Spatial distribution of urban basic education resources in Shanghai: Accessibility and supply-demand matching evaluation

Hongfu Yuan; Xiangguo Yang

doi:10.1515/geo-2022-0571

Article Open Access

Spatial distribution of urban basic education resources in Shanghai: Accessibility and supply-demand matching evaluation

Hongfu Yuan and Xiangguo Yang

Published/Copyright: December 11, 2023

Published by

Become an author with De Gruyter Brill

Submit Manuscript Author Information

From the journal Open Geosciences Volume 15 Issue 1

Abstract

This article focuses on Shanghai as a case study and utilizes various factors such as points of interest, nighttime lights, land use, road networks, and Digital Elevation Model to examine the spatial distribution of population. A random forest model is constructed to decompose the population of streets in 2022 into a 100-m grid. The study then assesses the spatial accessibility of basic education resources using a cost-weighted distance method and evaluates the supply-demand match of these resources using an improved potential model. The findings reveal the following: (1) At the street level, the spatialization of population distribution achieves a superior fit (R ² = 0.7679) with statistical data compared to the WorldPop dataset. The overall population distribution in Shanghai exhibits a spatial pattern characterized by “one main area, two sub-areas, and multiple scattered points,” effectively capturing the distribution characteristics. (2) The overall spatial accessibility of basic education resources in Shanghai is favorable, with 100% of residents able to reach the nearest primary school, junior high school, and high school within a 30-min travel time. However, significant urban–rural disparities are observed, as areas with dense facilities and well-developed transportation exhibit better accessibility. Streets with poorer accessibility tend to be concentrated in larger jurisdictional areas with abundant forests near the sea. (3) The main urban area of Shanghai and the districts of Songjiang and Fengxian demonstrate a relatively balanced supply and demand of basic education resources in several areas. However, there are still regions within these areas where resource allocation could be further strengthened.

Keywords: education; accessibility; Shanghai; spatial analysis; ArcGIS; DEM; POI

1 Introduction

Education is widely recognized as a crucial cornerstone of national development [1], with particular prominence in China, where it is regarded as a vital domain [2]. The progression of China’s education system has undergone significant transformation and development alongside the advancements of reform and opening-up policies [3]. In recent years, China has placed great emphasis on education, articulating a strategic objective of promoting the nation through science and education [4]. Government efforts have been directed toward substantial investments in education, bolstering educational infrastructure, and elevating both the quality and equity of education. Key focal points of educational reform encompass enhancing the inclusivity of education, elevating the caliber of the teaching workforce, refining pedagogical methodologies, and nurturing innovative talent.

However, as the urbanization process continues to accelerate, issues surrounding the spatial distribution of urban educational resources, alongside their accessibility and the alignment of supply and demand, have emerged as pivotal subjects within urban development [5]. Consequently, the study of educational resource spatial distribution has risen as a salient research direction in the field of education [6]. This line of research scrutinizes the geographic dispersion and spatial organization of educational resources, taking into consideration socioeconomic factors and policy influence. The accessibility and equitable allocation of educational resources constitute pivotal factors underpinning educational quality and societal advancement [7]. An effectively designed educational system should ensure that educational facilities are not only physically attainable but also readily accessible to all. Therefore, evaluating the accessibility of educational resources as well as the degree of alignment between supply and demand is of paramount importance in assessing the efficacy and fairness of the education system [8].

Within the context of educational resources, the concept of accessibility encompasses the convenience with which individuals, especially students and parents, can access educational facilities. Factors such as distance, transportation infrastructure, and travel time collectively influence the accessibility of educational resources [9]. Furthermore, issues pertaining to supply–demand mismatches involve whether existing educational resources adequately cater to the needs of the population they serve. Insufficient supply–demand alignment may result in resource overuse in some areas while wastage occurs in others, thereby giving rise to educational inequities. As pivotal elements supporting talent cultivation and societal progress, the optimized allocation of urban educational resources holds significance in enhancing residents’ quality of life and promoting social equity [10].

Shanghai, as one of China’s largest cities, boasts a high level of economic development, a dense population, and an abundance of basic educational resources. Consequently, the distribution of urban foundational educational resources and supply–demand alignment challenges in Shanghai serve as a prototypical case, offering valuable insights for other cities. Moreover, Shanghai possesses rich practical experience in urban planning, economic development, and social policies. Through an in-depth exploration of Shanghai as a representative urban case, this study aims to provide profound insights into local educational resource allocation and offer beneficial guidance for policymakers in other cities. Current research efforts have already initiated explorations into issues surrounding the spatial distribution of educational resources, thereby paving the way for further research expansion.

