Next Article in Journal
The Distribution Pattern and Spatial Morphological Characteristics of Military Settlements Along the Ming Great Wall in the Hexi Corridor Region
Next Article in Special Issue
Spatiotemporal Dynamics and Driving Factors of Urban Green Space in Texas (2001–2021): A Multi-Source Geospatial Analysis
Previous Article in Journal
Lack of Thermal Comfort Is a Matter of Life and Death: A Systematic Review for Older People
Previous Article in Special Issue
Research on the Dynamic Defensive Efficiency of Traditional Military Settlements in the Ming Dynasty: A Case Study of Ningbo Coastal Defense, China
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Research on the Evolution Characteristics of Building Space in the Central Urban Area of Tianjin Based on Multi-Source Data Collaboration: 2021–2024

by
Yicheng Zhang
1,
Guorui Chen
2 and
Xue Yang
3,*
1
School of Architecture, Tianjin University, Tianjin 300072, China
2
College of Architecture and Art, North China University of Technology, Beijing 100144, China
3
Institute for Urban Regeneration and Development, Tianjin University, Tianjin 300072, China
*
Author to whom correspondence should be addressed.
Buildings 2025, 15(7), 1142; https://doi.org/10.3390/buildings15071142
Submission received: 26 February 2025 / Revised: 25 March 2025 / Accepted: 28 March 2025 / Published: 31 March 2025
(This article belongs to the Special Issue Research towards the Green and Sustainable Buildings and Cities)

Abstract

:
Urban renewal faces critical challenges in balancing heritage protection and functional upgrades, particularly in dual-attribute cities like Tianjin that preserve industrial legacy while cultivating emerging functions. Existing studies exhibit three gaps: geographical bias toward megacities, fragmented analysis of functional–morphological interactions, and inadequate quantification of “protection-development” tensions. This study addresses these gaps through an integrated framework combining point-of-interest kernel density analysis and satellite imagery segmentation (2021–2024 data). The methodological innovations include: (1) Analysis of urban function changes based on Point of Interest density; (2) Analysis of urban spatial morphology changes based on the texture of buildings within plots; (3) Spatiotemporal coupling analysis of data. Key findings reveal: (a) The overall Point of Interest density in Tianjin increased by 127.2%, achieving further prosperity and development of the city; (b) The renewal of the central urban area exhibits characteristics of “edge aggregation and gradient diffusion”; (c) The historic urban area has reshaped its functional layout through micro-renewal and the relocation of industrial spaces, effectively balancing the conflict between preservation and development. This study systematically summarizes the experiences in resolving the conflict between preservation and development in the urban renewal of Tianjin, providing a reference case for cities undergoing similar dual-attribute renewal.

1. Introduction

As a strategic engine for addressing land scarcity, revitalizing aging infrastructure, and reconciling heritage preservation with modern development demands, urban renewal has emerged as a cornerstone of sustainable urbanization in the post-industrial era [1]. China’s “14th Five-Year Plan” (2021–2025) [2] elevates urban renewal to a national strategic level, aiming to address multiple challenges such as urbanization transformation, economic quality and efficiency improvement, enhancement of people’s well-being, ecological governance, and cultural heritage through systematic solutions. The core of this strategy is to shift the urban development model from “scale expansion” to “connotative development”. By optimizing urban spatial structure, improving public service facilities, protecting historical context, and promoting green and low-carbon development, it seeks to achieve a coordinated enhancement of economic, social, and ecological benefits. This marks a new development stage in China’s urbanization process, emphasizing both quality and efficiency [3,4,5].
By the end of 2024, it will have been four years since the comprehensive implementation of urban renewal initiatives, making it necessary to summarize the experiences and provide a theoretical basis for policy adjustments in the final year of the 14th Five-Year Plan. However, previous urban renewal studies still exhibit deficiencies in the following two aspects:
On the one hand, previous studies have been highly concentrated on the core urban areas of international metropolises such as Beijing, Shanghai, and Shenzhen [6,7,8,9,10,11], forming a case study paradigm characterized by “high-density built environments”. However, there has been insufficient attention to cities like Tianjin, which possess the dual attributes of “historical industrial heritage preservation” and “construction of emerging functional zones”. This has resulted in a gap in the explanatory power of theoretical models when addressing the synergistic mechanisms of “incremental renewal” (e.g., the revitalization of the Haihe Economic Belt) and “leapfrog development” (e.g., the construction of the Binhai New Area).
On the other hand, there is data fragmentation: a lack of a collaborative analysis framework for multi-source spatiotemporal data. Although POI (point of interest) data and satellite imagery have been used separately for urban function quantification and morphological analysis, their collaborative application is still in its infancy. In terms of functionality, while POI kernel density analysis [12,13,14] can depict the spatial clustering characteristics of commercial facilities, it is difficult to reveal the feedback effects of functional mixing on spatial morphology (such as gentrification triggered by excessive commercial expansion in historical districts). In terms of morphology, deep learning-driven interpretation of satellite and aerial imagery [15,16,17] can precisely identify changes in building textures but has not established dynamic association models with the evolution of POI functions. More critically, existing studies often adopt “static data slices” (such as matching POI and imagery from a single year) [18,19], lacking the capture of the “temporal response chain” in urban renewal processes. For example, although multi-agent models [20,21,22] can simulate the impact of traffic interventions on resident behavior, they have not integrated POI function evolution data to verify the spatiotemporal effects of business adjustments. Meanwhile, streaming POI-LSTM networks [23,24,25] achieve dynamic predictions for commercial site selection but do not couple spatial constraints from satellite imagery analysis. This data fragmentation resulted in a “black box” phenomenon [26] in the study of the linkage mechanisms between morphological reconstruction and functional regeneration.
This study aims to address two critical research gaps in dual-attribute cities like Tianjin: (1) How to quantitatively characterize the ’function-form’ co-evolution mechanisms under the dual constraints of industrial heritage protection and emerging functional cultivation; and (2) How to integrate multi-source spatiotemporal data (POI and satellite imagery) for dynamic monitoring of urban renewal processes. As the 14th Five-Year Plan enters its final year (2024), this study provides evidence-based indicators to directly support Tianjin’s future policy adjustments to optimize the allocation of urban renewal financial resources.

