EN中文

Journal of Geographical Sciences >

2023 , Vol. 33 >Issue 5: 1075 - 1094

DOI: https://doi.org/10.1007/s11442-023-2120-z

Research Articles

Spatiotemporal interaction pattern of the Beijing agricultural product circulation

  • ZHAO Yibo , 1, 2 ,
  • CHENG Shifen , 1, 2, * ,
  • LU Feng 1, 2, 3, 4
Expand
  • 1. State Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China
  • 2. University of Chinese Academy of Sciences, Beijing 100049, China
  • 3. Fujian Collaborative Innovation Center for Big Data Applications in Governments, Fuzhou 350003, China
  • 4. Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application, Nanjing 210023, China
*Cheng Shifen, E-mail:

Zhao Yibo, PhD Candidate, specialized in spatiotemporal data mining. E-mail:

Received date: 2022-07-27

  Accepted date: 2022-11-24

  Online published: 2023-05-11

Supported by

Innovation Project of LREIS(KPI003)

National Natural Science Foundation of China(42101423)

Strategic Priority Research Program of the Chinese Academy of Sciences(XDA23010202)

Fold

Abstract

Identifying the spatiotemporal interaction pattern of agricultural product circulation (APC) is crucial for agricultural resource adjustment and food security. Current studies are mostly based on static statistical data over an entire year or a specific period, which cannot describe the spatial pattern of APC and its seasonal variation on a fine spatiotemporal scale. Thus, this study extracts an APC trip chain based on national truck trajectory data and constructs the flow network of the Beijing APC with the city as the spatial unit and the season as the temporal unit. The spatial interaction pattern and seasonal variation in APC are then analyzed from the network spatial form, city node function role, and transportation corridors. The results are as follows: (1) Compared with methods based on static statistical data, the proposed method provides a more complete and refined depiction of the spatiotemporal interaction pattern of APC. (2) The flow network of the Beijing APC involves 316 cities in China, of which 143 cities play a major role with typical seasonal characteristics. These cities can be divided into perennial core cities, perennial major cities, core cities in winter-spring, major cities in winter-spring, core cities in summer-autumn, and major cities in summer-autumn, contributing 2.6%-40.3% to the Beijing APC. (3) There are 6 transportation corridors for the Beijing APC. The Beijing-Tianjin-Hebei corridor and coastal corridor contribute 53.5% and 12.8% of the annual supply, respectively, with a balanced supply in all seasons. The Beijing-Kunming corridor and Beijing-Guangzhou corridor contribute 14.3% and 9.0%, respectively, with much higher supplies in winter and spring. The northeast and northwest corridors contribute 7.3% and 3.3%, respectively, mainly in the summer and autumn. These results help deepen the understanding of agricultural product supply patterns and provide a reference for the design and optimization of agricultural product transportation routes.

Key words： agricultural product circulation; flow network; trajectory data mining; spatial interaction

Cite this article

ZHAO Yibo , CHENG Shifen , LU Feng . Spatiotemporal interaction pattern of the Beijing agricultural product circulation[J]. Journal of Geographical Sciences, 2023 , 33(5) : 1075 -1094 . DOI: 10.1007/s11442-023-2120-z

1 Introduction

Agricultural product trade is driven by the demand of producers and consumers, enabling the production and wide distribution of foods (Puma et al., 2015; Nagurney et al., 2019). Identifying the spatial interaction of agricultural product circulation (APC) is crucial for agricultural resource allocation optimization and food security (Jia et al., 2019). Currently, there are two main perspectives to analyze the spatial interaction of APC: (1) Agricultural product flows are estimated through optimization models based on statistical yearbook data such as grain outputs and self-sufficiency rates (Wang et al., 2019; Qian et al., 2020) and machine learning models (Lin et al., 2019; Deng et al., 2021). Based on agricultural product flows, researchers then capture the dynamic spatial pattern of APC (Sun et al., 2016; Wang et al., 2019), evaluate the stability (Sun et al., 2014; Fair et al., 2017) and safety (Ercseyravasz et al., 2012; Dalin et al., 2017) of agricultural trade, and analyze the resource, ecological and economic effects (Jiang et al., 2015; Ali et al., 2017; Mahjabin et al., 2020). (2) Based on trade statistic data such as import and export values, the agricultural product flow network is constructed through complex network theory. In this way, the circulation process of agricultural product trade can be viewed as a whole system. Researchers have also studied the spatial interaction patterns (Gephart et al., 2015; Dong et al., 2018) and evolutionary laws of the network (Wang et al., 2018; Li et al., 2021) and analyzed the determinants of APC (Wang et al., 2021; Yang et al., 2021). Spatial interaction analysis is of great significance for understanding the movement rules of individual flow components and the spatial characteristics of intercity linkages, which have become a new perspective for studying the patterns and mechanisms of geographic systems (Liu et al., 2020; Pei et al., 2020). Spatial interaction analysis has been applied in the fields of transportation (Wang et al., 2020; Chen et al., 2022), tourism (Li et al., 2020), trade (Zhao et al., 2020), urban planning (Jia et al., 2021; Kang et al., 2022; Yang et al., 2022a), and public health (Xin et al., 2022). The complex network method is an essential tool for spatial interaction analysis due to its application in quantitatively assessing complex economic and social linkages (Newman, 2003; Sartori et al., 2015). In the field of APC, analyzing the spatial interaction patterns of APC through the complex network method has received increasing attention.
However, most studies on the pattern of APC are mostly based on static statistical data for a whole year or a specific period to construct the agricultural product flow network. Due to the differences in geographical conditions and climates, the APC has strong seasonal characteristics (Guo et al., 2012). Considering seasonal differences or even monthly differences is necessary for finely portraying the spatiotemporal characteristics of APC so that the urban functional roles of APC and the temporal variation of the corridors can be identified, which is the basic information needed to understand the agricultural supply chain. The current statistical data are released over a long period with a time lag, making it difficult to capture dynamic temporal patterns. Furthermore, the static statistical data involve relatively coarse geographic units of spatial interaction, making it difficult to conduct spatial interaction analysis of APC on finer geographic units.
The intercity circulation of agricultural products is mainly realized by truck transportation. Statistics show that road freight transportation accounted for 75% of all freight transport modes in China during 20211（1 Analysis on the status quo of the freight transport industry in China in 2022 (https://www.chyxx.com/industry/1122913.html)）. With the popularization of Global Navigation Satellite Systems (GNSS), key operating trucks have achieved a full coverage of GNSS positioning devices in China. The truck trajectory data contain information such as location, time, and speed, reflecting the movement status of trucks and the transportation process of goods, which provides a new data source for the study of cargo circulation. Based on the truck trajectory data, previous studies have developed trajectory data mining methods to identify freight information (Zhu et al., 2021; Yang et al., 2022b), analyze freight activity (Gan et al., 2019; Siripirote et al., 2020), and evaluate freight emissions (Xu et al., 2021; Cheng et al., 2022). In this study, we first extracted the trip chain of APC through the national truck trajectory data in 2018 to construct the flow network of Beijing APC in four seasons. Then, we analyzed the spatial interaction pattern of the APC of Beijing from three aspects: the spatial form of networks, the functional role of cities, and the transportation corridors. This study is conducive to deepening the understanding of agricultural product circulation. Moreover, the results can provide guidance for agricultural product supply guarantees and early warnings, transportation route optimization, and related policy formulation.