In existing relevant research, the concept of accessibility primarily centers on the alignment of supply and demand as well as spatial layout, serving as a critical factor in effectively assessing the spatial distribution of urban public service facilities and supply–demand equilibrium [11]. To investigate the spatial distribution and accessibility of various urban public service facilities, many scholars have adopted methodologies such as Geographic Information System (GIS) technology and the two-step floating catchment area method. These approaches facilitate the visualization of research outcomes, enriching the array of planning research methods while offering a more scientifically intuitive portrayal of research analysis processes and results. For instance, Luo [12] employed spatial syntax and GIS spatial analysis methodologies to precisely characterize urban park and sports facility spaces in Shenzhen, conducting an accessibility analysis using representative cases. The research disclosed that while Shenzhen’s urban park and sports facility spaces were abundant, overall and local accessibility was favorable, yet significant regional disparities existed. With the expansion of service coverage, accessibility increased; however, perceived accessibility remained relatively weak, impacting residents’ prompt arrivals at destinations. Similarly, Wang and Xu [13] employed GIS spatial analysis methods to comprehend the scope of educational resource influence and spatial coverage. They employed GIS to map educational resource buffer zones across varying distance ranges in Guangzhou, thereby determining the areas influenced by resources. By comparing these findings with urban population distributions, it was possible to ascertain whether resources covered crucial urban regions, consequently guiding educational resource planning and distribution.

Moreover, Ren and Wang [14] proposed an enhanced two-step floating catchment area model, conducting an accessibility analysis of urban green spaces in Huangpu District. Initially, residential community points of interest (POI) information was gathered through web scraping, aggregated within hexagonal grids of 100-m sides, and used to calculate population unit figures and density. Subsequently, Baidu Map application programming interface (API) data furnished walking times between supply and demand points. By quantifying supply–demand configurations in terms of population demand, accessibility, hotspots, and blind spots, a comprehensive analysis was undertaken. Beyond GIS spatial analysis methods and the two-step floating catchment area method, a plethora of approaches exist for studying educational resource accessibility. For instance, Guo et al. [15] delved into equality in educational resource allocation across different regions in China, focusing on accessibility evaluation. They constructed an accessibility index, incorporating factors such as time and cost, to assess resource accessibility across diverse regions. By comparing index values across regions, researchers could uncover disparities in resource accessibility, thus providing recommendations for resource optimization. Similarly, Liu and Chen [16] investigated the accessibility of educational resources, focusing on arrival times for rural compulsory education resources in China’s Jiangsu Province. They employed the Doppler model, factoring in variables like traffic speed and road segment length, to predict arrival times for resources. Through comparisons with real-time traffic conditions, disparities in resource arrival times across different regions were revealed by scholars such as Zhang et al. [17], consequently unveiling variations in resource accessibility.

In conclusion, numerous accomplishments have been achieved in research on resource accessibility to date. However, research into the analysis of educational resource accessibility has rarely incorporated extensive population statistics. To investigate the issue of educational resource allocation, the concept of population spatialization must be introduced to uncover intricate details that statistical data alone cannot unveil, operating at a finer level of analysis [18]. By employing the methodology of population spatialization to scrutinize educational resource allocation issues, limitations in acquiring high-precision data, such as residential communities and buildings, can be mitigated while enhancing the precision of research analysis on a more comprehensive scale [19]. Especially in the case of international metropolises like Shanghai, characterized by exceptionally high population density and numbers, empirical research in this domain remains scarce. Consequently, this article employs diverse data sources, including POI, nighttime light data, land use data, road networks, Digital Elevation Models (DEMs), and population data, constructing a random forest model to simulate population distribution across a 100-m grid in Shanghai. Subsequently, utilizing cost-weighted distance methodology and population spatialization outcomes, an analysis of the spatial accessibility of basic educational resources is conducted. Finally, by implementing an enhanced potential model alongside population spatialization outcomes, an assessment of the supply–demand alignment of basic educational resources is conducted. Through an analysis of school geographical locations and quantities, characteristics of spatial distribution within various regions for basic educational resources are disclosed, alongside evaluations of resident distances to the nearest school and their levels of accessibility. Concurrently, the analysis considers factors such as transportation networks, road density, and modes of transportation, exploring resident travel times and convenience levels to schools, thus evaluating educational resource accessibility. This analysis contributes to probing the supply–demand alignment of basic educational resources in Shanghai’s urban setting. By comparing school capacities and student numbers, a further assessment of the balance between supply and demand for educational resources is conducted, uncovering potential imbalances in resource allocation. Research findings provide a foundation for comprehensive evaluations of public resource allocation in Shanghai and other cities, offering technical support and guiding paths for policymakers.

2 Research area and data processing

2.1 Research area

Shanghai is located in the eastern region of China, on the west coast of the Pacific Ocean, along the eastern edge of the Asian continent. It is part of the alluvial plain of the Yangtze River Delta, situated between 120°52′–122°12′ east longitude and 30°40′–31°53′ north latitude. Shanghai covers a total area of 6,340.5 km² and is divided into 16 districts. As of the end of 2022, the permanent population of Shanghai was 24.7589 million people. In recent years, Shanghai’s education system has become more developed and is at a world-leading level. As of the end of 2021, there were a total of 64 universities in Shanghai, while the number of regular schools decreased by 62 to 867 in 2020. There were 680 primary schools, which decreased by 4 compared to 2020, and 31 special education schools, which remained the same as in 2020 (Figure 1).

