf, Deciduous broadleaf and Coniferous tree species

.

2.2.2. Calculation Method for Rainfall Interception by Tree Canopies

The i-Tree 2024_6.1.51 software is used to calculate rainfall interception by tree canopies. Developed by the USDA Forest Service in 2006, i-Tree is a tool designed to analyze and evaluate the benefits of urban forests. i-Tree Eco model requires inputting tree data collected from field surveys, meteorological conditions, air quality, hourly rainfall, and meteorological information, air quality, hourly rainfall, and climate zones of Zhengzhou to accurately obtain the spatial distribution of rainfall interception in the city [47]. The model applies species-specific allometric growth equations to calculate the parameters of each tree [48,49,50]. These allometric growth equations are directly related to tree species, leaf area, crown cover, and tree health [51]. Species-specific parameters are sourced from the i-Tree Eco database, and for species missing from the database, the most morphologically and ecologically functionally similar alternative species were used [48,49,50]. The operation process of i-Tree Eco in Figure 3.

The principle for simulating interception is as follows:

$$I=P-TH-D$$

(1)

where I (mm) is the interception, P (mm) represents the total precipitation, TH (mm) refers the throughfall (rainfall passing through the canopy), and D (mm) indicates the drip rainfall from the canopy surface. Note: Stemflow was not considered separately in this study because it typically constitutes a very small portion of total rainfall (less than 1–3%) and accurately quantifying it requires long-term individual tree observations, which was beyond the scope and data availability of this study.

In this study, the rainfall interception capacity of communities in park green spaces was represented by that of tree canopies. The rainfall interception volume of the plot divided by the total green space area of the plot is the formula for calculating the average annual rainfall interception efficiency (RIE) (mm·yr ⁻¹) for a single plot. The formula is as follows:

$$SDi(RIE)={\displaystyle \frac{{\sum }_{i=1}^t{I}_i}{A_i}}$$

(2)

where SD_i represents the RIE for plot i, A_i refers the green space area of plot i, I_i denotes the average annual rainfall interception (ARI) (m³ yr⁻¹) of the i-th tree in the plot, and t represents the total number of trees in the plot.

2.2.3. Driving Factors Calculation

To explore the influence of community characteristics on canopy rainfall interception in Zhengzhou’s urban parks, this study selected 20 factors through a comprehensive literature review and field survey data analysis as explanatory variables for differences in rainfall interception capacity of trees and communities [52,53,54,55], including 8 morphological traits, 8 ecological characteristics, and 4 species diversity indices. The specific calculation methods for each factor are shown in

Table 2. Calculation Methods and Abbreviations of Influencing Factors.