2 Materials and methods

2.1 Study areas and data sources

Beijing is a national logistics hub and has many large wholesale markets for APC, such as Xinfadi Market in the Fengtai District, Jinxiudadi Market in the Haidian District, and Dayang Road General Market in the Chaoyang District. These wholesale markets are supersized agricultural product circulation centers that undertake the process of loading and unloading and serve the demand of daily agricultural product consumption in Beijing and its surrounding areas as important circulation hubs. For example, Xinfadi Market is one of the largest agricultural product wholesale markets in China, with an annual turnover of over 100 billion yuan (RMB)2（2 China agriculture information website (www.agri.cn)）.
Through the open data interface provided by the China Road Freight Vehicle Public Supervision and Service Platform (https://www.gghypt.net/), we obtained the national truck trajectory data passing through Beijing in January, April, July, and October 2018. The sampling frequency of the data was 2-30 seconds, including information such as truck ID, latitude, longitude, speed, and timestamp, with more than 74.1 billion data records (Cheng et al., 2020; Cheng et al., 2021). The data covered one busy month of agricultural trading in each season and over 300 cities in China, which can represent the APC status to some extent.

2.2 Methods

The analytical framework is shown in Figure 1. First, trip chains related to the APC were extracted. Second, flow networks of the Beijing APC in four seasons were constructed. Then, the spatiotemporal pattern of the Beijing APC was analyzed from three aspects: (1) The spatial form of the flow networks in four seasons was analyzed from the hierarchical structure and spatial distribution of the network. (2) The importance ranking of the nodes was analyzed, and the functional roles of the nodes were identified based on the importance ranking and seasonal variations. (3) The transportation corridors of APC were identified by the community detection algorithm, and the contribution of each corridor was calculated through the transportation scale in the four seasons.
Figure 1 Analytical framework for the spatiotemporal interaction pattern of the Beijing agricultural product circulation

2.2.1 Trip chain extraction

Since trajectory data lack information such as activity purpose and freight type, they cannot be directly used to build the APC flow network. In this study, the trip chain related to APC was identified through three steps: (1) identification of valid stop points; (2) construction and semantization of the trip chain; and (3) extraction of the trip chain related to agricultural products.
(1) Identification of valid stop points
A truck must first stop in order to perform any activity (loading and unloading, resting, refueling, etc.). Thus, a valid stop point contains specific semantic information representing the end of the previous trip and the beginning of the next trip. Therefore, this study considered the locations of truck activity as valid stop points. Referring to previous studies (You et al., 2019; Zhu et al., 2021), stop points were defined as when the speed was less than 1 km/h and the stop time was more than 20 minutes to eliminate situations such as traffic jams and traffic lights. Then, stop points at the same location were merged to a valid stop point through the DBSCAN algorithm (Figure 2a).
Figure 2 Trip chain extraction for agricultural product transportation
(2) Construction and semantization of trip chains
Trip chains were constructed by connecting valid stop points in chronological order. The semantization process refers to obtaining the semantic information contained in valid stop points based on the surrounding geographical elements to infer the purposes of truck activities (Gingerich et al., 2016; Sarti et al., 2017), which provides key information for the extraction of trip chains related to APC. The Amap open platform (https://lbs.amap.com/) provides information on points of interest (POIs) and areas of interest (AOIs) nationwide, such as their name, type, latitude, and longitude. Based on the POI/AOI data interface provided by Amap, we matched each valid stop point with a POI/AOI name, POI/AOI category, activity purpose, and activity content. The main steps were as follows: (1) We searched for the closest freight-related POI/AOI to the valid stop point and (2) inferred the purpose and content of the activity based on the POI/AOI category. The relationship between POI/AOI category, activity purpose, and activity content is shown in Table 1. For example, the semantic fields matched by the valid stop point shown in Figure 2b are the Xinfadi Wholesale Agricultural Market; shopping service; work; loading/unloading.
Table 1 Relationship between POI/AOI category, activity purpose, and activity content (POI, Point of Interest; AOI, Area of Interest)
POI/AOI name POI/AOI category Activity purpose Activity content
Mengrun Farm Enterprises Work Loading/Unloading
Lvpin Zhigu Industrial Park Commercial house Work Loading/Unloading
Xinfadi Agricultural Products Wholesale Market Shopping Work Loading/Unloading
Renqiu Service Area Gas Station Auto service Rest Refueling
Lingshi Service Area Road furniture Rest Dining/Resting
… … … …
(3) Extraction of trip chains related to agricultural products
Considering that the supply system of APC is “producer-primary wholesale market-secondary wholesale market-retail market”, the wholesale market is the main place for agricultural product trade (Chen et al., 2019). Trucks transporting agricultural products from other cities usually enter the agricultural product wholesale market for loading and unloading from 23:00 to 6:00 every day according to Beijing traffic restriction policy requirements. Therefore, a trip chain that passes through the Beijing agricultural product wholesale market between 23:00 and 6:00 was considered to be an agricultural product related trip chain. The extraction steps were as follows: (1) We filtered the trip chain passing through Beijing during the period of 23:00-6:00; and (2) extracted trip chains whose valid stop points contained “agricultural wholesale market” semantic information. One of the valid stop points of the trip chain shown in Figure 2b is in the Xinfadi Wholesale Agricultural Market, so it was determined to be a trip chain that was related to the Beijing APC.