2. Overview of Tianjin’s History and Cultural Heritage Protection

2.1. The Historical Context of Urban Development

Tianjin, as an important port city in northern China, has a history of architectural development that can be traced back to the Ming Dynasty’s establishment of defenses in 1404. It has undergone three major phases: the Ming and Qing dynasties’ canal transportation hub, the development of concessions in the modern era, and contemporary urbanization, forming a unique urban fabric that blends Chinese and Western styles and merges the ancient with the modern. It is known as the “Museum of World Architecture”. In terms of spatial distribution, the old city retains traditional Chinese architectural complexes represented by the Wen Miao (Confucian Temple, originally built in the 1430s) and the Tianhou Temple (originally built in the 1320s), whose wooden framework system and axial layout reflect the regulatory features of northern official architecture (Figure 1). After the port opened in 1860, the nine-nation concession area (which, at its peak, covered more than 15 square kilometers) displayed a distinct Western architectural style, exemplified by buildings such as the HSBC Building (constructed in the 1920s) and the former site of the Astor Hotel (constructed in the 1860s) (Figure 2). Currently, Tianjin has successfully transformed into a modern international port city with the Binhai New Area as its engine, and the skyline of Tianjin is continuously being updated. Modern buildings such as the Tianjin Eye (the only Ferris wheel in the world built on a bridge) and the Jin Tower have become new landmarks of Tianjin (Figure 3).

2.2. Cultural Heritage Protection System

The cultural heritage protection system in Tianjin has established a heritage management model that is both systematic and innovative through a multi-level institutional framework and practical approaches. As of 2024, Tianjin has designated 33 National Key Cultural Relics Protection Units and has implemented a three-tiered classification protection (special, key, and general) for 877 historical buildings, covering the diverse characteristics of architectural heritage. Additionally, through spatial planning integration, 14 historical and cultural districts, including Wudadao and Jiefangbei Road, have been incorporated into a holistic protection framework, while 58 buildings such as the Quanyechang and Guangdong Guild Hall have been selected for the “List of 20th Century Architectural Heritage in China”, highlighting their significance in the history of modern architecture. In terms of international heritage certification, the Tianjin section of the Grand Canal (from Sanchakou to Yangliuqing), as a crucial part of the Grand Canal of China, was inscribed on the World Heritage List in 2014. Its heritage elements include hydraulic facilities, dock sites, and historical districts along the route.
The architectural history of Tianjin encapsulates China’s transition from a traditional agrarian civilization to a modern industrial civilization. Through systematic protection and innovative utilization, Tianjin has maintained its urban characteristic of “coexistence of the ancient and the modern” during the modernization process, providing a model for the historical and cultural inheritance of port cities worldwide.

3. Materials and Methods

In response to the aforementioned issues, this study proposes a dual-dimension analytical framework (Figure 4) for decoding the “function-form” co-evolution mechanism in urban renewal, integrating multi-source spatiotemporal data through three methodological pillars:
  • Functional dimension: A diachronic POI density analysis [27] path is constructed using kernel density estimation to quantify the spatial fission (e.g., commercial expansion) and aggregation (e.g., industrial relocation) of urban functions.
  • Morphological dimension: Building texture changes are extracted using U-Net (Convolutional Networks for Biomedical Image Segmentation) [28] semantic segmentation, classifying plots into retained, demolished, and newly added types.
  • Spatiotemporal coupling: By overlaying changes in POI density with spatial morphological changes, dynamic monitoring of the impact of renewal is achieved. This framework advances previous unidimensional methods by addressing the issue of data fragmentation in urban renewal analysis.

3.1. Research Scope and Datasets

3.1.1. Research Area and Phase

Since 2021, Tianjin has actively responded to the national urban renewal strategy by successively introducing the “Implementation Plan for the Renovation and Upgrade of Old Houses and Old Residential Communities and Urban Renewal in Tianjin” [29] and the “Tianjin Urban Renewal Action Plan (2022–2025)” [30]. Relying on the rich historical and cultural resources of the central urban area, Tianjin has successively undertaken representative urban renewal projects such as the Quanyechang Shopping Mall and the Second Workers’ Sanatorium, providing important references for renewal practices in other cities across the country.
Choosing Tianjin as the research subject has triple typicality: the parallel regeneration of the Haihe industrial heritage belt and the construction of the Binhai New Area provides a natural laboratory to test the “stock renewal-incremental development” collaborative mechanism [31,32]; the 2021–2024 time-series data cover the main phases of the “14th Five-Year Plan” policy implementation (2021–2024), allowing clear capture of the function–form collaborative evolution path; the gentrification effects in micro-renewals within historic urban areas and the concentrated spatial planning outside these areas reflect the conflict and integration between historical preservation and development.
The study selects the central urban area within the outer ring road of Tianjin as the research scope, based on the following four considerations:
  • Typicality: As an important economic center and port city in northern China [33], Tianjin has experienced a rapid urbanization process and has now entered the stock adjustment stage, providing a rich basis of cases and data for urban renewal research.
  • Functional Diversity: The central urban area encompasses a variety of functional spaces including residential, commercial, industrial, and public services [34]. The renewal needs of different functional areas vary significantly, making it suitable for multidimensional comparative analysis.
  • Data Accessibility: The central urban area boasts abundant and easily accessible POI data and satellite imagery [35], providing reliable data support and enhancing the feasibility and scientific nature of the research.
  • Policy Support: Tianjin’s policy practices in the field of urban renewal are pioneering and exemplary [29,30], providing a favorable policy environment for the study of the central urban area.
This study focuses on the urban renewal process in the central urban area of Tianjin from 2021 to 2024. The reason is that Tianjin began to fully implement urban renewal in 2021, and the summary of experiences over these four years will help in adjusting the direction of subsequent urban renewal policies, theories, and practices.