Categorization	Variable	Abbr.	Units	Description	Formula
Morphological Traits	Average Diameter at Breast Height	ADBH	cm	The Average of the Sum of Diameters at Breast Height for the Same Tree Species.	$\begin{array}{c}ADBH={\displaystyle \frac{{\sum }_{i=1}^n{D}_i}{n}}\end{array}$	(3)
	Average Tree Height	ATH	m	The Average of the Sum of Tree Heights for the Same Tree Species.	$\begin{array}{c}ATH={\displaystyle \frac{{\sum }_{i=1}^n{T}_i}{n}}\end{array}$	(4)
	Average Branch Height	ABH	m	The Average of the Sum of Branch Heights for the Same Tree Species.	$\begin{array}{c}ABH={\displaystyle \frac{{\sum }_{i=1}^n{B}_i}{n}}\end{array}$	(5)
	Average Crown Width	ACW	m	The Average of the Sum of Crown Widths for the Same Tree Species.	$\begin{array}{c}ACW={\displaystyle \frac{\sum _{i=1}^n{C}_i}{n}}\end{array}$	(6)
	Diameter at Breast Height Disparity	DBH_D	cm	The Difference Between the Maximum and Minimum Diameter at Breast Height for the Same Tree Species.	$\begin{array}{c}DB{H}_{-}D=D_{max}-D_{min}\end{array}$	(7)
	Tree Height Disparity	TH_D	m	The Difference Between the Maximum and Minimum Tree Heights for the Same Tree Species.	$\begin{array}{c}T{H}_{-}D=T_{max}-T_{min}\end{array}$	(8)
	Branch Height Disparity	BH_D	m	The Difference Between the Maximum and Minimum Branch Heights for the Same Tree Species.	$\begin{array}{c}B{H}_{-}D=B_{max}-B_{min}\end{array}$	(9)
	Crown Width Disparity	CW_D	m	The Difference Between the Maximum and Minimum Crown Widths for the Same Tree Species.	$\begin{array}{c}C{W}_{-}D=C_{max}-C_{min}\end{array}$	(10)
Ecological Characteristics	Importance Value	IV	-	Ecological Importance of Species.	$\begin{array}{c}IV=\left(\begin{array}{c}RD+COV\\ +RF+RH\end{array}\right)\\ ÷ 400\end{array}$	(11)
	Relative Density	RD	%	The number of individuals per unit area.	$\begin{array}{c}RD={\displaystyle \frac{N_{species}}{N_{total}}}\times 100\%\end{array}$	(12)
	Coverage	COV	%	The proportion of the ground area covered by the vertical projection of tree canopies.	$\begin{array}{c}COV=\left({\displaystyle \frac{{CW}_1+{CW}_2}{4}}\right)\pi \\ \times 100\%\end{array}$	(13)
	Average Breast Height Cross-Sectional Area	ACS	m2	The ratio of the square of tree diameters at breast height (DBH) to the total number of trees in the stand.	$\begin{array}{c}ACS=\pi {\left({\displaystyle \frac{DBH}{2}}\right)}^2\end{array}$	(14)
	Significance Value	SIG	-	The Total Basal Area at Breast Height (BA) for a/all Specific Species.	$\begin{array}{c}SIG={\displaystyle \frac{{\sum }_{i=1}^n{ACS}_i}{{ACS}_{total}}}\end{array}$	(15)
	Relative Frequency	RF	%	The percentage of sample plots in which a species occurs.	$\begin{array}{c}RF={\displaystyle \frac{S_{species}}{S_{total}}}\times 100\%\end{array}$	(16)
	Relative Height	RH	%	The height of a specific plant species as a percentage of the total height of all species.	$\begin{array}{c}RH={\displaystyle \frac{T_{species}}{T_{total}}}\times 100\%\end{array}$	(17)
	Planting Density	TD	trees/m2	The number of plants per unit area within the green space plant community.	$\begin{array}{c}TD={\displaystyle \frac{{\sum }_{i=1}^t{T}_i}{A_i}}\end{array}$	(18)
Species Diversity Indices	Gleason Richness Index	GLE	-	A measure used to describe the diversity of species within a community.	$\begin{array}{c}GLE={\displaystyle \frac{S}{lnA}}\end{array}$	(19)
	Shannon-Wiener Index	SW	-	An indicator describing the disorder and uncertainty in the occurrence of individuals within species.	$\begin{array}{c}SIM=-{\displaystyle \sum _{i=1}^S}{P}_i{lnP}_i\end{array}$	(20)
	Simpson Dominance Index	SIM	-	Represents the probability that two individuals randomly sampled from a community belong to the same species.	$\begin{array}{c}SW=1-{\displaystyle \sum _{i=1}^S}{P}_i^2\end{array}$	(21)
	Pielou Evenness Index	PIE	-	Used to describe the relative abundance and evenness of different species within an ecosystem or community.	$\begin{array}{c}PIE={\displaystyle \frac{SW}{\mathit{ln}S}}\end{array}$	(22)

. Our study used a variety of data analysis methods to explore the correlation between the rainfall interception capacity of urban trees and their drivers, as well as the relative contribution of these factors.