2.2.2 Construction of the flow network of APC

This study constructs the flow network of APC based on a city spatial scale. The construction method is divided into two steps: (1) The valid stop point in the trip chain is merged with municipalities/prefecture-level cities or prefecture-level administrative regions (hereinafter referred to as cities) to form an origin–destination (O-D) sequence. As shown in Figure 3a, the valid stop points of the trip chain are in Baoding, Beijing, and Tianjin, forming an O-D sequence of “Baoding (O1)-Beijing (D1/O2)-Tianjin (D2)” (Figure 3b). Freight interactions between two cities were defined as the transportation process of agricultural products from the origin city to the destination city. (2) Using all cities as nodes, an undirected weighted network G=(V,E,W) is constructed based on the O-D sequences. $V=\left\{ {{v}_{\text{i}}}:i=1,2,\ldots,n \right\}$is the node set, where n is the number of nodes, representing the number of cities in the network.$E=\left\{ {{e}_{ij}}:\text{i},\text{j}=1,2,\ldots,\text{n} \right\}$is the set of all edges linking pairs of nodes in V, where eij represents whether there is a freight connection between city i and city j. If so, eij = 1; otherwise, eij = 0. $W=\left\{ {{w}_{ij}},\text{i},\text{j}=1,2,\ldots,\text{n} \right\}$is the connection strength set, where wij is the edge weight representing the number of trips between city i and city j.
Figure 3 O-D sequence construction from valid stop point sequences

2.2.3 City functional role division

Weighted degree centrality ($WD{{C}_{i}}$) is calculated based on the degree and strength of node i (Jiao et al., 2017), revealing the importance of cities in the agricultural product flow network. The formula is as follows:
$WD{{C}_{i}}={{D}_{i}}\text{*}{{\left( \frac{{{S}_{i}}}{{{D}_{i}}} \right)}^{1-\alpha }}$
where Di is the degree of node i, representing the influence of node i in the network. Si is the strength of node i, representing the activity of node i in the network. α is the tuning parameter, representing the difference in the effects of node degree and strength for node i in the network. In this study, the value of α is 0.5. The higher the$WD{{C}_{i}}$value is, the greater the importance of node i in a network. Di and Si are calculated as follows:
${{D}_{i}}=\underset{j\in N\left( i \right)}{\mathop \sum }\,{{e}_{ij}}$
${{S}_{i}}=\underset{j\in N\left( i \right)}{\mathop \sum }\,{{w}_{ij}}$
where N(i) is the neighbor node set of node$~i$.
The coefficient of variation (CVi) is a measure of dataset dispersion, which is used to evaluate the stability of WDC in four seasons in this study. From the value of CVi, we can assess whether the city has seasonal variations in the flow network of APC. The formula is as follows:
$C{{V}_{i}}=\frac{{{\sigma }_{i}}}{{{\mu }_{i}}}$
where μi is the average WDCi of node i in four seasons. σi is the standard deviation. Generally, the dataset tends to be discrete when CVi > 0.15. Therefore, 0.15 is regarded as the boundary of CVi in this study. When CVi > 0.15, the importance of city i is unstable in the four seasons, and city i has seasonal variations.

2.2.4 Measurement of network hierarchy

We analyzed the hierarchical structure of the network based on the rank-size rule (Zipf, 1949), with WDCi as the criterion. The rank-size rule can be formulated as follows:
${{p}_{i}}={{p}_{1}}*{{r}_{i}}^{-q}$
Taking the natural logarithm of both sides of Equation (5), the unitary linear regression equation can be obtained, as shown in Equation (6):
$\ln {{p}_{i}}=\ln {{p}_{1}}-q\ln {{r}_{i}}$
where pi represents the importance of city i. ri is the rank of city i. q is the Zipf index, representing the hierarchy of node importance. When q=1, the network has a hierarchical structure, and the number of cities at each level satisfies the fixed ratio suggested by the rank-size rule. When q>1, the importance of cities differs widely, and the importance structure of cities presents a primary distribution, indicating that cities with high importance have a great influence in the network. When q<1, the importance structure of cities shows a log-normal distribution.

2.2.5 Transportation corridor identification

The transportation corridor connects cities, ensuring the efficiency of goods transportation. The city connection in the corridor is relatively close and relatively sparse between corridors. Similarly, the subnetworks generated by complex network community detection are relatively independent and closely connected, characterizing the local agglomeration features in the network (Newman et al., 2004). Therefore, the Louvain community detection algorithm (Blondel et al., 2008) is used to identify the transportation corridors for APC. The closeness of the nodes in the corridor is calculated by the modularity Q (Newman et al., 2004). The formula is as follows:
$Q=\frac{1}{2m}\underset{ij}{\mathop \sum }\,\left[ {{w}_{ij}}-\frac{{{S}_{i}}{{S}_{j}}}{2m} \right]{{\text{ }\!\!\delta\!\!\text{ }}_{ij}}$
where δij represents whether city nodes i and j are in the same community. If so, δij = 1; otherwise, δij = 0. m represents the sum of the connection strengths of the entire network. The higher the Q value is, the better the quality of the community division ($Q\in \left[ -1,1 \right]$).

3 Results

3.1 Spatial form of the flow network

The flow networks of the Beijing APC in the four seasons are shown in Figure 4 and Table 2. The freight scale is represented by the sum of the weights of the edges in the network. The number of cities and freight scale of the network are relatively low in winter due to snowfall and foggy weather. Excluding winter, the statistical characteristics of the flow networks in the other three seasons are similar. There are 316 cities in the flow networks in the four seasons. WDC and weight of edge are used to measure the importance of cities and the connection strength between cities, which are normalized. The hierarchical structure of the networks is fitted using the rank-size rule. The values of goodness of fit in the four seasons are higher than 0.8, indicating a significant hierarchy in the flow network of the Beijing APC. The q values in the four seasons are greater than 1, indicating that the size rank of city importance has a primary distribution. Areas around Beijing, such as Tianjin and Baoding, are of high importance. Nearly 316 cities participate in the APC system, while a few cities play major roles. These cities are mostly located in the peripheral areas of Beijing, which are affected by distance decay. In addition, the average CV in the four seasons is over 1.20, indicating that there are seasonal variations in the flow networks of the Beijing APC.
Figure 4 Flow networks of Beijing agricultural product circulation in four seasons
Table 2 Statistical characteristics of flow networks of Beijing agricultural product circulation
Season Number of cities Freight scale Average WDC Maximum WDC Average CV Zipf Goodness of fit (R2)
Spring 305 6945 20.50 825.23 1.20 1.0927 0.833
Summer 295 5587 21.83 845.29 1.25 1.1017 0.863
Autumn 284 6541 23.97 904.85 1.24 1.2082 0.888
Winter 290 4460 18.75 742.53 1.25 1.0758 0.868
The spatial patterns of the flow networks of the Beijing APC in the four seasons are dense in the east and sparse in the west, with high values of the two indicators radiating around Beijing. For seasonal variations, the high values of the two indicators radiate to the south and southwest regions in spring and winter and to the north and northeast in summer and autumn.