3.1.2. Dataset Construction

  • POI Data
POI data, as an important component of geographic information, hold significant value in fields such as urban planning, business analysis, and tourism recommendations. This study obtained POI data for Tianjin through the Amap API, which offers advantages such as rich data, real-time updates, and easy access. Using Python programming (PyCharm Community Edition, Version 2024.3; Python 3.13.2), the study completed steps including coordinate picking, POI data retrieval, format conversion, and coordinate transformation to ensure data accuracy and consistency. To meet the research needs, the POI data were reclassified into 10 subcategories based on functionality (Table 1).
2.
Satellite Imagery Data
Satellite imagery provides high-resolution information on topography, urban texture, and historical evolution for urban renewal research. There are various satellite remote sensing data products available globally, such as Google Earth, Planet Labs, and Maxar Technologies. Based on the research needs, including high resolution, extensive data coverage, rich historical information, and lower data acquisition costs, this study uses Google Earth historical satellite imagery as the data source. Since Google Earth does not support direct downloading of images for specific areas or selecting resolution levels, this study employs Python to download image tiles from the Google Earth web platform and stitch them together to generate high-resolution satellite imagery of the central urban area of Tianjin for the years 2021 and 2024.

3.2. Model Training and Evaluation

3.2.1. Overview of the Model

U-Net is a convolutional neural network (CNN) architecture widely used for image segmentation tasks, proposed by Ronneberger et al. in 2015 [36]. Inspired by the Fully Convolutional Network (FCN), it adopts a unique U-shaped structure, consisting of a contracting path (encoder) and a symmetric expanding path (decoder). The contracting path captures the contextual information of the image through successive convolutional layers and pooling layers, while the expanding path gradually restores the spatial resolution of the image through upsampling and convolutional layers. A core feature of U-Net is the “skip connections”, which connect feature maps from the encoder with corresponding layers in the decoder, thus preserving more positional information and details during the segmentation process. This architecture enables U-Net to perform exceptionally well when processing images with complex boundaries and details, and it is widely applied in fields such as medical image analysis, architectural structure monitoring, and remote sensing image processing [37].

3.2.2. Training Process

The prediction of buildings is a binary classification problem, where all parts of the dataset, except the target, are labeled as background. The training process included the following steps:
  • Data Processing
Use the os and shutil libraries in Python to clean and format the training and test data, standardizing it into a VOC format dataset.
2.
Image Cropping
To facilitate model inference, crop large remote sensing images into smaller images of 512 × 512 pixels using the PIL library for cropping operations.
3.
Model Training
Train the model based on public datasets to ensure the model’s generalization ability.
4.
Model Optimization
Use the remote sensing images of Tianjin’s central urban area from 2021 and 2024 to test the model. If the prediction accuracy does not meet the requirements, optimize it by changing the dataset or adjusting parameters until the model’s performance satisfies the research needs. The final model’s mIoU (mean Intersection over Union) reaches 70–80, indicating that the model is well-trained, as shown in Figure 5.

3.3. Data Analysis

3.3.1. POI Data Analysis

This study utilizes the kernel density analysis tool in ArcGIS 10.8 to systematically analyze POI data in Tianjin, aiming to reveal the characteristics of functional evolution in the central urban area. Kernel density analysis effectively reflects the spatial aggregation degree of different types of points of interest by calculating the distribution density of POIs within a specific region. By setting appropriate search radius and weight parameters, density maps for various types of POIs are generated, thereby identifying their spatial distribution patterns. By comparing the density distribution of different categories of POIs, a deeper understanding of the macro layout of functional spaces in Tianjin and their impact on urban development can be achieved.

3.3.2. Satellite Imagery Data Analysis

  • Morphological Prediction
Use the optimized U-Net model to perform semantic segmentation on remote sensing images of the central urban area of Tianjin in 2021 and 2024, extracting building contour information.
2.
Morphological Comparison
Using the PIL library to compare the building prediction results of 2021 and 2024 on a pixel-by-pixel basis, the changes are categorized into three classes:
  • Retained: Building footprints that persist from 2021 to 2024;
  • Demolished: Building footprints that disappear from 2021 to 2024;
  • Added: Building footprints that are newly added from 2021 to 2024.
3.
Image Fusion
Use the PIL library to stitch the cropped smaller images by rows and columns, restoring them to a large image with dimensions consistent with the original image.
4.
Data Post-Processing
To optimize the segmentation results considering potential errors caused by factors such as image quality, building shadows, seasonal changes, and the complexity of ground objects, the plots with texture variation (referred to as “updated plots”) are categorized into three types based on research needs (Figure 6):
  • Texture Change: The existing building footprints within the plots change from 2021 to 2024 but do not completely disappear;
  • Texture Disappear: The existing building footprints within the plots completely disappear from 2021 to 2024;
  • Texture Appear: New building footprints appear within vacant plots from 2021 to 2024.