n represents the total number of trees of the same species within a sample plot. i represents the i-th tree. D_i, T_i, B_i, C_i represent the DBH, TH, BH and CW of the i-th tree, respectively. D_max, D_min represent the maximum and minimum DBH values. T_max, T_min represent the maximum and minimum TH values. B_max, B_min represent the maximum and minimum BH values. C_max, C_min represent the maximum and minimum CW values. N_species represents the total number of trees of a specific species. N_total represents the total number of trees of all tree species. S_species represents the number of sample plots where a specific species occurs. S_total represents the total number of all sample plots. CW₁ represents the east–west crown width. CW₂ represents the north–south crown width. T_species represents the total height of a specific tree species. T_total represents the total height of all tree species. t represents the trees within a sample plot. T_i represents the number of tree species within the i-th sample plot. A_i represents the area of the i-th sample plot.

3. Results

3.1. Vegetation Characteristics

3.1.1. Tree Structural Characteristics

This study surveyed 17,626 trees across 805 sample plots, identifying 38 families, 69 genera, and 99 species. The top 15 tree species by abundance, account for 59.94% of the total tree population. Ligustrum lucidum, Ginkgo biloba, and Cedrus deodara are identified as the dominant tree species in urban parks of Zhengzhou’s main district (Figure 4a). The trees in Zhengzhou’s parks are primarily small to medium-sized, with average DBH, TH, BH, and CW of 16.79 cm, 7.62 m, 2.05 m, and 4.28 m, respectively (Figure 4d–g). Of the three tree categories, DECT is the most prevalent type in Zhengzhou’s parks, comprising 41.8% of all community types (Figure 4b). Among the seven community types, E-D is the most common, representing 25.5% of all community types (Figure 4c).

3.1.2. Differentiation of Tree Species and Community Characteristics

Vegetation characteristics vary significantly among different tree species types (Figure 5). The ACS, SIG, and RF values of EVET are significantly higher than those of the other two tree species types, with values of 12.98 m², 0.0128, and 0.0148%, respectively. DECT has the highest IV and COV, with values of 4.9% and 19.57%, respectively. The values of CW_D (10.45 m) for CONT were significantly higher than those of the other tree species types. Significant differences among tree species types are observed only in ADBH, ACS, and SIG, with EVET showing notably higher values than the other two types.

Significant differences in vegetation characteristics are observed among the seven community types (Figure 6). The ADBH, ATH, and ACW values of CON are significantly higher than those of E-D-C, while DEC exhibits the highest ABH value of 2.4 m. The TH_D value of E-D-C is notably higher than that of D-C, at 7.8 m. Significant differences are observed in SW and SIM values among E-D-C, E-D, E-C, and DEC, with a declining trend. E-D-C also shows a significant difference in TD compared to the other six communities, with the highest value, averaging 0.075 trees per square meter.

3.2. Rainfall Interception Status

The RIE of urban park green spaces in Zhengzhou is 66.35 mm∙yr⁻¹, with individual sample plots ranges from 0.25 to 547.5 mm∙yr⁻¹ (Figure 7a). Rainfall interception capacity shows little regional variation, though Jinshui and Erqi districts include more areas with higher interception levels. The ARI of per tree for the 17,626 trees ranges from 0.1 to 32.3 m³∙yr⁻¹. The top 15 species in terms of average rainfall interception per tree are presented in Figure 7b. DECT dominates this ranking, with Paulownia fortunei, Populus tomentosa, and Zelkova serrata exhibiting higher annual interception values of 5.50 m³∙yr⁻¹, 4.41 m³∙yr⁻¹, and 3.83 m³∙yr⁻¹, respectively. Additionally, Platanus orientalis has the highest individual annual rainfall interception value, at 32.3 m³∙yr⁻¹. The estimated annual ARI and RIE values are derived from a single-month sampling period and may be influenced by seasonal variability in rainfall patterns and canopy conditions. Therefore, these figures should be interpreted as approximations rather than precise annual values.

Among the 805 sample plots, the plot with the highest annual rainfall interception has a total interception of 219 m³∙yr⁻¹ and the RIE of 547.5 mm∙yr⁻¹. The top 15 plots with the highest RIE, as shown in

Table 3. Information on RIE clusters in the top 15 positions.