3.2 City functional role

3.2.1 Rank division and spatial distribution

The importance of cities in the flow networks of the Beijing APC are divided into five ranks. Among them, the highest rank contains the primary city (Beijing). The lowest-ranked cities appear less frequently in the APC system. Apart from these two ranks, the other 143 cities constitute three ranks of importance. The spatial distribution of cities with different ranks in the four seasons is shown in Figure 5. The highest-ranked cities, which are in the Beijing-Tianjin-Hebei (BTH) region and are represented by Langfang, Tianjin, and Baoding, account for 1.7% of the network. From the perspective of seasonal variation, Cangzhou and Shijiazhuang are included in winter and spring. Zhangjiakou is included in summer and autumn. The second-highest ranking cities are mainly located in Hebei, Shandong, and Liaoning Provinces. Typical cities include Tangshan, Chengde, Handan, Hengshui, and Xingtai in Hebei Province; Weifang, Linyi, and Yantai in Shandong Province; Shenyang and Huludao in Liaoning Province; and a few provincial capitals in central and southern China, such as Guangzhou and Chengdu. Furthermore, there are obvious differences between the four seasons. The second-highest ranking cities account for 7.3% of the network, which are in summer and autumn and mainly in the northeastern region, such as Jinzhou, Anshan, and Changchun. The spatial distribution of the second-highest ranking cities extends northeastward to Harbin in summer and northward to Xilin Gol League and Chifeng in autumn. The second-highest ranking cities in winter and spring are mainly in the provincial capitals in the central, southern, and southwestern regions, such as Xi’an, Wuhan, Kunming, and Zhanjiang. The third-highest ranking cities in the four seasons are roughly the same, accounting for 23.3% of the network, and are mainly dispersed near first- and second-highest ranking cities, such as Chongqing, Xiangyang, Ganzhou, Foshan, Shanghai, and Suzhou.
Figure 5 Spatial distribution of city ranks for Beijing agricultural product flow in four seasons
The importance of cities in flow networks presents a significant spatial agglomeration effect. The global Moran’s I value is positive, and the Z score is higher than 2.58 (Table 3), passing the significance test at a 1% level. Specifically, the distribution of cities with high values in summer and autumn is mainly concentrated in the northeast region, while the cities in winter and spring are scattered in the Yangtze River basin, southwest region, and southern region. Thus, the values of Moran’s I in winter and spring are slightly lower than those in summer and autumn. This result can be explained by the climate. In northern China, the mild climate in summer and autumn is suitable for growing agricultural products, so Beijing and its surrounding areas can be self-sufficient. In contrast, the cold climate in winter and spring results in less production of agricultural products in the north, requiring a supply from warmer areas in the south. Therefore, the cities with high values in spring and winter are scattered in the south.
Table 3 Global Moran’s I of the spatial distribution of city ranks for Beijing agricultural product flow in four seasons
Season Global Moran’s I Z score
Spring 0.212 20.054
Summer 0.226 18.338
Autumn 0.255 18.898
Winter 0.185 16.318

3.2.2 City functional role division

The functional roles of 143 cities are determined by their importance and seasonal characteristics. As shown in Figure 6a, the importance of cities is divided into two categories based on the annual average WDC of the cities: core cities (annual average WDC > 0.1) and major cities (annual average WDC < 0.1). Furthermore, the seasonal characteristics are distinguished based on the CV of the cities. For cities with a high CV (CV > 0.15), the average WDC in summer-autumn and winter-spring are calculated, with the former being higher for summer-autumn cities and the latter being higher for winter-spring cities. Consequently, the functional roles of cities in the Beijing agricultural product flow network are divided into six categories: perennial core cities, perennial major cities, core cities in winter-spring, major cities in winter-spring, core cities in summer-autumn, and major cities in summer-autumn. Figure 6b shows the spatial distribution of the functional role of each city. Table 4 shows the contribution of the functional role of each city to the Beijing APC based on the node strength.
Figure 6 Division criteria and results of city functional role (SW: spring and Winter, SA: spring and autumn)
Table 4 Seasonal contributions of cities to the Beijing agricultural product circulation
City role Perennial core cities Perennial major cities Core cities in winter-spring Major cities in winter-spring Core cities in summer-autumn Major cities in summer-autumn
Number of cities 6 56 6 56 1 18
Proportion 40.27% 22.82% 6.65% 13.34% 2.62% 5.17%
Contribution per city 5.75% 0.41% 1.11% 0.24% 2.62% 0.29%
Perennial core cities and perennial major cities serving the Beijing APC are mainly located in the Bohai Rim region and the Huang-Huai-Hai region, with provinces including Hebei, Liaoning, Shandong, and Inner Mongolia. These cities take advantage of their distance from Beijing, ensuring the year-round agricultural product demand of Beijing and its surrounding areas. A total of 6 perennial core cities, including Tianjin, Baoding, Tangshan, Langfang, Cangzhou, and Guangzhou, are predominant in the flow network of the Beijing APC year-round. Except for Guangzhou, all these cities are from the BTH region, showing a significant spatial agglomeration. The number of perennial core cities accounts for 5.2% of the 143 cities in total, while the contribution of these perennial core cities to the circulation system is 40.3%, revealing the dominance of these cities in the network. The 56 perennial major cities, including Linyi, Dezhou Jinan, Rizhao, Shenyang, Hengshui, Huludao, Qinhuangdao, and Chifeng, are less important than the perennial core cities. These cities are mostly in Northeast China and Shandong Province, revealing that the distribution of agricultural products is partly influenced by distance. The contribution of these cities to the APC system is 22.8%. In addition, perennial major cities contain provincial capitals in the eastern region and some transportation hub cities, such as Wuhan, Xi’an, Taiyuan, Yueyang, and Foshan. These cities are indispensable hubs in the flow network of the Beijing APC.
For the winter-spring cities serving the Beijing APC, core cities in winter-spring include 6 cities, namely, Shijiazhuang, Handan, Xingtai, Weifang, Zhengzhou, and Chengdu. Major cities in winter-spring include Kunming, Nanning, Zhanjiang, Shenzhen, Haikou, Changsha, and Hangzhou, with a total of 56 cities. These winter-spring cities contribute 20.0% to the Beijing APC. Spatially, the winter-spring cities are mainly located in the Yangtze River basin and southern region, where the climate is warm in winter and spring. These winter-spring cities have hub roles in regional agricultural product transportation.
For the summer-autumn cities serving the Beijing APC, the only core city in summer-autumn is Zhangjiakou in Hebei Province. Major cities in summer-autumn include Changchun, Songyuan, Anshan, Tieling, Xilin Gol League, and Urumqi, with a total of 18 cities. Summer-autumn cities contribute approximately 7.8% to the Beijing APC. Spatially, summer-autumn cities are mainly located in the northeast and northwest regions, where the climate is moderate in summer and autumn. Summer-autumn cities have important contributions to the Beijing APC in winter and spring as the key hubs for the northeast and northwest regions to transport agricultural products to Beijing.

