A Multi-Level Evaluation of the 3-30-300 Rule Using Individual- and Building-Level Analyses in the Japanese Urban Context: A Case Study of Himeji City

Study Area and Data

This study focused on the urban area of Himeji City, Hyogo Prefecture, Japan (Figure 1). With a population of 500,000, Himeji City boasts a dual character: it serves as the economic center of Western Hyogo Prefecture and functions as a suburb of the greater Osaka metropolitan area. The area around the Himeji Station is highly urbanized. The Northern part of the station features facilities, such as the World Heritage-listed Himeji Castle and art museums, whereas the Southern part features an office district, including a city hall, surrounded by residential buildings. In the Central and Southern areas, rail and bus networks are well developed, with growing suburban residential zones. The Southern part is known for the Keihanshin Industrial Zone. The Northern part of the city features scattered settlements with fields and farmlands nestled in mountainous areas and valleys. Hills dot the urban landscape, and some of these undeveloped areas within the city have been designated as parks.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 1.

Study area: urban areas in Himeji City, Japan.

The data required for this study was obtained from the datasets available for the target area (Table 1). For the location and attributes of all individuals, the Population Attribute Distribution Data from the Pseudo People Flow synthetic population data developed at the national level in Japan was utilized (CSIS 2023; Kashiyama et al. 2024). The dataset refines national statistical data from the 2020 Census Administrative Units, originally aggregated by subregion, to represent information at the level of individual persons (Kashiyama et al. 2024). The estimation model behind this downscaling process reflected on the building’s type, gross floor area, and family type (Kashiyama et al. 2024). Each synthetic individual is assigned latitude and longitude coordinates of the representative points of each residential building, as well as attributes such as household size, age, sex, and occupation (Figure 2) (CSIS 2023). Therefore, it can be used to simulate the residential patterns of individual residents at the micro level, serving as an alternative to censuses that utilize small geographic areas as their unit of aggregation and lack microdata. The accuracy (R² = 0.81) compared with mobile phone data was reported for the population distribution on a 1,000 m × 1,000 m grid (Kashiyama et al. 2024). These data are publicly available to researchers as part of open data initiatives and have been used in academic research (Garcia-Gabilondo et al. 2024; Ma et al. 2024; Ami et al. 2025).

• Download figure
• Open in new tab
• Download PowerPoint

Figure 2.

Example of the building data and synthetic population data. The coordinates correspond to representative points of buildings in both datasets; however, the building data contains one building point and its attributes for each building, while the synthetic population data contains multiple individual points and their attributes for each building.

Analytical Framework

This study consisted of 3 steps of evaluation: (1) a multi-level assessment of the 3-30-300 rule both at the individual and building levels; (2) examination of how achievement of the 3-30-300 rule varies across population groups residing in different residential building types and urban densities; and (3) examination of how achievement of the 3-30-300 rule varies across population groups with different attributes.

In the first step, this study compared two evaluation units: the building-level assessment and individual-level assessment. The building-level assessment of the 3-30-300 rule is to examine whether each representative point of residential buildings meet each component. Analysis at the building level allows us to understand the distribution of compliance with the 3-30-300 rule. However, population size varies considerably among residential buildings, with apartments typically accommodating more residents than detached houses, preventing us from accurately identifying the number of residents, differences in their attributes, and those who are unable to access urban green spaces and trees. The individual-level assessment, which is the focus of this study, is to examine whether the specific point coordinates of each population meet each component. In the synthetic population data, the location of each individual has been estimated with high accuracy. The achievement status (met/unmet) of each component of the 3-30-300 rule and the number of components met (unmet, only one met, only two met, and all met) with the 3-30-300 rule using building- and individual-level assessment were examined. Statistical analysis revealed the proportion of the rule met based on the total number of buildings and total number of people as denominators for building-level analysis and individual-level analysis, respectively.

In the second step, residential building types were categorized into 3 groups: apartments, detached houses, and others. Others include those classified as office or “unknown” in building data but considered residential building in the synthetic population data. Urban density was calculated as a continuous variable using the floor area ratio computed for each 100-m mesh.

In the third step, sex and age in the synthetic population data were converted into categorical (male/female) and dummy variables, respectively, with 10-year increments.

Esri ArcGIS Pro 3.4 (Redlands, CA, USA) was used for spatial analysis at each step, and jamovi 2.6.26.0 (The jamovi project, Sydney, Australia) was used for statistical analysis.

Assessment of the 3-30-300 Rule

Examination Across Population Groups and Urban Densities

The relationship between the 3 residential building types and the achievement status (met/unmet) of each component as well as the number of components met (unmet, only one met, only two met, and all met) were verified using chi-square test. The building type was identified based on the building data.