Community Composition	Community Type	Total Rainwater Interception Volume (m3∙yr−1)	RIE (mm∙yr−1)
Populus tomentosa	DEC	219	547.5
Salix babylonica + Populus tomentosa	DEC	199.3	498.25
Salix babylonica + Platanus orientalis	DEC	160.2	400.5
Salix babylonica	DEC	158.9	397.25
Platanus orientalis + Ligustrum lucidum + Ginkgo biloba	E-D	150.5	376.25
Zelkova serrata	DEC	141.7	354.25
Populus tomentosa	DEC	138.1	345.25
Cedrus deodara + Ligustrum lucidum	E-C	134.3	335.75
Populus tomentosa +Robinia pseudoacacia + Juglans regia	DEC	129.9	324.75
Cedrus deodara + Bischofia polycarpa	D-C	127.5	318.75
Fraxinus chinensis	DEC	124	310
Platanus orientalis	DEC	122.3	305.75
Ginkgo biloba	DEC	119.6	299
Paulownia fortune + Populus tomentosa + Robinia pseudoacacia Linn.	DEC	117.7	294.25
Platanus orientalis + Salix babylonica	DEC	114.1	285.25

, are predominantly dominated by DEC.

ARI varies across tree species types (Figure 7c), with DECT exhibiting the highest ARI of 1.28 m³∙yr⁻¹ among the three tree species types. In comparison, both EVET and CONT have ARI values below 1 m³∙yr⁻¹, with EVET showing the lowest ARI.

Among the seven community types, CON has the highest RIE, at 90.7 mm∙yr⁻¹, followed by DEC, at 76.9 mm∙yr⁻¹. The RIE values for D-C and E-C is 66.3 mm∙yr⁻¹ and 57.5 mm∙yr⁻¹, respectively. E-D-C and E-D have identical RIE values, while EVE exhibits the lowest RIE (Figure 7d). These results indicate that canopy interception efficiency varies across different community types.

3.3. Correlation Between Vegetation Characteristics, Tree Species Types ARI, and RIE of Different Communities

Across all tree species, the correlation coefficients between ATH and RH are the highest, both at 0.802, followed by ACW with a coefficient of 0.787. For EVET, RH and ATH, as well as COV and IV, show identical correlation coefficients, demonstrating strong positive correlations with ARI, at 0.831 and 0.788, respectively. In DECT, the correlation coefficients for RH, COV, IV, ACW, and ATH all exceed 0.79. For CONT, ACW exhibits the strongest correlation, with a coefficient of 0.715. Compared to EVET and CONT, DECT exhibits more correlated factors. Specifically, CW_D, BH_D, TH_D, and ABH are only correlated with DECT’s ARI. Meanwhile, COV, IV, ACW, and ATH exhibit strong positive correlations across all three tree species types (Figure 8a).

Across all sample plots, ADBH and ACW show the strongest positive correlations, with coefficients of 0.667 and 0.64, respectively. In E-D-C, ACW has the highest correlation coefficient of 0.663. In EVE, only TD, ATH, and ACW are significantly correlated with the community RIE. E-D and DEC share the same primary correlation factor, ATH, with correlation coefficients of 0.651 and 0.685, respectively. In E-C, only TD shows a significant positive correlation with RIE, with a coefficient of 0.92. In D-C, ACW has the highest correlation coefficient of 0.626, followed by ATH and TH_D. In CON, ATH and ACW show significant positive correlations with RIE, with coefficients of 0.746 and 0.699, respectively. Notably, species diversity indices show no correlation with RIE in any of the seven community types (Figure 8b).

3.4. Analysis of the Factors Influencing Tree and Community Rainfall Interception Capacity

Among all trees, ATH (63.9%) has the greatest influence on ARI, with IV also contributing. For EVET, RH (66.6%) is the most significant factor affecting ARI. In DECT, multiple factors influence ARI, with ATH (67%) having the greatest impact, followed by IV (13.1%). For CONT, both ACW (46.3%) and ATH (31.5%) significantly impact ARI (Figure 9a).