2.2 Data sources

The data sources include the following:

2022 POI data: Obtained by web scraping using Python programming techniques, consisting of 632,729 records.
2022 Luojia-1 nighttime light data: Sourced from the Gaofen Hubei Center, with a resolution of 130 m.
2022 Land use data: Sourced from Tsinghua University’s FROM_GLC dataset, with a resolution of 10 m.
30-m Resolution DEM and the derived slope data: Sourced from the Geographic Spatial Data Cloud.
Basic education facility data: Derived from the POI data, filtered to include primary, junior high, and high schools. Data cleansing was performed to remove errors and duplicates, ensuring consistency with the school counts in the statistical yearbook.
Administrative division and road data: Sourced from the Shanghai Planning Bureau.
2022 Population statistics data: Representing the total permanent population without age structure characteristics, obtained from the statistical yearbook and the Shanghai Statistical Bureau.
2022 WorldPop dataset: Sourced from the official website of the WorldPop project, with a resolution of 100 m.

2.3 Data preprocessing

Table 1

Optimal bandwidth and correlation coefficient between POI density and population density

POI type	Number	Optimal bandwidth (m)	Correlation coefficient
Sports and recreation	29,202	2,200	0.879
Government and social organizations	37,939	1,300	0.941
Educational services	42,948	1,400	0.838
Transportation facilities	71,378	1,400	0.848
Food and beverage services	85,742	1,800	0.947
Lifestyle services	46,090	600	0.933
Medical services	19,722	2,100	0.958
Shopping services	97,949	2,000	0.812
Companies and enterprises	197,494	1,800	0.868
Tourist attractions	4,265	6,500	0.941

The POI data were used for derivative analysis, focusing on two derived indicators: POI density and DtN-POI (Distance to the Nearest POI). Kernel density estimation was employed with 100-m intervals and different bandwidths (ranging from 500 to 7,000 m) to obtain the POI density data that exhibited the highest correlation with population density (Table 1). The DtN-POI data represents the Euclidean distance from each location in the study area to the nearest POI. After combining the various POI data with population density data, a random forest model was trained, and the weights were calculated based on the importance of the variable factors [13]. After merging the different types of POI data, this study calculated the average POI density and DtN-POI for each town.
After mosaicking and calibration, this study calculated the annual mean value of the nighttime light data and applied denoising techniques. Subsequently, the data were resampled to a resolution of 100 m × 100 m, and the average nighttime light brightness for each town was computed. In addition, a comprehensive index method based on nighttime light and POI data was employed to extract built-up areas. After extracting the built-up areas, the Euclidean distance from each grid cell to the nearest built-up area was calculated, and the average distance from each street to the nearest built-up area was determined [16].
Furthermore, for the raster-format land use data, the proportion of each land use type within each street was calculated [12]. The calculation formula is as follows:
(1) E tk = F tk ∑ k = 1 n F tk ,
where E tk represents the proportion of the k th land-use type in street t , F tk represents the number of grid cells with the k th land-use type in street t .
For the road network data, the road network density (RND) is used as a modeling feature, using the calculation formula proposed by Luo [12]:
(2) RND = ( 3 × N r + 3 × N ne + 2 × N pe + N cr + 0.4 × N tr ) / D ,
where RND represents the road network density (km/km²); N r , N ne , N pe , N cr , and N tr represent the lengths of the roads of different categories that connect to the street; and D represents the area of the street. The mileage of each road category is converted into the length of a standard county road. The coefficients 3, 3, 2, 1, and 0.4 in the equation are the conversion factors.
Calculate the Euclidean distance from each grid within the study area to the nearest road network and compute the average distance from each street to the nearest road network. For the DEM data and slope data, calculate the average elevation and average slope for each street. Regarding the population statistics data, calculate the population density for each street by dividing the population by the area of the street [20].

3 Methodology

3.1 Population spatialization method

This study employs an advanced approach to accurately predict and distribute the population within Shanghai’s 100-m grid cells. By combining the random forest model with factors like POI density and nighttime light, it captures intricate population patterns. The 100-m grid resolution provides localized insights and resource trends. Extending predictions to grid cells enhances comprehension of resource allocation, aiding decisions for additional facilities [21]. Utilizing formula (3), the estimated values from the random forest model serve as weights to allocate the total population of each street to the grid cells. This method aims to predict street-level population density using multiple variable factors and simulate it across Shanghai’s 100-m grid using the random forest model. The use of formula (3) allows the estimated values obtained from the random forest model to be employed as weights, ensuring a more accurate allocation of the total population to each grid cell and obtaining more precise results for population spatial distribution.

(3) POP i = POP t × Q i / Q t ,

where POP i represents the population count of grid i , Q i represents the weight of the population distribution in that grid, Q t represents the sum of the population distribution weights of all grids within the same street as the grid, and POP t represents the total population of the street where the grid is located. The mean absolute error (MAE) and root mean square error (RMSE) are used to evaluate the accuracy of population spatialization. MAE reflects the actual error of the estimated values, while RMSE measures the deviation between the estimated values and the statistical values. These evaluation metrics, including MAE and RMSE, enhance the reliability of predictions. This method holds policy implications, guiding urban planners in optimal resource allocation. The integration of multiple data sources enriches the analysis of population distribution [22].

(4) MAE = 1 M ∑ | P v − O v | RMSE = 1 M ∑ ( P v − O v ) 2 ,

where M represents the number of street administrative units used for accuracy testing, P v represents the v th street, and O v represents the statistical value of the population in the v th street.

3.2 Accessibility analysis

In the framework of this study, the accessibility analysis assumes a critical role in evaluating the reach and ease of access to essential educational resources within Shanghai. This analysis method employs a sophisticated GIS-based cost-weighted approach, quantifying accessibility levels and providing vital insights for informed decision-making.