The relationships between floor area ratio as an urban density measure and the achievement status as well as the number of components met were investigated using Mann-Whitney U test and Kruskal-Wallis test, respectively.

Examination Across Population Groups with Different Attributes

Chi-square tests were used to examine several relationships: between achievement status (met or unmet) and sex; between achievement status and age; between number of components met (unmet, only one, only two, or all met) and sex; and between number of components met and age.

Status of the 3-30-300 Rule

The analysis revealed that the study area contained 3,194,902 trees, 152.2 km² of green cover, and 1,168 parks for urban green spaces and trees. It also comprised 88,910 residential buildings and 487,845 individuals.

The analysis demonstrated that 21.7% of residential buildings and 27.5% of individuals met component 3, the capability to see 3 trees. The median number of visible trees was 1.0 per residential building and 1.0 per individual while the average was 2.88 trees per building and 3.47 trees per individual. In both cases, individual-level evaluations exceeded building-level evaluations. Over 35% of the population could not see a single tree from residential buildings, whereas over 20% could see only one tree (Figure 3).

• Download figure
• Open in new tab
• Download PowerPoint

Figure 3.

Histogram of visible tree count, green cover ratio, and distance to green spaces at the individual level.

Achieving component 30 was the most difficult. Only 21.0% of the residential buildings and 17.8% of the population could enjoy 30% green cover within 300 m of their residential buildings. The median green cover rates were 11.5% (building level) and 10.0% (individual level), whereas the means were 18.4% (building level) and 16.7% (individual level). In both cases, individual-level assessments exceeded building-level assessments. Conversely, approximately 30% of the population resided in areas with only 5% to 10% of green cover, and over 20% had less than 5% cover (Figure 3).

Achieving component 300 was the easiest. Green space within 300 m was accessible for 65.2% of the residential buildings and 66.3% of the population. The median distance to green space for individuals was 229 m (building level) and 228 m (individual level), whereas the average distance was 292 m (building level) and 285 m (individual level). Most of the population had an access distance of approximately 200 m (Figure 3).

Only 5.5% of the residential buildings and 5.1% of the individuals met all 3 components of the 3-30-300 rule (Table 2); 19.0% of the residential buildings and 21.9% of the population met two components, whereas 53.6% of the residential buildings and 52.4% of the population met only one component. Among these, 7.8% of the residential buildings and 12.9% of the population met components 3 and 300 but failed to meet component 30, indicating that green cover rate was a limiting factor. None of the requirements were met by 22.1% of the residential buildings and 20.6% of the population.

Figure 4 shows that residential buildings that met component 3 were mixed with those that did not, extending from the South to the Central area. Component 30 is concentrated along the urban fringe, Himeji Castle, and rivers. Component 300 demonstrated a mixed distribution of compliant and noncompliant buildings. The areas that failed to meet any of the 3-30-300 rule were located in the Central region, along the coast, and in scattered villages in the North. Conversely, areas meeting all the components were located on the urban fringe.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 4.

Maps showing the achievement status of each component of the 3-30-300 rule. (A) Building type; (B) Floor area ratio; (C) Visible tree count; (D) Green cover ratio; (E) Distance to green spaces; (F) Analysis of components.

Residential Building Type and Urban Densities

The Chi-square test revealed that differences in residential building type were significantly associated with meeting the 3-30-300 rule (Table 3). Apartments outperformed detached houses for components 3 and 300, whereas detached houses outperformed apartments for component 30. The proportion of residential building types capable of meeting component 30 was significantly lower in apartments than in detached houses. Only approximately 3.9% of apartments met all 3 components of the 3-30-300 rule compared to approximately 6.0% of detached houses (Table 4). Individuals who met none of the components were 17.8% for apartments and 21.2% for detached houses.

The Mann-Whitney U test revealed a significant relationship between the surrounding floor area ratio and compliance with the 3-30-300 rule (Table 3). Figure 4 demonstrates the average surrounding floor area ratio at the 100-m mesh level. The Kruskal-Wallis test revealed differences among multiple components: individuals meeting all 3 components were located in areas with 43.9% floor area ratio, whereas those who failed to meet any component were located in areas with 62.1% floor area ratio (Table 4).

Individual Attributes

The Chi-square test displayed no significant differences in sex for components 3 or 300; however, a higher proportion of males met component 30 (Table 3). Approximately 5.1% of males and 5.0% of females met all 3 components, and 20.5% of males and 20.613.4% of females failed to meet any component (Table 4).