Across all sample plots, ATH (44.4%) has the greatest influence on RIE, followed by ACW and TD. In E-D-C, RIE is primarily influenced by ACW (43.3%). ATH (40.8%) and TD (32.2%) are the critical factors for EVE. In E-D, ATH (46.7%) is the decisive factor, contributing the most. In E-C, TD accounts for 83.4% of the variation, making it the key factor in determining RIE. ATH (46.7%) is the primary driving factor for DEC, while ACW (38.6%) and TD (24.3%) contribute the most to RIE in D-C. For CON, RIE is mainly influenced by ATH (53.5%) and ADBH (19.8%) (Figure 9b).

4. Discussion

4.1. The Importance of Rainfall Interception by Urban Trees

This study finds that the annual average rainfall interception efficiency of trees in urban park green spaces in Zhengzhou is 66.35 mm∙yr⁻¹, further illustrating the significant role of urban trees in stormwater management. Based on canopy cover data [56], we estimated that trees in Zhengzhou intercept approximately 553,171 m³ of rainfall annually. Assuming typical runoff coefficients for impervious urban surfaces, the intercepted 553,171 m³ could prevent an equivalent volume from contributing to peak stormflows, thereby alleviating pressure on urban drainage infrastructure during heavy rainfall events. Furthermore, this volume of water is enough to meet the water demand of 3000–4000 households for one year, further highlighting the crucial role of urban trees in urban water resource management, particularly in mitigating urban flooding and improving stormwater management. In comparison, this annual interception volume significantly exceeds that of urban trees in Santa Monica, California (193,168 m³) [57]. This difference can be attributed to the variation in local dominant tree species. Livesley et al. found that Eucalyptus saligna and Eucalyptus nicholii intercepted 29% and 44% of annual rainfall, respectively, corroborating the key role of tree species selection. This difference may be related to the higher tree density and canopy cover in Zhengzhou. Compared to international cases, Qingfu Xiao et al. found that for the same rainfall intensity and duration, the rainfall interception rate of North American maple, a deciduous tree species in Santa Monica, California, was 70.5% during the growing season, compared to only 5.5% during the winter defoliation period. In addition, Santa Monica has a Mediterranean climate characterized by mild, wet winters and dry summers. This climatic contrast with Zhengzhou is one of the factors contributing to the difference in annual rainfall interception between the two cities. With the acceleration of urbanization, the hydrological regulation function of urban trees will play an increasingly important role in sustainable urban development.

This study, using the i-Tree model, estimated the ARI of 1.23 m³/year for a single tree in Zhengzhou. Under the same model, the result is higher than the rainfall interception of a single tree in Luohe (0.67 m³ year⁻¹) [58], but lower than the rainfall interception of each street tree in Manhattan, New York, USA (2.228 m³ year⁻¹) [59]. In Zhengzhou, most trees have a DBH between 10 and 20 cm, while Luohe has many young trees, with a DBH ranging from 7.7 to 15.2 cm. In contrast, many trees in Manhattan were planted earlier, with 43% of street trees having a DBH of 15–30 cm. Manhattan has a humid subtropical climate with higher rainfall in summer. Studies have shown that rainfall amount is positively correlated with canopy rainfall interception in a stepwise pattern [38]. In addition, urban form and policy factors may also contribute to this difference. This confirms the feasibility of the i-Tree model and indicates that DBH is a key factor influencing the rainfall interception Additionally, the Jinshui and Erqi in Zhengzhou exhibit higher rainfall interception. The two districts have earlier urban construction and relatively rich plant communities, which fundamentally improves the rainfall interception efficiency in these areas. Local environmental factors, building density, and roadway form play an important role in the rainfall interception capacity of trees [60]. It is worth noting that simply increasing density may be counterproductive. When planting density is too high, restricted space for single tree growth can hinder crown development, reducing the RIE [61]. Therefore, scientifically selecting high-retention tree species, optimizing community vertical structure, and maintaining reasonable planting density should be key strategies for sponge city construction.