Accessibility is gauged by assigning time cost values, expressed as minutes per meter (min/m), to various land types. By integrating data on highway and rail transit speeds, as well as human walking speeds, the approach calculates time costs for distinct road levels (refer to Table 2). This methodology intricately captures transportation modes and terrain nuances in assessing accessibility [23].

Table 2

Time cost assignment for different types of land use

Land use types	Speed (km/h)	Time cost (min/m)
	100	0.0006
Expressway	60	0.0010
National road	40	0.0015
Provincial road	30	0.0020
County road	20	0.0030
Township road	35	0.0017
Rail transit	5	0.0120

Integrating GIS and the cost-weighted analysis, the study comprehensively evaluates accessibility across Shanghai’s urban fabric. By factoring in time costs from road networks and land types, the methodology unveils nuanced accessibility disparities. Utilizing formula (5) provides a tangible, adaptable approach to calculating accessibility that is tailored to Shanghai’s urban intricacies.

In the context of this research, the application of this accessibility analysis method is imperative. It enables the evaluation of educational resource accessibility, revealing discrepancies in transportation networks and facility distribution. Given education’s pivotal role in urban development, this analysis informs policymakers and urban planners in optimizing resource allocation for enhanced quality of life. Employing this method, the study addresses a crucial urban aspect, contributing to informed decision-making and sustainable urban progress.

(5) T = 1 2 ∑ i = 1 n ( B i + B i + 1 ) , horizontal direction 2 2 ∑ i = 1 n ( B i + B i + 1 ) , diagonal direction ,

where T represents the time cost, B i represents the cost value of the i th pixel, B _i + 1 represents the cost value of the ( i + 1 )th pixel in a certain direction, and n represents the total number of pixels.

3.3 Matching degree

The matching degree analysis employs a potential model, embodying the principles of proximity and convenience in school enrollment, thereby aiding in a comprehensive understanding of allocation effectiveness [24]. Recognizing the fixed nature of student school choices and the finite capacity of each institution, this method assesses the sufficiency of resource allocation within a designated service radius. It is postulated that greater proximity to educational facilities corresponds to increased educational opportunities. Hence, evaluating whether resource allocation aligns with the local population distribution becomes a pivotal aspect of urban educational planning. Thus, assessing whether the allocation share provided by basic education facilities within their service radius matches the population count of each grid unit is essential. To calculate the allocation share, an improved potential model is employed, taking into account the facility’s service radius and service capacity attenuation coefficient. The specific calculation formula is as follows:

(6) F i = ∑ j = 1 n S j R ij d ij β R ij = 1 − d ij D j β ,

where F i represents the resource allocation share for grid unit i in basic education; S j represents the service capacity of basic education facility j , which is determined based on the number of schools and enrolled students in primary, junior high, and senior high schools, with a service capacity ratio of approximately 1:2:3; R ij represents the decay coefficient that varies with distance; d ij represents the travel distance from grid unit i to basic education facility j through the transportation network; D j represents the service radius of basic education facility j ; and β represents the impedance parameter.

The matching degree of basic education resource supply and demand is the ratio between the sum of allocation shares obtained by grid units from basic education facilities and the population of the respective grid unit. It evaluates the rationality of basic education resource allocation in Shanghai from a “supply-demand matching” perspective. A high value of the matching degree indicates that the resource allocation in the grid unit is sufficient to meet the demand, while a low value indicates that the resource allocation falls short of the demand. A moderate value of the matching degree suggests a relatively balanced supply and demand in the grid unit. The formula for calculating the matching degree involves normalizing the two sets of values and is as follows:

(7) E i = N F i N P i ,

where E i represents the matching degree of basic education resource supply and demand for grid unit i , N F i represents the normalized sum of allocation shares obtained by grid unit i from various basic education facilities, and N P i represents the normalized population count of grid unit i .

4 Results

4.1 Population spatialization

Overall, the spatial distribution of Shanghai’s population in 2022 follows a spatial pattern characterized by a main center, two subcenters, and multiple scattered points (Figure 2). The population is concentrated in areas with active economic activities, convenient transportation, and well-developed infrastructure. The central urban area of Shanghai serves as the main population center, with subcenters emerging in the Songjiang and Fengxian districts on its eastern and western sides, respectively. Street-level populations are scattered in a star-like pattern throughout the city. Due to data availability, this study only utilizes population statistics at the village (community) level in a specific district for accuracy evaluation (Figure 3). At the village (community) scale, the linear regression (R ²) between the simulated population density (referred to as CSPop) and the statistical data population density reaches 0.7679 (Figure 3a), while the R ² for WorldPop’s simulation is only 0.5476 (Figure 3b), indicating a higher accuracy of the results obtained in this study.

Figure 1

An overview of Shanghai.

Figure 2

Spatial distribution of population in Shanghai city.

Figure 3

Accuracy assessment at village (community) scale: (a) CSPop and (b) WorldPop.