3.3 Transportation corridor

An agricultural transportation corridor is a cross-regional, long-distance corridor for goods circulation, which can reflect the transportation routes and seasonal variations of other regions supplying agricultural products to Beijing. We construct a year-round flow network of the Beijing APC by integrating data from four seasons and identify the transportation corridors of the Beijing APC through the Louvain community detection algorithm. The result is shown in Figure 7. In this study, the transportation corridors of the Beijing APC are named the Northeast Corridor, Beijing-Tianjin-Hebei (BTH) Corridor, Coastal Corridor, Beijing-Guangzhou (BG) Corridor, Beijing-Kunming (BK) Corridor and Northwest Corridor. The main cities of each corridor are shown in Table 5. Using the connection strength between cities in the corridor as the transportation scale, we analyze the contribution of the corridors to the Beijing APC, as shown in Figure 8.
Figure 7 Spatial distribution of the Beijing agricultural product transportation corridor
Table 5 Main cities of each transportation corridor
Corridor Perennial core/major cities Core/major cities in
spring-winter		 Core/major cities in
summer-winter
BG corridor Guangzhou, Wuhan, Dongguan, Puyang, Chongzuo, Foshan, Yueyang, Hebi, Yichang Handan, Zhengzhou, Nanning, Zhanjiang, Haikou, Shenzhen, Changsha, Jingzhou, Kaifeng, Anyang, Guilin, Xinxiang, Nanyang, etc. /
Northeast corridor Shenyang, Huludao, Jinzhou, Dalian, Harbin / Changchun, Anshan, Yingkou, Songyuan, Tieling, Suihua, Siping, Tongliao, Liaoyang, Jilin
BK corridor Xi’an, Yuncheng, Weinan, Chongqing, Yuxi, Lanzhou, Xianyang, Linfen, Ziyang Chengdu, Kunming, Guiyang, Honghe Hani and Yi Autonomous Prefecture, Meishan, Hanzhong, Changde, Sanmenxia, Qujing, Xishuangbanna Dai Autonomous Prefecture, Chuxiong Yi Autonomous Prefecture, Ya’an, Dali Bai Autonomous Prefecture, Dehong Dai and Jingpo Autonomous Prefecture, Baoshan /
BTH corridor Langfang, Tianjin, Baoding, Cangzhou, Tangshan, Hengshui, Qinhuangdao, Dezhou, Jinan, Chifeng, Ulanqab, Hohhot, Baotou, Chaoyang, Datong Shijiazhuang, Xingtai Zhangjiakou, Chengde, Xilingol League, Shuozhou
Coastal corridor Linyi, Xuzhou, Qingdao, Shanghai, Binzhou, Jining, Zhangzhou, Suzhou, Shangqiu, Suqian, Zaozhuang, Suzhou, Nanjing, Jinhua, Weihai, Jiaxing, Lianyungang, Ningbo, Fuzhou, Rizhao Weifang, Liaocheng, Yantai, Heze, Hangzhou, Tai’an, Ganzhou, Hefei, Dongying, Wuxi, Fuzhou, Wenzhou, Quanzhou, Taizhou, Huai’an, Nanchang, Jiujiang /
Northwest corridor Taiyuan, Ordos, Jinzhong, Bayannur / Urumqi, Luliang, Changji Hui Autonomous Prefecture, Hami
Figure 8 Proportion of transportation scale of each corridor in each season
The BTH corridor contributes approximately 53.5% to the Beijing APC, which is much higher than the other corridors, indicating that the agricultural product supply in Beijing and its surrounding areas can be basically self-sufficient. The self-sufficiency rate of agricultural products in summer and autumn exceeds 55.0% and is slightly higher than that in winter and spring, which is because the northern climate in summer and autumn is more suitable for producing crops.
The BG corridor contributes approximately 14.3% to the Beijing APC. The transportation scale in winter and spring is twice that in summer and autumn. Spatially, the corridor extends north to Hebei Province and south to Hainan Province, passing through the Central Plains, the middle reaches of the Yangtze River and the Pearl River Delta, containing cities with various climate types. In particular, benefiting from the warm climate in the Yangtze River Basin and South China, the BG corridor replenishes Beijing with a large amount of agricultural products in winter and spring.
The coastal corridor contributes approximately 12.8% to the Beijing APC, and the supply is relatively balanced over the four seasons. Cities in the coastal corridor are mainly in Shandong and the Yangtze River Delta, which have adequate sunlight and hot temperatures in all seasons, thus producing various agricultural products. In addition, as the most densely populated and industrial area in China, the coastal area has convenient transportation and a complete infrastructure, resulting in flourishing freight activity in the coastal corridor.
The BK corridor contributes approximately 9.0% to the Beijing APC. The corridor passes through central China, reaching the Inner Mongolia Autonomous Region in the north and Yunnan Province in the south. The warm climate in Southwest China is advantageous in winter and spring. Conversely, the cool climate in the Yunnan-Guizhou Plateau and Loess Plateau is advantageous in summer and autumn. The transportation scale in winter and spring is approximately 1.57 times that in summer and autumn.
The northeast and northwest corridors contribute approximately 7.3% and 3.3%, respectively, which are lower than those of the other corridors. Both corridors are in the northern part of China at high latitudes. The climate is suitable for growing crops in summer and autumn. Thus, the contribution in summer and autumn is approximately twice as high as that in winter and spring.