4.2. Differences in Rainfall Interception Capacity of Tree Species and Key Influencing Factors

Our study found that DECT, such as Paulownia fortune (5.5 m³∙yr⁻¹) and Populus tomentosa (4.41 m³∙yr⁻¹) had the highest ARI. In contrast, CONT has lower total interception, with an ARI of less than 2.0 m³∙yr⁻¹ per tree, although most CONT intercept rainfall year-round. The flat leaves of DECT store water as a film and generally exhibit larger CW and leaf area, which intercept more rainfall than the needle-like leaves of CONT. However, previous studies have indicated that CONT tend to exhibit higher rainfall interception than both DECT and EVET [62]. It should be noted that annual interception provides a long-term assessment, whereas interception during individual rainfall events reflects short-term performance. The overall ARI of EVET was lower than that of other two species, EVET typically have leaves with smoother surfaces and fewer stomata, limiting their ability to absorb and interrupt rainfall [63,64]. Consequently, DECT can more effectively reduce surface runoff and relieve stormwater pressure, and are suitable for application to cities like Zhengzhou where seasonal rainfall is concentrated, while CONT is more appropriate for areas with uneven rainfall throughout the year.

This study indicated that tree morphological characteristics strongly influence the rainfall interception capacity of tree species among the 20 factors selected. Correlation analysis revealed that ATH, ACW, ABH, TH_D, and RH exhibit strong positive relationships with the ARI of tree species. Viewed as a whole, ATH (63.9%) contributed the most, which indicates that the characteristics of tree species affect the rainfall interception capacity [65,66]. Our findings on the effects of arborvitae species characteristics on their rainfall interception capacity are consistent with previous studies [67]. IV, COV, and RH also showed strong positive correlations with the ARI of tree species. Tree species with higher IV tend to be larger, widely distributed and more abundant, which has a greater impact on rainfall interception in the community. Species with greater coverage have denser canopies, which reduces direct rainfall reaching to the ground and enhances interception within the canopy layer [33,68]. In EVET, RH (66.6%) contributed the most, indicating that RH is a decisive factor in its rainfall interception capacity. In previous research, Anys and Weiler found no correlation between DBH and TH in small-leaved linden trees, likely because the experiment, conducted from June to September, overlooked seasonal variations [69]. Seasonal changes can lead to leaf shedding, which subsequently affects the ARI of tree species [70]. Our study did not account for the impact of different rainfall events on interception volumes, whereas previous research have shown that low-intensity, long-duration rainfall events yield higher interception volumes [17,35].

4.3. Differences in Rainfall Interception Capacity of Communities and Key Influencing Factors

There are significant differences in rainfall interception capacity across community types, and the structural complexity of plant communities is a critical factor in rainfall interception efficiency [71,72]. This study found that the RIE of CON exceeded that of the other six community types, consistent with previous research [73]. This is likely because CON have higher ADBH, ATH, ACW, and lower BH_D, while CONT have denser, more overlapping branches and leaves, and also maintain interception of rainwater during winter. In contrast, most DECT lose their leaves in winter [67,74]. Evergreen broadleaf mixed communities showed higher RIE than single-species communities, suggesting that species diversity and structural complexity enhance the overall urban rainfall interception capacity. Ilek demonstrated that mixed forests have a greater rainfall interception capacity than pure forests, and tree-shrub mixed forests with a richer structural hierarchy intercept more rainfall than mixed tree forests [75]. Kermavnar and Vilhar also found that the strongest optimal stand structure for rainfall interception capacity occurs in mixed forests, regardless of season [71]. In designing urban green spaces, community structure should be optimized to enhance ecological benefits. Notably, DEC also exhibited a high RIE of 76.9 mm∙yr⁻¹. The higher RIE of deciduous broadleaf pure forest communities, compared to mixed deciduous broadleaf communities, suggests that the greater ADBH, ATH, ABH, and ACW exert a stronger effect. At the same time, DECT with large vertical growth and horizontal extension intercept more rainfall, which is extremely beneficial during summer [67].