To gain a more detailed understanding of population spatial distribution, we selected three representative regions for detailed analysis. Region A in Figure 2 is located in the main urban area of Shanghai, primarily consisting of commercial and residential land. The small size of administrative units (typically streets) and a large population result in high population density. The boundaries between administrative units exhibit smooth transitions without noticeable jagged patterns. Region B is situated in the Songjiang district. Compared to the densely populated main urban area, Songjiang district has a shorter history of development, a larger area, and a more evenly distributed population, reflecting its actual characteristics. Region C is located in the Fengxian district. In this area, the population is concentrated in the town center, and the population density gradually decreases with increasing distance from the town center. The surrounding rural settlements have a relatively sparse population distribution, highlighting the transitional characteristics between the town and countryside and showcasing the diversity of population spatial distribution.

4.2 Spatial accessibility

The spatial accessibility of basic education facilities in Shanghai was analyzed using the cost-weighted distance method to calculate the time required to reach the nearest facility within the Shanghai metropolitan area. The results were divided into time intervals, as shown in Figure 4. The analysis reveals a spatial distribution pattern of basic education resource accessibility in Shanghai, with higher accessibility in the main urban area and lower accessibility in the suburbs. Specifically, the accessibility extends outward from the facility points along the road network. The maximum travel time is observed in the Chongming district, where it can exceed 60 min.

Figure 4

Spatial accessibility of basic education resources in Shanghai: (a) primary schools, (b) middle schools, and (c) high schools.

4.2.1 Overall accessibility of the city

In the overall accessibility analysis of the entire city, the coverage of basic education resources and the corresponding population percentages in Shanghai were calculated and summarized in Table 3. The findings are as follows: (1) Within the 0–5 min travel time range, the coverage of primary schools accounts for 51.49% of the total population, the coverage of junior high schools accounts for 54.55%, and the coverage of high schools accounts for 34.18%. (2) Within the 5–10 min travel time range, the coverage of primary schools accounts for 32.79% of the total population, the coverage of junior high schools accounts for 31.15%, and the coverage of high schools accounts for 34.41%. (3) Within the 10–20 min travel time range, the coverage of primary schools accounts for 13.64% of the total population, the coverage of junior high schools accounts for 12.30%, and the coverage of high schools accounts for 24.15%. At the same time, only 0.04, 0.03, and 0.45% of residents in Shanghai are unable to reach the nearest primary, junior high, and high schools within 60 min. Overall, the spatial accessibility of basic education resources in Shanghai is good, primarily due to the increase in the number of schools and the improvement of transportation infrastructure. Since the beginning of this century, both in urban and suburban areas, Shanghai has adhered to the principles of “fair admission” and “admission based on proximity” and has promptly constructed new schools in areas with insufficient basic education resources.

Table 3

Spatial accessibility of basic education resources in Shanghai City based on spatialization results and corresponding population size

	Primary school			Middle school			High school
Time cost (min)	Population coverage (10,000s)	Percentage (%)	Cumulative percentage (%)	Population coverage (10,000s)	Percentage (%)	Cumulative percentage (%)	Population coverage (10,000s)	Percentage (%)	Cumulative percentage (%)
0–5	148.95	51.49	51.49	157.83	54.55	54.55	98.87	34.18	34.18
5–10	94.88	32.79	84.28	90.12	31.15	85.71	99.56	34.41	68.59
10–20	39.46	13.64	97.92	35.57	12.30	98.00	69.86	24.15	92.74
20–30	4.63	1.60	99.52	4.41	1.53	99.53	13.87	4.79	97.53
30–45	0.92	0.32	99.84	1.12	0.39	99.92	4.49	1.55	99.08
45–60	0.34	0.12	99.96	0.17	0.06	99.97	1.35	0.47	99.55
>60	0.12	0.04	100.00	0.08	0.03	100.00	1.31	0.45	100.00

4.2.2 Neighborhood accessibility

The spatial accessibility values of basic education resources in each grid are aggregated and averaged according to the administrative units (streets). Combining the concept of a 60-min travel radius, the average accessibility values are divided into seven levels (Figure 5). The urban–rural disparity in the spatial accessibility of basic education resources in Shanghai is evident, with streets along national highways, provincial highways, and county roads generally having better accessibility. The relationship between the accessibility of basic education resources and the corresponding street scale is shown in Table 4. Within the city, there are 226 streets where the travel time to primary schools is less than 30 min, accounting for 98.69% of the total. There are 226 streets where the travel time to junior high schools is less than 30 min, accounting for 95.69% of the total. There are 209 streets where the travel time to high schools is less than 30 min, accounting for 91.26% of the total. These streets are mostly located in the main urban area and the central areas of the suburban districts, where accessibility gradually decreases with increasing distance in the surrounding suburbs. The main reasons for this are as follows: (1) The main urban area and central areas of the suburban districts have relatively more basic education resources, with denser facilities, which reduces the spatial distance between residents and basic education facilities. (2) The transportation network in the main urban area and central areas of the suburban districts is relatively well developed, with a dense road network, which reduces the time cost for residents to reach basic education facilities. Therefore, the accessibility of basic education resources in these regions is better.

Figure 5

Spatial accessibility of basic education resources in Shanghai: (a) primary schools, (b) middle schools, and (c) high schools.