4. Results

4.1. Characteristics of Urban Functions Evolution

4.1.1. Characteristics of Spatial Differentiation in Synchrony

Based on the spatial differentiation characteristics of comprehensive kernel density analysis (Figure 7), the POI distribution in Tianjin from 2021 to 2024 exhibits a significant “one core dominance, secondary core support, and multi-point coordination” radiation development model. The overall POI density reaches 799.1 points/km2, among which:
  • The core area (central urban area), as a traditional city center, carries the dual mission of historical and cultural preservation and functional renewal. The POI density gradient shows a characteristic of “dense inside, sparse outside”, with commercial facility POIs having the highest proportion and an average density of 529.6 points/km2 and the core’s highest density reaching approximately 2000 points/km2.
  • The secondary core area (Binhai New Area) is primarily driven by industrial functions, forming a “large-scale, cluster-style” renewal model by leveraging the advantages of national-level new area policies. The POI for logistics and warehouse facilities has grown rapidly, with an overall density increase of 226.1% compared to 2021, creating a “mutual nourishment of twin cities” effect with the central urban area.
  • The multi-point areas (such as Wuqing and Baodi, which are remote suburban areas) exhibit a “patchy” distribution characteristic, with a significant increase in green space POI density (Δ = 364.6%), reflecting the development philosophy that “lucid waters and lush mountains are invaluable assets”.

4.1.2. Evolutionary Patterns of Temporal Functionality

Between 2021 and 2024, the overall POI density in the central urban area increased by 84.3%. Through the analysis of kernel density of different functional types of POIs (Figure 8), it was found that some functional types showed distinct evolutionary paths:
  • Public space fission: The POI density of green space increased by 762.7%, forming a “Haihe Green Corridor + Community Park” nested structure; the POI density of plaza and street space increased by 116.7%, reflecting the Tianjin municipal government’s absolute emphasis on environmental protection and the livelihood sector during the “14th Five-Year Plan” period.
  • Improvement of public services: Benefiting from the continuous investment of special funds for the renovation of old residential areas, the POI density of public service facilities increased by 15.9% on the basis of 62.6 points/km2 in 2021, and the POI density of public utilities also increased by 30.9%. The construction effectiveness of “complete communities” was quite significant.
  • Industrial space restructuring: Industrial facilities were relocated from the central urban area, resulting in a 44.3% decrease in POI density, while a dual-core cluster formed in the Binhai New Area, and the Xiqing-Wuqing industrial corridor began to take shape. The POI density of commercial facilities increased by 193.3%, with significant growth in both the density and area of core commercial districts. This corroborated the “dual-city multi-node” urban functional spatial pattern established by Tianjin’s “14th Five-Year Plan”, as well as the land-use planning and control measures guiding industrial relocation and service sector agglomeration.

4.2. Characteristics of Urban Morphology Evolution

In November 2022, the Tianjin Municipal Planning and Natural Resources Bureau announced the “Tianjin Historical and Cultural City Protection Plan (2021–2035)”, which clearly delineated the scope of the historical urban area of Tianjin. As a city recognized for its historical and cultural significance in China, the renewal and development of Tianjin’s central urban area are influenced by multiple factors, with distinct differences observed inside and outside the historical urban area.

4.2.1. Typological Characteristics of Spatial Texture

The “Tianjin Historical and Cultural City Protection Plan (2021–2035)” [38] delineates the historical district with reference to the built-up area of the city in 1949, covering an area of approximately 54 square kilometers. This district includes the old city, nine foreign concessions, and contemporary business districts, highlighting the more pronounced contradiction between protection and development. This study analyzes the urban fabric from the “inner-outer” dimensions of the historical district (Figure 9):
  • Inside the Historical District
In terms of texture types, traditional hutongs, Western-style blocks, and modern high-rise clusters coexist along the banks of the Haihe River, forming a composite structure of “Haihe Axis + Grid Units”. Regarding functional distribution, residential areas (40.8%), public services (30.4%), and commercial areas (14.5%) create a “residential-dominated, service-supported” functional ratio, with industrial land accounting for only 1.4%. Additionally, the historical district retains a large number of historical buildings and cultural sites, which not only enrich the city’s cultural heritage but also provide residents with ample cultural activities and leisure spaces. In terms of spatial utilization, the historical district exhibits high density and compactness, with high land use efficiency. Moreover, due to the well-developed transportation network and comprehensive public transportation facilities within the historical district, residents enjoy convenient travel, further promoting the region’s economic development and population concentration.
2.
Outside the Historical District
In terms of texture type, a “layered radiation” structure is presented, with residential clusters on the inside and industrial parks on the outside. In terms of functional composition, residential areas (68.8%) and industrial areas (11.4%) dominate, while public service facilities account for 10.9%, reflecting the characteristics of industry–city integration. Additionally, outside the historical district, the texture shows a clear trend towards modernization, with high-rise residential areas and emerging industrial parks gradually rising, bringing new vitality and economic growth points to the city. In the spatial layout, the area outside the historical district exhibits low-density, decentralized characteristics, with relatively flexible land use, which is conducive to urban expansion and renewal. It is noteworthy that the transportation network outside the historical district is also continuously improving, with the construction of highways, subways, and other modern transportation facilities further enhancing the area’s convenience and attractiveness. Meanwhile, with the advancement of urbanization, a number of commercial streets and cultural venues with local characteristics have emerged outside the historical district, providing residents with diverse options for consumption and entertainment. These emerging commercial and cultural venues not only enrich the city’s cultural connotations but also promote interaction and integration between the historical district and external areas.