Our results show that ATH significantly influences the RIE of communities. For CON, DEC, EVE, and E-D, ATH consistently contributed the most to their RIE, which shows that ATH within a community strongly influences its rainfall interception capacity. Empirical studies show that the tree height determines the distribution of leaves in space [76], which in turn affects overall interception efficiency. ATH influenced the leaf area index (LAI) and canopy coverage in communities. Research has shown that LAI, canopy coverage, and canopy rainfall interception are positively correlated [77]. Taller trees typically exhibit larger canopy coverage, which enhances the interception capacity of the entire community. In the mixed D-C and E-C communities, RIE was primarily affected by TD, which indicated that the planting strategies should be fully considered when planting broad-leaved trees with conifers. The planting density of trees in communities directly affects the canopy coverage [78]. High-density planting creates more continuous and complete canopies, increasing the interception time and redistribution of rainfall, which improve overall interception efficiency. Morphological and physiological differences between conifers and broadleaf trees also can improve overall interception capacity. RIE is more influenced by ACW in DEC. CW directly affects stand structure, like the hierarchical structure and density of the canopy [79]. In mixed conifer-broad forests, tree species with larger CW may form an upper canopy, while tree species with smaller CW may form a lower canopy, and this structure has an important effect on rainfall interception and distribution.

4.4. Limitations and Future Outlook

To further translate the quantitative findings of this study into practice management and planning recommendations, we identified specific strategies to enhance the rainfall interception efficiency of urban tree communities based on the response relationships of the key drivers.

Table 4. Strategies for improving rainfall interception capacity of urban tree communities based on driving factors.

Driving Factor	Influence Mechanism	Optimization Strategy	Applicable Urban Contexts
ATH	Taller trees increase vertical canopy depth and enhance initial interception	Prioritize medium-to-tall tree species; enhance maturity; reduce prevalence of low-stature monocultures	Main park corridors, flood-prone greenspaces, campus cores
ACW	Wider crowns provide greater interception surface and delay infiltration	Select native species with broad crowns; space trees to promote lateral crown expansion	Street greenbelts, riverside strips, residential peripheries
RH	Multi-layered canopies enhance rainfall redistribution and multi-phase interception	Establish vertical stratification using tall canopy trees and lower understory vegetation	Peri-urban greenbelts, park edges, ecological buffer zones
TD	Denser planting increases canopy coverage and retention but may restrict growth	Adjust density by species and function; promote compact arrangements in constrained areas	Flood-prone districts, compact micro-greenspaces
CW_D	Variability in crown size fosters spatial heterogeneity and layered interception	Combine species with diverse crown shapes; avoid uniform layouts	Park understories, urban courtyards, multifunctional greenspaces
IV	Dominant species contribute significantly to community-level interception	Increase the proportion of high-performance species (e.g., Paulownia, Populus tomentosa, Koelreuteria)	Species transition zones, functional green infrastructure sites

summarizes the strategies of each critical factor, providing a reference for sponge city construction and sustainable urban development.