Table 4

Spatial accessibility of basic education resources in Shanghai and corresponding street scale

	Primary school			Middle school			High school
Time cost (min)	Neighborhood number	Percentage (%)	Cumulative percentage (%)	Neighborhood number	Percentage (%)	Cumulative percentage (%)	Neighborhood Number	Percentage (%)	Cumulative percentage (%)
0–5	87	37.99	37.99	96	41.92	41.92	63	27.51	27.51
5–10	74	32.31	70.31	64	27.95	69.87	65	28.38	55.90
10–20	58	25.33	95.63	55	24.02	93.89	64	27.95	83.84
20–30	7	3.06	98.69	11	4.80	98.69	17	7.42	91.27
30–45	1	0.44	99.13	2	0.87	99.56	11	4.80	96.07
45–60	1	0.44	99.56	1	0.44	100.00	3	1.31	97.38
>60	1	0.44	100.00	0	0.00	100.00	6	2.62	100.00

Moreover, there are 3, 3, and 20 streets in the city where the travel time to primary schools, junior high schools, and high schools exceeds 30 min, respectively, with significant regional disparities, mainly concentrated in Chongming District (Figure 6). The main reasons for this are as follows: (1) Compared to the central districts, Chongming District has the largest area in the city, which leads to a sparse spatial distribution of basic education facilities and greater difficulty in resource allocation. (2) Chongming District is mostly covered by forests and is close to the sea, resulting in a sparse road network and inconvenient transportation. Residents in this area need to spend more time to reach the nearest basic education facilities. Considering the population spatialization data, the total population of these streets is calculated based on administrative divisions. The proportion of streets with a population of less than or equal to 10,000 people exceeds 100% and is concentrated in the range of 10,000–50,000 people (Table 5).

Figure 6

Number of streets in Shanghai with travel time to basic education resources >30 min.

Table 5

Total population and corresponding scale of streets in Shanghai with travel time to basic education resources >30 min

	Primary school		Middle school		High school
Population size (in 10,000 s)	Number of streets	Percentage (%)	Number of streets	Percentage (%)	Number of streets	Percentage (%)
0–0.1	1	33.33	2	66.67	2	10.00
0.1–0.5	2	66.67	1	33.33	17	85.00
>1	0	0.00	0	0.00	1	5.00

4.3 Matching degree

Figure 7 illustrates that in the central main urban area and densely populated areas such as Fengxian and Songjiang districts, the distribution of basic education resources is generally in a state of supply–demand balance. However, in suburban areas with lower population density, there are multiple high-value regions of supply–demand matching. To further investigate the internal differences and urban–rural disparities, we select the main urban area of Shanghai, the urban areas of Fengxian district, and Songjiang district as representative regions for detailed analysis.

The basic education resources in the main urban area of Shanghai are mostly in a state of supply–demand balance. The main reason is that the area has a dense road network and convenient transportation, abundant basic education resources, and a population density that generally matches the high values, which can meet the needs of residents. Due to local differences in allocation quotas and population density, the supply–demand matching of basic education resources within the area also varies (Area A in Figure 7). The main urban area is not only a densely distributed area of basic education facilities but also a high-density area in terms of population density. Therefore, the matching degree of grid units is relatively good in areas such as Huangpu District, Putuo District, Changning District, Xuhui District, Jing’an District, and Hongkou District. There are also areas with low matching degrees, mainly distributed along railway lines near train stations. Basic education facilities need to maintain a certain distance from the railway for safety reasons, resulting in smaller allocation quotas for basic education resources on both sides of the railway. At the same time, the population density in these areas is high, so the matching degree is relatively low.
There is significant spatial variation in the supply–demand matching of basic education resources in Songjiang District. The main reason is that the population distribution in the area is relatively even, but there are significant differences in the distribution of basic education facilities and road networks, resulting in significant differences in resource allocation quotas, which in turn affect the satisfaction of residents’ needs (Area B in Figure 7). The allocation quotas of basic education resources in the central area of Songjiang District are relatively consistent with the distribution of population density, resulting in a good matching degree of grid units. However, there are also areas with low matching degrees, mainly located in transitional zones along streets. These areas have an insufficient allocation of basic education resources, with smaller allocation quotas. Newly developed urban areas are expansion areas of the core urban area and can be divided into two situations. The first situation is that they are located in suburban streets but not in areas with concentrated distribution of basic education facilities, so they cannot obtain sufficient allocation quotas, resulting in a lower matching degree. The second situation is that they are located in streets undergoing urbanization, with relatively low population density and sparse distribution of basic education facilities and road networks, forming an independent ring structure with a higher matching degree.
There are multiple areas in Fengxian District with a high supply–demand matching degree of basic education resources. The main reason is that there are many scattered primary and secondary schools in the area, especially in the locations where the district government is located. However, the population density in the entire area is generally low, resulting in a star-shaped distribution of high-matching-degree areas. In rural areas, the population density is low and the road network is sparse, so the supply–demand matching degree of basic education resources is mainly determined by allocation quotas. Therefore, the matching degree presents high-value areas centered on the locations of basic education facilities, gradually decreasing with increasing distance from the facilities, forming a ring structure until it exceeds the service range or connects with the ring structure of adjacent facilities (Area C in Figure 7).

Figure 7

Supply and demand match of basic education resources in Shanghai.

In summary, the basic education resources in the main urban area of Shanghai, as well as Songjiang and Fengxian districts, are mostly in a state of supply–demand balance, but there is still a need to further strengthen resource allocation in some areas. The supply–demand matching degree of basic education resources in suburban areas is relatively high, while in urban–rural areas, the matching degree is generally high within their service radius but gradually decreases with increasing distance from the facilities. To expand the scope of student enrollment, it is necessary to strengthen the provision of school buses and the capacity to accommodate boarding students within the region. This can better meet the demand and improve the overall balance of basic education resources.