4.2.2. Spatiotemporal Differentiation of the Update Process

Through the detection of changes in architectural texture (Figure 10), it was found that the renewal intensity of the central urban area presents a characteristic of “edge aggregation and gradient diffusion”:
  • Overall Renewal Characteristics (2021–2024):
The renewal plots in the central urban area account for 4.47% of the total area, mainly concentrated in the edge regions. These edge regions often have significant land development potential and policy support, thus becoming key areas for urban renewal projects. This helps to alleviate spatial pressure within the central urban area while promoting rational urban spatial layout and functional optimization. Among them, renovation-type renewal plots (1.71%) mainly involve the renovation of industrial facilities, public facilities, and public service facilities, aiming to enhance the efficiency of existing infrastructure and promote the optimal allocation and sustainable use of resources. Demolition-type renewal plots (0.40%) mainly target old industrial facilities and shantytowns, reflecting the urban renewal policies of old city renovation and shantytown redevelopment, aiming to improve residents’ living environment and enhance the overall image of the city. New construction-type renewal plots (2.36%) involve the addition of residential, commercial, and public service facilities, reflecting the response to new functional demands and the expansion of spatial layout in the process of urban development.
2.
Specificity of Historical Urban Area Renewal:
In order to protect historical buildings and cultural heritage, the renewal of historical urban areas is strictly restricted, requiring a balance between preservation and development. Moreover, the historical urban area is primarily positioned for residential, commercial, and public service functions, with renewal projects focusing more on enhancing the efficiency and quality of existing functions rather than large-scale structural changes. Between 2021 and 2024, renewal plots within the historical urban area accounted for only 1.51%, with the renovation-type mainly involving the activation of inefficient public spaces (0.37%), demolition-type for shantytown redevelopment (0.05%), and new construction-type mainly focusing on public service and transportation facilities (1.09%). Given limited resources, the government may prioritize allocating more renewal resources to the external areas of the central urban area to promote balanced development of the entire city.

4.3. Feature Summary

The evolution of the functions and forms of the central urban area of Tianjin exhibits a complex characteristic of parallel protection and development. Functionally, the core area has established a spatial pattern centered on historical and cultural preservation dominated by modern service functions by relocating industrial facilities and enhancing the concentration of commerce and services. It also builds a “residence-service-ecology” nested system by relying on the fission of public spaces and the improvement of public services. The morphological evolution presents a gradient differentiation characteristic: within the historical city, a composite structure of traditional texture and modern high-rises is maintained, and inefficient spaces are revitalized through micro-updates, with strict restrictions on large-scale development. The external areas promote industrial–urban integration through a concentric expansion model, achieving industrial relocation and a residential–industry cluster layout through concentrated renewal of peripheral areas, forming a coordinated development path of “core conservation-outer expansion”. This spatial restructuring not only maintains the spatial genes of a historical and cultural city but also responds to the needs of sustainable urban development through functional replacement and gradient development, providing a valuable experience and reference for the renewal and development of other cities.

5. Discussion

This study, through the integration and analysis of data from multiple sources, successfully reveals the synergistic evolution patterns between function and form in the urban renewal process of Tianjin. The research results indicate that urban renewal is not merely a transformation of physical space but a process of mutual interaction and promotion between function and form. However, despite achieving certain research findings, there is still room for optimization in terms of the breadth and depth of data collection, the accuracy and efficiency of data cleaning, and the precision and applicability of algorithms. These issues may potentially affect the accuracy and reliability of the research results. Therefore, to further enhance the quality and depth of the research, it is necessary to delve into these potential impacts and find corresponding improvement paths. These include, but are not limited to, optimizing data collection methods, increasing the automation and intelligence level of data cleaning, and improving and innovating algorithms to adapt to the complex and ever-changing scenarios of urban renewal.

5.1. The Temporal and Spatial Limitations of Data Collection

In terms of data collection, the resolution of Google Earth imagery (0.5–1 m) makes it difficult to accurately identify small building modifications (such as illegal shacks and temporary structures), resulting in a potential 15–20% miss rate for plots requiring demolition updates. This particularly affects the spatial recognition accuracy of micro-renewal projects in historic urban areas (e.g., renovation projects of less than 50 m2 within courtyard alleys may not be effectively captured). Moreover, information on the renovation of building facades, municipal pipelines, and other such details is also difficult to obtain through satellite imagery. Additionally, the POI update cycle for Amap is approximately 3–6 months, leading to a data lag for new facilities (for example, the POI annotation for a new community service center in 2024 may be delayed), and its classification system does not reflect the contribution values of new hybrid functional spaces such as “industrial heritage revitalization” and “shared offices”. Improvement directions include integrating drone oblique photography (0.1 m resolution) and nighttime light data (NPP/VIIRS) to build a multi-scale spatial perception system, as well as introducing crowdsourced geographic data (such as consumer review site merchant tags) and planning approval data to construct a multi-source database for urban function perception. Moreover, considering the impact of different seasons and weather conditions on satellite image quality, it is necessary to develop targeted image screening standards to reduce interference with data collection accuracy caused by cloud cover, shadow casting, and other factors. Furthermore, to address the lag in POI data, establishing a real-time data-sharing mechanism with local government planning departments can be considered to ensure the timely inclusion and classification update of new facilities, thereby improving data timeliness and accuracy. These improvement measures are expected to further enhance the spatial recognition accuracy and functional analysis depth of urban renewal projects.