4 Discussion

4.1 Trajectory data mining provides more detailed insight into agricultural product circulation between regions

Many studies have focused on the flow network of APC for agricultural supply security and food security based on static statistical data (Wang et al., 2018; Lin et al., 2019). For example, Wang et al. (2021) constructed a food trade network by using countries as the nodes and the trade between two countries as the edge based on international food trade data from 1994 to 2016. Ben et al. (2016) simulated the provincial agricultural trade by the linear optimization model based on the grain supply and demand data of each province from 2010 to 2012. Because static statistical data are mainly at provincial and national scales and the publication cycle is long, these studies usually lag by 2-4 years and have a low spatial resolution, which makes it challenging to analyze the interaction patterns of agricultural trade at finer spatiotemporal scales.
Unlike previous studies, we provided a more detailed means for building an APC flow network. With the advantage of the high accuracy of national truck trajectory data on the spatiotemporal scale and reference to agricultural product transportation characteristics, we extracted the trip chain of APC through the identification of valid stop points, construction and semantization of trip chain, and construction of a fine-scale agricultural product flow network. The spatiotemporal resolution can be dynamically adjusted according to the demand, which makes up for the shortcomings of existing studies. The results show that the APC flow network based on this method considers the temporal variations in APC, which provides support for the study of the spatiotemporal interaction pattern of APC. Furthermore, the proposed method of extraction of agricultural transportation trip chains can be applied to the study of agricultural transport behavior patterns and agricultural transport demand prediction, which is valuable for the improvement of agricultural product transportation efficiency, agricultural product supply security and early warnings. With the rapid development of mobile localization technology and the gradual popularization of vehicle trajectory data, the proposed method can be applied to other cities in China.

4.2 Seasonal variation patterns of Beijing agricultural product circulation from a network perspective

Unlike existing studies based on an entire year or period, we construct the flow networks of the Beijing APC in four seasons. Thereafter, we analyze the spatiotemporal interaction pattern of the Beijing APC through three aspects: the spatial form of the networks, the network form, the functional role of cities, and the transportation corridors. Specifically, the network form demonstrates the spatial pattern and statistical characteristics of the flow network. The functional role of cities demonstrates the contribution of other cities to the Beijing APC. Transportation corridors portray the transportation routes of agricultural products supplied to Beijing from other cities, reflecting the seasonal variations in the agricultural supply process.
The flow networks in the four seasons involve 316 cities in total. The spatial pattern of the Beijing APC is dense in the east and sparse in the west, with seasonal variations. Cities of high importance are mainly distributed in the north and northeast in summer and autumn, whereas they are distributed in the south and southwest in winter and spring. Furthermore, the flow network of the Beijing APC has a clear hierarchical structure, with a primary distribution. There are 143 cities mainly serving the Beijing APC. The contribution of each functional role in the network ranges from 2.6% to 40.3%. Specifically, perennial core cities and perennial major cities are mainly from the area around Beijing and the provincial capitals of some developed provinces. Winter-spring cities are mainly developed cities in the southern BTH region, with a scattered distribution, while summer-autumn cities are mainly located in the northern BTH region, with a concentrated distribution. There are 6 transportation corridors in the Beijing APC. The BTH and coastal corridors contribute 53.5% and 12.8%, respectively, to the Beijing APC throughout the year, with a balanced supply in all seasons. The BG and BK corridors contribute 14.3% and 9.0%, respectively, with much higher supplies in winter and spring. The northeast and northwest corridors contribute 14.3% and 9.0%, respectively, mainly in the summer and autumn.
Furthermore, compared with the circulation nodes and national backbone circulation corridors in the National Circulation Node City Layout Plan (2015-2020) (hereafter referred to as the Plan)3（3 Ministry of Commerce, PRC. the National Circulation Node City Layout Plan (2015-2020) (http://www.mofcom.gov.cn/article/ae/ai/201506/20150600998472.shtml)）, approximately 54.5% of the cities in the flow network of the Beijing APC are circulation nodes at or above the regional level. The BK corridor, BG corridor and northwest corridor in the flow network of the Beijing APC are basically consistent with the corridor layout in the Plan. However, due to the particularity of the APC, other corridors are different from those in the Plan.
The spatial distribution of city functional roles and the seasonal variations in the supply of agricultural products through the corridors can be interpreted by the geographical location and climate. First, the BTH region is the main consumer of the Beijing APC and the main producer of agricultural products, which is consistent with the findings of Li et al. (2015). Second, compared with the climate in summer and autumn, the climate in winter and spring in northern China is cool, making it unsuitable for growing agricultural products. Therefore, Beijing and its surrounding areas need more external supplementation in winter and spring. Coastal cities play important roles in the Beijing APC throughout the year due to their high productivity levels and abundant light and heat resources. Third, South China and Southwest China are warm in winter and spring, providing Beijing with warm-weather agricultural products, such as watermelon and tomato. The Yangtze River basin, with its mild climate, provides Beijing with cool-weather agricultural products, such as cabbage, celery, and lettuce. Therefore, the winter and spring cities of the Beijing APC are located in the south and southwest regions, constituting the BK and BG corridors. In summer and autumn, the cool climate of the northern high latitudes allows for the cultivation of basic agricultural products, such as peppers, cucumbers, and onions, and specialty agricultural products, such as grapes and melons, without requiring shade and cooling. Therefore, the summer and autumn cities of the Beijing APC are mainly located in the northern region, constituting the northwest corridor and the northeast corridor.
In addition, the results have practical implications for governmental agricultural production policies, enterprise investments, and agricultural product transportation scheduling. For example, the functional role of cities in the flow network of the Beijing APC provide a reference for the improvement of the national backbone agricultural logistics network, and the transportation corridors provide a reference for the design and optimization of agricultural transport routes.