The Chi-square test revealed age-related differences for all components (Table 3). Components 3 and 30 exhibited higher proportions of being met among the older age groups. The green cover ratio was relatively high across the younger age groups, with notably low values observed in the working-age population. Conversely, component 300 was higher among younger age groups and the working-age population and lower among older age groups. Approximately 5.5% of the 60 to 69 age group met all components, although low values were observed in the working-age population (Table 4).

Status of the 3-30-300 Rule

The results for Himeji City show that only 5.5% of the buildings and 5.1% of the individuals met all aspects of the 3-30-300 rule. While a precise comparison is not possible owing to differences in methodology across studies, the score in Himeji City outperforms Barcelona (Nieuwenhuijsen et al. 2022), Florence (Lopez et al. 2025), and Xiamen (Zheng et al. 2024); however, it significantly lags behind Singapore (Croeser et al. 2024) and the Netherlands (Helbich et al. 2025). The results indicate that, despite the 3-30-300 rule being established as a target for the well-being of all residents (Konijnendijk, 2022; Konijnendijk, 2023), Himeji City reports insufficient urban green spaces and trees.

The findings of this study indicate that achieving Component 30 within the rule is particularly challenging. This trend is consistent with the findings from Turin (Italy)(Battisti et al. 2024), Barcelona (Nieuwenhuijsen et al. 2022), and a comparison of 6 global cities (Croeser et al. 2024), but the underlying mechanisms differ fundamentally. In Europe, high building density and historic street layouts often limit the space for canopy expansion. By contrast, Himeji’s challenges are likely influenced by the transformation of residential gardens and street trees. Although our spatial analysis did not directly assess temporal changes, evidence shows that reductions in trees in the backyards of residential buildings are observed because of the subdivision of long-standing estates into multiple detached houses or redevelopment projects involving apartments (Shiraishi and Terada 2024). Street trees face issues such as insufficient planting space and excessive pruning, preventing tree canopy expansion (Fujii 2022; Egerer et al. 2024), and a fundamental lack of street-tree planting, particularly in areas distant from stations.

The relative ease of achieving Component 300 is possibly related to urban planning. Residential building development in Japan often occurs on a small scale and involves the conversion of farmland, vacant lots, or large single houses into detached housing areas (Takatori 2000; Nakahara and Hoshino 2006). In Japan, the provision of parks or similar facilities is required in residential building development above a certain size. However, some small parks remain unused, fail to meet local needs, and provide almost no ecosystem services (Okamoto et al. 2024). This suggests that meeting Component 300 alone is insufficient if the quality and usability of urban green spaces are lacking. Future implementation of the 3-30-300 rule in the Japanese context should consider not only the distance but also the functional capacity of these urban green spaces to ensure their alignment with the rule’s purpose.

Residential Building Type and Urban Densities

Residential building type and the 3-30-300 rule were related. Meeting Component 30 and the overall 3-30-300 rule were more advantageous for detached houses. Component 30 can only be met on the fringes of urban areas, where apartments are rarely constructed. The fringe of urban areas was characterized by a “satoyama” landscape, where residential buildings and forests coexisted (Takeuchi 2010; Takeuchi et al. 2016). This environment was used to support activities such as firewood and charcoal collection as well as agricultural work before industrialization. In addition, during the period of rapid economic growth, the construction of new towns and housing developments progressed in the fringe area of urban locations (Kato and Takizawa 2024). Conversely, apartments have advantages in components 3 and 300. Apartments are frequently situated along main roads in central urban areas. Residents of these residential buildings may observe street trees, with notable variations in street tree density among different street classes (Smart et al. 2020). Additionally, some apartments have trees planted on their grounds. Conversely, some detached houses only face narrow streets, making it more difficult for them to benefit from street trees. The number of detached houses without gardens and residents who plant less trees in their gardens have increased in recent years (Itahana and Terada 2022; Shiraishi and Terada 2024).

A relationship between urban density and the 3-30-300 rule also exists. Specifically, a higher floor area ratio made it more difficult to meet Component 30 and all components, but it was easier to meet Component 300. Component 30 was met in the urban fringe, which can provide an extensive canopy in satoyama landscape (Takeuchi 2010; Takeuchi et al. 2016), whereas Component 300 was met in the urban center, wherein parks have been developed.

The disparities observed across residential building types and urban densities in Himeji City align with the global discussion of loss of public and private urban green space in compact city and densification (Haaland and Konijnendijk 2015), wherein high-density apartment areas facilitate park access but limit canopy cover. However, differences between Himeji City and other cities exist. In Europe and the United States, low-density detached houses are typically located in affluent areas characterized by substantial tree cover and access to parks (Locke et al. 2021). Some cities such as Singapore implement policies that promote vertical greening, rooftop greening, and large-scale green spaces (McDonald et al. 2023), mitigating the lack of green spaces in dense apartment areas.