5.2. Challenges of Data Cleaning Complexity

In terms of data cleaning, the U-Net model has a misclassification rate of up to 12.3% for building shadows in high-density urban areas (as shown in Figure 11), with some contiguous shadows being mistakenly identified as demolition-type plots. This leads to a potential overestimation error of 5–8% in the demolition-type update area (0.40%) in central urban areas. Additionally, some POI (point of interest) labels (such as “community convenience store”) exhibit functional alienation (actually also serving as parcel collection points, community canteens, etc.), leading to a deviation of up to ±0.05 in the functional mix index of kernel density analysis, affecting the accuracy of public service facility density assessment. Improvement methods include integrating multi-temporal imagery and LiDAR point cloud data, verifying segmentation results through three-dimensional geometric features, and employing the BERT model to perform semantic parsing on POI names and user reviews, thereby constructing a fine-grained functional label system. Furthermore, in low-density residential and industrial areas at the urban fringe, due to the scattered distribution of buildings and high vegetation coverage, the U-Net model may experience breaks and misalignments in identifying building outlines, resulting in biased estimates of building density and floor area ratio in these areas. To address this issue, super-resolution reconstruction technology from deep learning can be introduced to enhance the detailed features of images, thereby improving the accuracy of building identification. Additionally, combining field surveys and drone aerial data can facilitate manual verification and correction of the model’s output to ensure the quality and accuracy of data cleaning.
Moreover, due to the diversity and complexity of urban renewal projects, the renewal models and functional orientations of different areas vary, posing higher demands on data cleaning. Therefore, during the data cleaning process, it is essential to fully consider the characteristics and needs of urban renewal and develop targeted cleaning strategies and methods to ensure data accuracy and usability. For example, for renewal projects in historical urban areas, emphasis should be placed on the protection and inheritance of architectural styles, as well as the excavation and display of historical culture. In contrast, in emerging industrial clusters, attention should be given to the aggregation and allocation of innovative elements, as well as the construction and optimization of the industrial ecosystem.

5.3. Spatial Heterogeneity of Algorithm Accuracy

In terms of algorithm accuracy, the U-Net model shows a significant performance difference between historical urban areas (mIoU = 78.2%) and industrial parks (mIoU = 65.4%), with the segmentation accuracy for industrial factory sloped roofs being 17.6 percentage points lower than that for residential buildings. This may lead to conclusions about changes in the POI density of industrial facilities in the Binhai New Area being affected by shape misdetection. Additionally, the time reference inconsistency between POI data (updated quarterly) and satellite imagery (updated annually) results in a 3–6-month lag effect in function–morphology correlation analysis, reducing the confidence in causality inference regarding the growth in logistics and warehousing facilities’ density and spatial expansion. Improvement directions include introducing the Transformer architecture to enhance feature extraction capabilities, adding prior knowledge of building roof types (such as sloped roof and flat roof classifiers) and constructing a Space-Time Cube model for sliding window matching on a monthly basis. To further reduce the impact of spatial heterogeneity on algorithm accuracy, a multi-scale fusion strategy could be adopted by integrating satellite imagery data of different resolutions to capture multi-level information from local details to the overall layout. Additionally, considering that varying seasonal lighting conditions and environmental changes may affect image quality, exploring seasonal adaptive algorithms that automatically adjust model parameters according to seasonal changes could improve segmentation accuracy across different seasons. Finally, to more comprehensively evaluate the algorithm’s performance, more evaluation metrics such as Precision, Recall, and F1 Score could be introduced to comprehensively assess the model’s performance in different scenarios.

6. Conclusions

This study employs an analytical approach that combines POI data with satellite imagery to deeply explore and reveal the spatiotemporal patterns of the collaborative evolution of functionality and form in the central urban area of Tianjin:
  • In terms of functionality, this study finds that the concentration of commercial activities and the dispersed reorganization of industrial activities have reshaped the urban spatial structure, forming a new pattern of “dual-city with multiple nodes”. At the same time, the fission of public spaces and the improvement in infrastructure quality have strongly supported the construction of “complete communities”.
  • In terms of form, this study points out the gradient differences between the concentrated renewal of peripheral areas and the micro-renewal of historical urban areas, reflecting the sustainable development path the city has taken in “core conservation and peripheral expansion”.
  • Methodologically, this study combines POI kernel density analysis with U-Net architectural texture detection technology to build an integrated framework, providing a new technical paradigm for dynamic monitoring of urban renewal.
Nevertheless, this study also highlights the limitations of the current data resolution and the inadequacies of algorithms in addressing spatial heterogeneity, issues that require further optimization through the integration of multi-source data. Looking forward, the research can be further expanded to assess the functional mix of industrial heritage and simulate renewal policies, aiming to provide more systematic and comprehensive decision support for high-quality urban development. Additionally, there is a need for ongoing collaboration between researchers, policymakers, and stakeholders to ensure that the insights gained from this study are effectively translated into actionable strategies. By fostering interdisciplinary dialogue, we can better understand the complex interplay between industrial heritage, urban renewal, and sustainable development.

Author Contributions

Conceptualization, Y.Z. and X.Y.; methodology, G.C.; software, Y.Z.; validation, Y.Z. and X.Y.; formal analysis, G.C.; investigation, Y.Z.; resources, X.Y.; data curation, Y.Z.; writing—original draft preparation, Y.Z.; writing—review and editing, Y.Z.; visualization, G.C.; supervision, X.Y.; project administration, X.Y.; funding acquisition, X.Y. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the 14th Five-Year National Key Research and Development Program of the Ministry of Science and Technology of the People’s Republic of China, “Construction of Urban Renewal Typology and Renewal Methods for Typical Types” (2022YFC3800303).

Data Availability Statement

The datasets presented in this article are not readily available because the data are part of an ongoing study.