4.3 Limitations and future perspectives

Although reliable trip chain extraction rules for APC have been designed in this research, the lack of specific information on the agricultural products transported and the scale of loading and unloading leads to some limitations in the study. Future research will combine multiple data sources (survey data, enterprise type data, land use data, etc.) to improve the accuracy of trip chain extraction. Moreover, with the analytical framework proposed in this study, some comparative analysis on the spatiotemporal characteristics of agricultural trade for different regions will be conducted. Furthermore, the evolutionary patterns of the national agricultural flow network will be investigated to support the improvement of the capacity and efficiency of cross-regional agricultural product circulation in China.

5 Conclusions

This study extracts the trip chains related to the Beijing APC based on national truck trajectory data to construct the flow network of the Beijing APC. Thereafter, the spatiotemporal interaction pattern of the Beijing APC is analyzed from three aspects: the spatial form of the flow networks, the functional role of city nodes, and the transportation corridors. The study conclusions are as follows:
(1) Compared with methods based on static statistical data, the agricultural product flow network constructed by the massive truck trajectory data more completely and finely depicts the spatiotemporal interaction process of APC.
(2) Beijing APC has formed a backbone network, with cities surrounding Beijing such as Tianjin, Baoding, and Langfang as the core cities (40.3%) and the capital cities of developed provinces connected as the major cities (22.8%). The network shows obvious seasonal variations, and cities with seasonal characteristics contribute 27.8% of the flow in the Beijing APC. The spatial pattern of the network is dense in the east and sparse in the west, with a clear hierarchical structure. The transportation distance is the main factor affecting the contribution scale of regions in the APC.
(3) There are 6 transportation corridors to ensure the Beijing APC in different seasons. The BTH corridor and coastal corridor contribute 53.5% and 12.8%, respectively, to the Beijing APC, which are balanced across the four seasons. The BG corridor and BK corridor contribute 14.2% and 12.8% to the Beijing APC, respectively, with a more critical role in supplementing Beijing agricultural products in winter and spring. The Northeast and Northwest corridors mainly function in summer and autumn, contributing 7.3% and 3.3%, respectively, to the Beijing APC.
These results can provide a reference for the construction and improvement of the national backbone agricultural product logistics network and the design and optimization of agricultural product transportation routes. Future research can be conducted on the spatial pattern and evolution characteristics of the agricultural product flow network at the national scale and on the horizontal comparison of the spatiotemporal characteristics of the agricultural product flow network in megacities of different regions to improve the capacity and efficiency of the cross-regional circulation of agricultural products in China.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
Ali T, Huang J, Wang J et al., 2017. Global footprints of water and land resources through China’s food trade. Global Food Security, 12: 139-145.

DOI

[2]
Ben P, Wu S, Li X et al., 2016. China’s inter-provincial grain trade and its virtual cultivated land flow simulation. Geographical Research, 35(8): 1447-1456. (in Chinese)

[3]
Blondel V D, Guillaume J-L, Lambiotte R et al., 2008. Fast unfolding of communities in large networks. Journal of Statistical Mechanics: Theory and Experiment, 2008: P10008.

DOI

[4]
Chen X, Zhang Z, Zhang X, 2019. Study on the upgrading model and path of China’s agricultural product wholesale market: The experiences from five common operation principles of world agricultural product wholesale market. China Business and Market, 33(2): 30-37. (in Chinese)

[5]
Chen Z, Wang J, Li Y, 2022. Intercity connections by expressway in metropolitan areas: Passenger vs. cargo flow. Journal of Transport Geography, 98: 103272.

DOI

[6]
Cheng S, Lu F, Peng P, 2020. A high-resolution emissions inventory and its spatiotemporal pattern variations for heavy-duty diesel trucks in Beijing, China. Journal of Cleaner Production, 250: 119445.

DOI

[7]
Cheng S, Lu F, Peng P et al., 2021. Emission characteristics and control scenario analysis of VOCs from heavy-duty diesel trucks. Journal of Environmental Management, 293: 112915.

DOI

[8]
Cheng S, Zhang B, Zhao Y et al., 2022. Multiscale spatiotemporal variations of NOx emissions from heavy duty diesel trucks in the Beijing-Tianjin-Hebei region. Science of the Total Environment: 854: 158753.

DOI

[9]
Dalin C, Wada Y, Kastner T et al., 2017. Groundwater depletion embedded in international food trade. Nature, 543(7647): 700-704.

DOI

[10]
Deng X, Cao S, Horn A L, 2021. Emerging applications of machine learning in food safety. Annual Review of Food Science and Technology, 12(1): 513-538.

DOI

[11]
Dong C, Yin Q, Lane K J et al., 2018. Competition and transmission evolution of global food trade: A case study of wheat. Physica A: Statistical Mechanics and Its Applications, 509: 998-1008.

DOI

[12]
Ercseyravasz M, Toroczkai Z, Lakner Z et al., 2012. Complexity of the international agro-food trade network and its impact on food safety. PLoS ONE, 7(5): e37810.

[13]
Fair K R, Bauch C T, Anand M, 2017. Dynamics of the global wheat trade network and resilience to shocks. Scientific Reports, 7(1): 1-14.

DOI

[14]
Gan M, Nie Y M, Liu X et al., 2019. Whereabouts of truckers: An empirical study of predictability. Transportation Research Part C: Emerging Technologies, 104: 184-195.

DOI

[15]
Gephart J A, Pace M L, 2015. Structure and evolution of the global seafood trade network. Environmental Research Letters, 10(12): 125014.

DOI

[16]
Gingerich K, Maoh H, Anderson W, 2016. Classifying the purpose of stopped truck events: An application of entropy to GPS data. Transportation Research Part C: Emerging Technologies, 64: 17-27.

DOI

[17]
Guo H, Cai J, Wang D, 2012. Analysis of tempo-spatial patterns of Beijing external vegetable supply and its effects under massive logistical system. Economic Geography, 32(3): 96-101. (in Chinese)

[18]
Jia P, Liu A, Cheng S et al., 2019. Pattern changes of China’s agricultural trade and countermeasures for the utilization of overseas agricultural resources. Journal of Natural Resources, 34(7): 1357-1364. (in Chinese)

DOI

[19]
Jia T, Yu X, Li X et al., 2021. Identification and analysis of urban influential regions using spatial interaction networks. Transactions in GIS, 25(6): 2821-2839.

DOI

[20]
Jiang Y, Cai W, Du P, et al., 2015. Virtual water in interprovincial trade with implications for China’s water policy. Journal of Cleaner Production, 87: 655-665.