5 Discussion

The findings of this study provide critical insights into the spatial dynamics of population distribution, the spatial accessibility of basic education resources, and the alignment of resource allocation with demand. In this section, we will delve into the practical implications and potential applications of the research outcomes, emphasizing their significance for urban planning, education policy formulation, and resource allocation decisions. A detailed analysis of population spatialization revealed a distinctive urban pattern of population distribution in Shanghai. Concentrations of residents in economically active and well-connected areas underscore the intricate relationship between economic activities and population agglomeration. This understanding can guide urban planners to proactively identify areas with potential high population density, facilitate targeted development measures, and foster balanced urban growth.

The investigation into spatial accessibility unveiled disparities in the accessibility of basic education resources across different urban sectors. Notably, the discrepancies between urban and rural accessibility highlight the importance of equitable resource distribution. This insight is crucial for urban planning, guiding the layout of educational institutions and public facilities to ensure convenient access to basic education resources for all residents, regardless of their geographical location. The assessment of resource allocation matching underscores the alignment between educational supply and local demand. The equilibrium of supply and demand in central urban areas and densely populated zones reflects the effectiveness of resource allocation strategies. Conversely, the high matching degree in suburban areas underscores successful resource allocation efforts. This information is invaluable for education policymakers as it guides the optimization of resource allocation, thereby facilitating the expansion and development of educational infrastructure. The significance of this study extends to a broader realm of urban research, showcasing the utility of advanced methods in dissecting complex urban dynamics. The integration of GIS analysis, statistical modeling, and data synthesis highlights the interdisciplinary approach’s application in addressing urban challenges. This comprehensive methodology serves as a model for future research, addressing similar challenges across various urban landscapes.

In conclusion, our research offers insights into urban planning, education policy, and resource allocation decisions, which can play a pivotal role in fostering inclusive urban development and formulating effective education policies. By bridging the gap between theoretical insights and practical applications, our study lays the foundation for informed decision-making in urban development and education policy.

6 Conclusion

This study conducted a comprehensive analysis of population spatialization, spatial accessibility, and the matching degree of basic education resources in Shanghai. The findings contribute to understanding critical issues concerning resource distribution, accessibility, and alignment within the city’s education system. The conclusions derived from this study are as follows:

6.1 Population spatialization

At the street level, the linear regression (R ²) between estimated and statistical population densities reached 0.7476, surpassing the WorldPop dataset and indicating high precision. The overall population distribution in Shanghai exhibits a spatial pattern characterized by a main center, two subcenters, and scattered points. This distribution underscores the influence of economic activity and infrastructure on urban development and population density.

6.2 Spatial accessibility

The overall spatial accessibility of basic education resources in Shanghai is notable. Within 30 min, 100% of residents can access the nearest primary, junior high, and high schools. Urban–rural disparities are evident, with areas characterized by dense facilities and well-developed transportation networks exhibiting superior accessibility. In contrast, streets with poor spatial accessibility are concentrated in Chongming District due to its extensive forested areas and proximity to the sea.

6.3 Matching degree

Resource allocation strategies in Shanghai largely align with demand in the central urban area and densely populated districts. However, disparities are observed in rapidly urbanizing regions and areas with diverse road networks, reflecting the complex dynamics of urban development.

This study makes significant contributions to urban planning and educational policy research by providing a comprehensive understanding of resource allocation dynamics. Accurate population spatialization, accessibility analysis, and matching degree evaluation offer policymakers a holistic perspective for targeted resource allocation and infrastructure development, ensuring equitable education opportunities across different urban areas.

Nevertheless, several limitations must be acknowledged. The spatial accessibility assessment considers only the design speeds of various road levels, potentially introducing deviations from actual conditions. Future studies could leverage internet mapping APIs to account for real-time traffic conditions and special road types. The calculation of resource allocation shares is constrained by the availability of school enrollment data. Further data collection methods such as phone surveys and field investigations could enhance accuracy. In addition, the resource supply–demand matching discussed here is confined to intra-city comparisons and does not provide an assessment of the citywide matching level. Considering the prevalence of school district policies and cross-district enrollment, future research could incorporate district boundaries for more nuanced analyses within urban areas.

Acknowledgments

No funding was received for the present study.

Funding information: This study was supported by the Zhejiang Province’s “14th Five-Year Plan” Teaching Reform Project for Ordinary Undergraduate Colleges (jg20220667) and the Zhejiang Province’s 2022 Provincial-Level First-Class Social Practice Course: “Creative Marketing Competition” Practice.
Author contributions: Y.H. designed and carried out the experiments; Y.H. and Y.X.: formal analysis; Y.X.: investigation and data curation; Y.H.: writing original draft preparation; Y.H. and Y.X.: writing review and editing.
Conflict of interest: Authors state no conflict of interest.
Data availability statement: Data will be available upon reasonable request.