Acknowledgments

In addition, the authors would like to thank Kun Song for his guidance on the research direction and Wanqi Jing and Zehao Liu for their assistance with the specific work.

Conflicts of Interest

The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

References

  1. Bi, C.; Little, J.C. Integrated assessment across building and urban scales: A review and proposal for a more holistic, multi-scale, system-of-systems approach. Sustain. Cities Soc. 2022, 82, 103915. [Google Scholar] [CrossRef]
  2. Outline of the 14th Five-Year Plan for National Economic and Social Development of the People’s Republic of China and the Long-Range Objectives Through the Year 2035. Available online: https://www.gov.cn/xinwen/2021-03/13/content_5592681.htm (accessed on 20 March 2025).
  3. Zhang, J.; Wang, G.; Xiong, L. Using a Data Mining Method to Explore Strategies for Improving the Social Interaction Environment Quality of Urban Neighborhood Open Spaces. Architecture 2023, 3, 128–136. [Google Scholar] [CrossRef]
  4. Sun, J.; Li, H.; Xiao, R.; Yao, G.; Zou, F. Dynamics of Heat Island Intensity in a Rapidly Urbanizing Area and the Cooling Effect of Ecological Land: A Case Study in Suzhou, Yangtze River Delta. Sustainability 2024, 16, 4422. [Google Scholar] [CrossRef]
  5. Ouyang, Y.; Bai, X.; Wang, X.; Chen, Y.; Huang, G.; Xie, D. Case Study on Cultural Industry Empowerment in Urban Renewal: A Focus on Guangzhou, China. Sustainability 2025, 17, 439. [Google Scholar] [CrossRef]
  6. Li, Q.; Lv, S.; Cui, J.; Zhang, J.; Liu, Y. Environmental Perception about Pedestrian Environment on Cultural Visitation Roads. Sustainability 2024, 16, 7097. [Google Scholar] [CrossRef]
  7. Zhang, Z.; Sun, Q. Integrated Design Methods for Sustainable Public Seating in Urban Communities—A Shanghai Case Study. Sustainability 2024, 16, 9096. [Google Scholar] [CrossRef]
  8. Zhang, P.; Zhang, T.; Fukuda, H.; Ma, M. Evidence of Multi-Source Data Fusion on the Relationship between the Specific Urban Built Environment and Urban Vitality in Shenzhen. Sustainability 2023, 15, 6869. [Google Scholar] [CrossRef]
  9. Zhang, W.; Xin, J. Green Spaces and the Spontaneous Renewal of Historic Neighborhoods: A Case Study of Beijing’s Dashilar Community. Sustainability 2023, 15, 16566. [Google Scholar] [CrossRef]
  10. Jiang, C.; Huang, R.; Huang, S.; Shen, T. Unveiling Sustainable Co-Creation Patterns in Entrepreneurial Ecosystems of Shanghai’s High-Density Urban Communities. Sustainability 2024, 16, 10642. [Google Scholar] [CrossRef]
  11. Lee, C.-C.; Wang, Y.; Tang, C.; Li, X.; Yin, J. A Visual Preference Analysis of Building Façades in Street View Images: A Case Study of Nanshan District, Shenzhen. Buildings 2025, 15, 611. [Google Scholar] [CrossRef]
  12. Šoltésová, M.; Iannaccone, B.; Štrba, Ľ.; Sidor, C. Application of GIS Technologies in Tourism Planning and Sustainable Development: A Case Study of Gelnica. ISPRS Int. J. Geo-Inf. 2025, 14, 120. [Google Scholar] [CrossRef]
  13. Yao, C.; Li, G.; Yan, S. Design Strategies to Improve Metro Transit Station Walking Environments: Five Stations in Chongqing, China. Buildings 2024, 14, 1025. [Google Scholar] [CrossRef]
  14. Guo, X.; Cui, X. GIS-Based Study on the Suitability of Site Selection for Community Complexes in Old Residential Areas in Xi’an, China. Buildings 2025, 15, 551. [Google Scholar] [CrossRef]
  15. Agustiyara, A.; Mutiarin, D.; Nurmandi, A.; Kasiwi, A.N.; Ikhwali, M.F. Mapping Urban Green Spaces in Indonesian Cities Using Remote Sensing Analysis. Urban Sci. 2025, 9, 23. [Google Scholar] [CrossRef]
  16. Zhang, W.; Yu, M.; Chen, X.; Zhou, F.; Ren, J.; Xu, H.; Xu, S. Combining Deep Fully Convolutional Network and Graph Convolutional Neural Network for the Extraction of Buildings from Aerial Images. Buildings 2022, 12, 2233. [Google Scholar] [CrossRef]
  17. Fu, X.; Wu, M.; Ponnarasu, S.; Zhang, L. A Hybrid Deep Learning Approach for Real-Time Estimation of Passenger Traffic Flow in Urban Railway Systems. Buildings 2023, 13, 1514. [Google Scholar] [CrossRef]
  18. Li, J.; Chen, Y.; Zhao, D.; Zhai, J. The Impact of Built Environment on Mixed Land Use: Evidence from Xi’an. Land 2024, 13, 2214. [Google Scholar] [CrossRef]
  19. Zhu, D.; Dang, X.; Shi, W.; Chen, Y.; Li, W. Urban Functional Zone Mapping by Integrating Multi-Source Data and Spatial Relationship Characteristics. Proceedings 2024, 110, 17. [Google Scholar] [CrossRef]