DOI

[21]
Jiao J, Wang J, Jin F, 2017. Impacts of high-speed rail lines on the city network in China. Journal of Transport Geography, 60: 257-266.

DOI

[22]
Kang C, Jiang Z, Liu Y, 2022. Measuring hub locations in time-evolving spatial interaction networks based on explicit spatiotemporal coupling and group centrality. International Journal of Geographical Information Science, 36(2): 360-381.

DOI

[23]
Li T, Liu X, Wang R et al., 2021. Dynamics of global pork trade networks during 2000-2019 and its implications for China: A complex network analysis. Journal of Natural Resources, 36(6): 1557-1572. (in Chinese)

[24]
Li T, Wang J, Huang J et al., 2020. Exploring temporal heterogeneity in an intercity travel network: A comparative study between weekdays and holidays in China. Journal of Geographical Sciences, 30(12): 1943-1962.

DOI

[25]
Lin X, Ruess P J, Marston L et al., 2019. Food flows between counties in the United States. Environmental Research Letters, 14: 084011.

DOI

[26]
Liu Y, Yao X, Gong Y et al., 2020. Analytical methods and applications of spatial interactions in the era of big data. Acta Geographica Sinica, 75(7): 1523-1538. (in Chinese)

DOI

[27]
Mahjabin T, Mejia A, Blumsack S et al., 2020. Integrating embedded resources and network analysis to understand food-energy-water nexus in the US. Science of The Total Environment, 709: 136153.

[28]
Nagurney A, Besik D, Nagurney L S, 2019. Global supply chain networks and tariff rate quotas: Equilibrium analysis with application to agricultural products. Journal of Global Optimization, 75(2): 439-460.

DOI

[29]
Newman M E, 2003. The structure and function of complex networks. SIAM Review, 45(2): 167-256.

DOI

[30]
Newman M E, Girvan M, 2004. Finding and evaluating community structure in networks. Physical Review E, 69: 026113.

DOI

[31]
Pei T, Shu H, Guo S, 2020. The concept and classification of spatial patterns of geographical flow. Journal of Geo-Information Science, 22(1): 30-40. (in Chinese)

[32]
Puma M J, Bose S, Chon S Y et al., 2015. Assessing the evolving fragility of the global food system. Environmental Research Letters, 10(2): e1002963.

[33]
Qian H, Engel B A, Tian X et al., 2020. Evaluating drivers and flow patterns of inter-provincial grain virtual water trade in China. Science of the Total Environment, 732: 139251.

DOI

[34]
Sarti L, Bravi L, Sambo F et al., 2017. Stop purpose classification from GPS data of commercial vehicle fleets. 2017 IEEE International Conference on Data Mining Workshops. IEEE, 2017: 280-287.

[35]
Sartori M, Schiavo S, 2015. Connected we stand: A network perspective on trade and global food security. Food Policy, 57: 114-127.

DOI

[36]
Siripirote T, Sumalee A, Ho H, 2020. Statistical estimation of freight activity analytics from Global Positioning System data of trucks. Transportation Research Part E: Logistics and Transportation Review, 140: 101986.

DOI

[37]
Sun C, Han X, Qin X, 2014. Research on pattern stability of virtual water flow embedded in major crops trade among regions in China. Geographical Research, 33(3): 478-489. (in Chinese)

[38]
Sun S, Wang Y, Engel B A et al., 2016. Effects of virtual water flow on regional water resources stress: A case study of grain in China. Science of the total Environment, 550: 871-879.

DOI

[39]
Wang J, Dai C, Zhou M et al., 2021. Research on global grain trade network pattern and its influencing factors. Journal of Natural Resources, 36(6): 1545-1556. (in Chinese)

DOI

[40]
Wang J, Du D, Huang J, 2020. Inter-city connections in China: High-speed train vs. inter-city coach. Journal of Transport Geography, 82: 102619.

DOI

[41]
Wang L, Anna H, Zhang L et al., 2019. Spatial and temporal changes of arable land driven by urbanization and ecological restoration in China. Chinese Geographical Science, 29(5): 809-819.

DOI

[42]
Wang W, Zhuo L, Li M et al., 2019. The effect of development in water-saving irrigation techniques on spatial-temporal variations in crop water footprint and benchmarking. Journal of Hydrology, 577: 123916.

DOI

[43]
Wang X, Qiang W, Niu S et al., 2018. Analysis on global agricultural trade network and its evolution. Journal of Natural Resources, 33(6): 940-953. (in Chinese)

DOI

[44]
Xin R, Ai T, Ding L, et al., 2022. Impact of the COVID-19 pandemic on urban human mobility:A multiscale geospatial network analysis using New York bike-sharing data. Cities, 126: 103677.

[45]
Xu G, Lv Y, Sun H et al., 2021. Mobility and evaluation of intercity freight CO2 emissions in an urban agglomeration. Transportation Research Part D: Transport and Environment, 91: 102674.

DOI

[46]
Yang J, Wang J, Xu C et al., 2021. Rice supply flows and their determinants in China. Resources, Conservation and Recycling, 174: 105812.

[47]
Yang L, Wang J, Yang Y, 2022a. Spatial evolution and growth mechanism of urban networks in western China: A multi-scale perspective. Journal of Geographical Sciences, 32(3): 517-536.

DOI

[48]
Yang Y, Jia B, Yan X-Y et al., 2022b. Identifying intercity freight trip ends of heavy trucks from GPS data. Transportation Research Part E: Logistics and Transportation Review, 157: 102590.

DOI

[49]
You S I, Ritchie S G, 2019. Tour-based truck demand modeling with entropy maximization using GPS Data. Journal of Advanced Transportation, 2019: 5021026.

[50]
Zhao P, Liu X, Shi W et al., 2020. An empirical study on the intra-urban goods movement patterns using logistics big data. International Journal of Geographical Information Science, 34(6): 1089-1116.

DOI

[51]
Zhu Z, Ren H, Ruan S et al., 2021. ICfinder: A ubiquitous approach to detecting illegal hazardous chemical facilities with truck trajectories. Proceedings of the 29th International Conference on Advances in Geographic Information Systems, ACM, 2021: 37-40.

[52]
Zipf G K, 1949. Human Behavior and the Principle of Least Effort: An Introduction to Human Ecology. Cambridge, MA: Addison-Wesley.

Options
Outlines

/

〈 〉