References

[1] Fägerlind I, Saha LJ. Education and national development: A comparative perspective. Oxford, UK: Pergamon Press; 2016.Search in Google Scholar

[2] Tsang MC. Education and national development in China since 1949: Oscillating policies and enduring dilemmas. China Rev; 2000;2000:579–618.Search in Google Scholar

[3] Wang X. Education in China since 1976. Jefferson, North Carolina, USA: McFarland & Co.; 2010.Search in Google Scholar

[4] De Wit H, Altbach PG. Internationalization in higher education: Global trends and recommendations for its future. In: Higher Education in the Next Decade. Leiden, The Netherlands: Brill; 2021. p. 303–25.10.1163/9789004462717_016Search in Google Scholar

[5] Lembani R, Gunter A, Breines M, Dalu MTB. The same course, different access: the digital divide between urban and rural distance education students in South Africa. J Geogr High Educ. 2020;44(1):70–84.10.1080/03098265.2019.1694876Search in Google Scholar

[6] Liang W, Ma C. Modelling the spatial distribution differences of compulsory education resource. Discret Dyn Nat Soc. 2021;2021:1–8.10.1155/2021/8342789Search in Google Scholar

[7] Willems J, Bossu C. Equity considerations for open educational resources in the glocalization of education. Distance Educ. 2012;33(2):185–99.10.1080/01587919.2012.692051Search in Google Scholar

[8] Mundy K, Dryden-Peterson S. The comparative education of refugees: Maps, myths, and possibilities. Comp Educ Rev. 2017;61(2):191–210.Search in Google Scholar

[9] Sun Y, Shen Y, Li X. A study on the accessibility of urban public facilities in a historical district: A case study of Hankou old street area in Wuhan. Sustain Cities Soc. 2020;53:101973.Search in Google Scholar

[10] Park HD, Lee SW, Kim YH. Analyzing the effects of educational investment and urbanization on regional economic growth in South Korea. Sustainability. 2021;13(12):6500.Search in Google Scholar

[11] Chen Y, Wang D. Measuring the spatial accessibility of urban public service facilities: A case study of Jinan, China. Sustainability. 2019;11(6):1616.Search in Google Scholar

[12] Luo H. Analysis of sports space accessibility in urban parks based on spatial syntax and GIS. Group Res. 2021;29(8):132–7.Search in Google Scholar

[13] Wang Y, Xu Z. Measuring and mapping spatial accessibility of educational resources: A case study in Guangzhou, China. Sustainability. 2022;14(3):732.10.3390/su15010732Search in Google Scholar

[14] Ren J, Wang Y. Analysis of spatial accessibility of parks and green spaces in Huangpu District, Shanghai, based on improved two-step floating catchment area method. Prog Geogr. 2021;40(5):774–83.10.18306/dlkxjz.2021.05.005Search in Google Scholar

[15] Guo YZ, Zhou Y, Liu YS. The inequality of educational resources and its countermeasures for rural revitalization in southwest China. J Mt Sci. 2020;17(2):304–15.10.1007/s11629-019-5664-8Search in Google Scholar

[16] Liu Y, Chen R. Evaluating the supply and demand balance of rural compulsory education resources in China: A case study of Jiangsu Province. Int J Environ Res Public Health. 2021;18(3):1133.Search in Google Scholar

[17] Zhang WH, Yuan Q, Cai H. Unravelling urban governance challenges: Objective assessment and expert insights on livability in Longgang District, Shenzhen. Ecol Indic. 2023;155:110989.10.1016/j.ecolind.2023.110989Search in Google Scholar

[18] Lechner AM, Singleton AD. A practical comparison of methods for enhancing spatial population distribution using aggregated census data. Comput Environ Urban Syst. 2019;75:126–36.Search in Google Scholar

[19] Zhang S, Liu X, Li X. Urban land use simulation and analysis based on high-resolution population data. Sustain Cities Soc. 2022;94:103752.Search in Google Scholar

[20] Fang Z, Guan D, Wu J, Liu Z. Exploring driving factors of energy-related CO2 emissions in Chinese provinces: A case of Liaoning. Appl Energy. 2017;190:172–83.Search in Google Scholar

[21] Zhang C, Gao J, Fan J, Li X. Predicting fine-scale population distribution using a combination of geospatial big data and machine learning techniques. Comput Environ Urban Syst. 2021;89:101648.Search in Google Scholar

[22] Hasanzadeh M, Poorzahedy H. Estimation of air pollution concentration using machine learning methods and dispersion modeling: A comparison study. Environ Pollut. 2020;258:113808.Search in Google Scholar

[23] Geurs KT, van Wee B. Accessibility evaluation of land-use and transport strategies: review and research directions. J Transp Geogr. 2004;12(2):127–40.10.1016/j.jtrangeo.2003.10.005Search in Google Scholar

[24] Wang Z, Li H. Spatial distribution of educational resources and housing prices in the metropolitan area: A case study of the Beijing-Tianjin-Hebei Metropolitan area, China. ISPRS Int J Geo-Inf. 2021;10(4):240.Search in Google Scholar

Received: 2023-07-01

Revised: 2023-10-11

Accepted: 2023-10-22

Published Online: 2023-12-11

This work is licensed under the Creative Commons Attribution 4.0 International License.

Articles in the same Issue

https://doi.org/10.1515/geo-2022-0571

Keywords for this article

education; accessibility; Shanghai; spatial analysis; ArcGIS; DEM; POI

Creative Commons

BY 4.0