Individual Attributes

Negligible differences were observed in terms of sex; however, differences according to generation were observed. Components 3 and 30 were achieved more frequently by the elderly, whereas younger people and the working-age population achieved Component 300. Component 30 is thought to be influenced by the aging of urban fringe areas, whereas Component 300 is believed to be influenced by recent developments that have secured access to small parks. Improved access to parks is important for younger generations, as it helps secure children’s play areas and educational opportunities (Chawla 2015). However, it also plays a significant role for the elderly by providing walking paths and opportunities for physical activity (Zhai et al. 2020).

These findings indicate disparities in the equitable access to urban green spaces (Wen et al. 2013; Yu et al. 2023). However, the situation varies by city. In Hannover, Germany, studies have revealed that older adults have limited access to green spaces as well as disparities in access to the types of green spaces that offer age-related requirements (Weber et al. 2023). However, in Szeged, Hungary, research has revealed that areas with abundant green spaces tend to boast higher concentrations of older adults (Kolcsár et al. 2022). The case of Himeji City illustrates a unique mismatch between people and space, characterized by the aging of suburban areas and the return to the city center of the working-age population. Populations that are unable to meet the 3-30-300 rule may face heightened health risks and increased vulnerability to social isolation. Therefore, conducting analyses from a social inclusion perspective, focusing specifically on child-rearing households and low-income groups, is necessary.

Building- and Individual-Level Evaluations

This study evaluated the 3-30-300 rule at the individual level. Discrepancies were observed between the building- and individual-level evaluations. For components 3 and 300, individual-level evaluations exceeded building-level evaluations. However, only 5.5% of the residential buildings and 5.1% of the population fully met all components. For those failing to meet any component, individual-level evaluations were lower than building-level evaluations: 22.1% of the residential buildings and 20.6% of the population.

Aggregating data into administrative and building units often masks intraneighborhood disparities, a phenomenon known as the Modifiable Areal Unit Problem (MAUP). By using synthetic population data, this study overcomes this limitation and visualizes microscale environmental injustices. This big data approach to mitigating the MAUP (Dalton and Thatcher 2015; Luo et al. 2025) allows urban planners to identify specific neighborhoods and demographic clusters, enabling more targeted resource allocation to achieve the 3-30-300 rule.

The individual-level evaluation conducted in this study indicates the need for distinct greening strategies to address specific environmental injustice. In detached housing areas, where private gardens are shrinking (Itahana and Terada 2022; Shiraishi and Terada 2024), incentives to preserve existing trees and prevent full-site paving to enhance tree visibility are suggested. Conversely, for apartments in high-density areas with working-age populations, public strategies include providing street trees along commuting corridors and enhancing green roofs and vertical greening rather than new park provision.

Limitations

This study utilized synthetic population data to identify individual locations. While synthetic population data offers great accuracy (Kashiyama et al. 2024), it does not capture actual residential conditions. Because it is generated using models based on statistical data, individuals who significantly deviate from these models, such as those in large families, may not be accurately represented. Nevertheless, for the purpose of identifying individual locations, this dataset is superior to others in terms of accuracy and sample size. Despite its limitations, the dataset remains valuable for analyzing the regional context of the 3-30-300 rule.

The evaluation methods for the 3-30-300 rule also present certain limitations. For Component 3, the building perimeter was divided into 1-m intervals, and lines were drawn from these points to estimate trees within 30 m. Instances where the lines did not intersect the building were considered visible as per the methodology of an existing study (Croeser et al. 2024). This approach does not account for the actual number of floors or the presence of objects other than buildings. Furthermore, tree locations were determined using tree estimation based on high-resolution DSM and DEM (Okatani et al. 2013; Iwahashi et al. 2014; Geospatial Information Authority of Japan 2015). Owing to the difficulty of obtaining actual data, this simplified evaluation method is commonly employed. However, the discrepancy between this method and real-world conditions should be acknowledged as a methodological limitation.

For Component 30, the definition of a neighborhood remains contested: small blocks delineated by administrative boundaries and fixed-distance buffer zones surrounding each residential building. This study adopts a 300-m buffer zone around each residential building as the operational definition of a neighborhood (Nieuwenhuijsen et al. 2022; Owen et al. 2024; Lopez et al. 2025). Consequently, comparing regions without having to worry about block size or shape is now easy.

The data used in this study is cross-sectional in nature, and thus identifying causal relationships based on time series data, such as the decline in gardens or the pruning of street trees, was not possible. Further analysis using such time series data is required for prediction of trends based on current vegetation and urban planning.