Based on these findings, we propose a multi-scale optimization strategy across micro, meso, and macro levels. At the micro level, while ensuring the normal growth of trees, rainfall interception efficiency can be enhanced by selecting tree species with larger DBH, TH and CW, and by scientifically adjusting the planting density to form a continuous, complete canopy [67]. DECT, such as Paulownia fortunei, Populus tomentosa, Zelkova serrata, and Koelreuteria paniculata exhibit high ARI per tree. Therefore, these species should be prioritized in urban planting. At the meso level, efficient and stable community structures should be constructed to improve overall rainfall interception capacity. CON provide high RIE, meaning they can consistently maintain high interception efficiency throughout both rainy and dry seasons. Consequently, CON are well-suited for sloped areas or locations where rainwater accumulates in winter, enhancing both interception and infiltration efficiency. In contrast, DECT are suited for open green spaces and park plazas, where they help manage seasonal rainfall. However, considering the biodiversity and ecological balance, it is recommended to plant a mixture of CONT and DECT to form a mixed forest in order to improve the structural complexity and ecological efficiency. The REI of evergreen broadleaf mixed community is better than that of single community, so the rainfall interception capacity of the community can be improved by increasing the number of plant species and constructing a multi-level plant community structure [46]. At the macro level, Zhengzhou has a certain capacity for rainwater management, but there are significant differences in the structure, management, and construction standards of various green spaces. Strengthening the spatial layout and integration of the green space system is essential. Priority should be given to improving areas with lower rainfall interception efficiency by concentrating green spaces, increasing planting density, and fostering the formation of rich plant communities in newly built green spaces [80]. High RIE tree communities should be prioritized in areas prone to urban flooding, and interception functions should be integrated into sponge city planning, park landscapes, and street greening. This systems-based approach enhances the overall rainfall interception capacity of urban trees. These findings bridge ecological insights with policy-oriented spatial design, offering guidance for plant selection and layout in urban parks, and contributing to the reduction in urban flood risk and alleviation of water scarcity.

While the i-Tree Eco model effectively captures the relationship between community structure of urban trees and hydrologic benefits, several limitations remain. Firstly, the ARI and RIE estimates are informative, they are derived from a single month of data and therefore do not capture how rainfall patterns and seasonal changes affect the ability of trees to intercept rainfall. Future work should adopt multi-season or year-round monitoring to better characterize how this ability varies over time [81]. Second, the model’s species database is primarily based on North American tree species [82], which may differ from Zhengzhou’s native species. Additionally, the use of city-wide average meteorological data overlooks microclimatic variability within the study area [58]. Future work should develop localized databases for Zhengzhou and other Chinese cities, and integrate tools such as i-Tree Hydro to simulate stormwater management scenarios under varying rainfall and climate conditions. Lastly, this study focused exclusively on the tree interception, but urban rainwater management needs to take into account the contribution of other factors such as ground cover, grasses, shrubs, and permeable paving [83,84]. Future research should adopt a vertical structural perspective to evaluate how diverse vegetation layers contribute to rainfall management across various urban spatial configurations, thereby informing more comprehensive urban ecosystem planning.

5. Conclusions

The rainfall interception effect of urban trees significantly enhances urban hydrological regulation services to mitigate water-related risks. This study, focused on Zhengzhou, utilized remote sensing and field survey data to analyze the rainfall interception capacity of 805 plots. It quantified the annual rainfall interception capacity of urban trees and their communities at the city level.

The results show that the RIE of Zhengzhou’s parks is 66.35 mm∙yr⁻¹, with the ARI of per tree being 1.23 m³∙yr⁻¹, totaling approximately 550,000 m³ of annual interception. This highlights the significant ecological role of urban trees in regulating urban hydrological processes. Stepwise regression analysis revealed several key drivers, including ATH, ACW, TD, and RH. ATH had a greater impact on the ARI of DECT and CONT, while EVET, with a higher RH, shows a higher ARI. The RIE of communities is most influenced by ATH, followed by ACW. TD is the decisive factor for the RIE of E-C. These factors, spanning structural characteristics, functional attributes, and community composition, provide various pathways for enhancing rainfall interception capacity through species selection, spatial configuration optimization, and community structure adjustments. Our study constructed a multi-scale enhancement strategy. At the micro scale, priority should be given to species with high interception rates, like Paulownia and Populus, as optimizing species selection and density can enhance interception efficiency per unit area. At the meso scale, composite communities should be developed to enhance functional complementarity among species. At the macro scale, high-efficiency interception communities should be prioritized in areas prone to flooding, park squares, and green space renewal zones, creating green stormwater buffer zones.

This study bridges the gap between the quantitative assessment of ecological processes and urban sustainability practices, offering valuable insights for urban stormwater management and sponge city development. Future research should establish long-term monitoring systems, integrating multi-layered vegetation structures and multi-model simulations, to systematically evaluate the comprehensive role of urban trees in sustainable urban development through a multi-scale, dynamic, and contextual approach.