  20. Pan, L.; Yang, N.; Zhang, L.; Zhang, R.; Xie, B.; Yan, H. Assessment of the Impact of Multi-Agent Model-Based Traffic Optimization Interventions on Urban Travel Behavior. Electronics 2025, 14, 13. [Google Scholar] [CrossRef]
  21. Kalyuzhnaya, A.; Mityagin, S.; Lutsenko, E.; Getmanov, A.; Aksenkin, Y.; Fatkhiev, K.; Fedorin, K.; Nikitin, N.O.; Chichkova, N.; Vorona, V.; et al. LLM Agents for Smart City Management: Enhancing Decision Support Through Multi-Agent AI Systems. Smart Cities 2025, 8, 19. [Google Scholar] [CrossRef]
  22. Liang, Z.; Várady, G.; Zagorácz, M.B. Sustainable Application of Automatically Generated Multi-Agent System Model in Urban Renewal. Sustainability 2023, 15, 7308. [Google Scholar] [CrossRef]
  23. Zou, J.; Zhang, X.; Cong, Y.; Gao, Z.; Shi, J. Research and Modeling of Commercial Location Selection Based on Geographic Big Data and Mobile Signaling Data—A Case Study of the Central Urban Area of Beijing. ISPRS Int. J. Geo-Inf. 2024, 13, 432. [Google Scholar] [CrossRef]
  24. Li, Y.; Huang, Y.; Liu, Z.; Zhang, B. A Distributed Scheme for the Taxi Cruising Route Recommendation Problem Using a Graph Neural Network. Electronics 2024, 13, 574. [Google Scholar] [CrossRef]
  25. Chen, J.; Feng, Q.; Fan, D. Vehicle Trajectory Prediction Based on Local Dynamic Graph Spatiotemporal–Long Short-Term Memory Model. World Electr. Veh. J. 2024, 15, 28. [Google Scholar] [CrossRef]
  26. Roodenrijs, J.C.M.; Hegger, D.L.T.; Mees, H.L.P.; Driessen, P. Opening up the Black Box of Group Decision-Making on Solar Energy: The Case of Strata Buildings in Amsterdam, the Netherlands. Sustainability 2020, 12, 2097. [Google Scholar] [CrossRef]
  27. Zhang, T.; Zhu, X.; Liu, Y.; Jia, C.; Bai, H. Research on the Spatial Structure of the Beijing–Tianjin–Hebei Urban Agglomeration Based on POI and Impervious Surface Coverage. Buildings 2024, 14, 1793. [Google Scholar] [CrossRef]
  28. Tyvoniuk, V.; Trach, R.; Trach, Y. Integration of Probability Maps into Machine Learning Models for Enhanced Crack Segmentation in Concrete Bridges. Appl. Sci. 2025, 15, 3201. [Google Scholar] [CrossRef]
  29. Notice from the General Office of the Tianjin Municipal People’s Government on Issuing the Implementation Plan for the Renovation and Upgrade of Old Residential Buildings and Neighborhoods and Urban Renewal in Tianjin. Available online: https://www.tj.gov.cn/zwgk/szfwj/tjsrmzfbgt/202106/t20210624_5486108.html (accessed on 20 March 2025).
  30. “Tianjin City Renewal Action Plan (2022–2025)” Open for Public Comments. Available online: https://www.tj.gov.cn/zmhd/dczj/ptwz/202206/t20220609_5900229.html (accessed on 20 March 2025).
  31. Fang, Y.-N.; Zeng, J.; Namaiti, A. Landscape Visual Sensitivity Assessment of Historic Districts—A Case Study of Wudadao Historic District in Tianjin, China. ISPRS Int. J. Geo-Inf. 2021, 10, 175. [Google Scholar] [CrossRef]
  32. Wang, Q.; Sun, X.; Wang, R.; Zhou, L.; Zhou, H.; Di, Y.; Li, Y.; Zhang, Q. Research on Urban Energy Sustainable Plan under the Background of Low-Carbon Development. Sustainability 2023, 15, 14206. [Google Scholar] [CrossRef]
  33. Qu, T.; Zhang, L.; Wu, Z. City Spatial Structure Evolution Affected by Special Events in Different Historical Periods: A Case Study of Tianjin. Sci. Geogr. Sin. 2014, 34, 656. [Google Scholar]
  34. Liu, W.; Yan, J.; Sun, X.; Song, R. Research on “Identification–Cognition–Perception” of the Pedestrian Spaces Around Subway Stations near Popular Tourist Attractions from the Tourists’ Perspective: A Case Study of Tianjin. Land 2025, 14, 145. [Google Scholar] [CrossRef]
  35. Zhang, D.; Song, K.; Zhao, D. Leveraging Multi-Source Data for the Trustworthy Evaluation of the Vibrancy of Child-Friendly Cities: A Case Study of Tianjin, China. Electronics 2024, 13, 4564. [Google Scholar] [CrossRef]
  36. Ronneberger, O.; Fischer, P.; Brox, T. U-Net: Convolutional Networks for Biomedical Image Segmentation. In Proceedings of the 2015 Medical Image Computing and Computer-Assisted Intervention, Munich, Germany, 5–9 October 2015. [Google Scholar]
  37. Dong, X.; Liu, Y.; Dai, J. Recognition of Concrete Surface Cracks Based on Improved TransUNet. Buildings 2025, 15, 541. [Google Scholar] [CrossRef]
  38. Draft of the Protection Plan for the Historical and Cultural City of Tianjin (2021–2035) Public Notice. Available online: https://ghhzrzy.tj.gov.cn/ywpd/cxgh_43015/ghgs/202211/t20221102_6024105.html (accessed on 20 March 2025).
Figure 1. Traditional Chinese architectural representatives: (a) Confucian Temple; (b) Tianhou Palace.
Figure 1. Traditional Chinese architectural representatives: (a) Confucian Temple; (b) Tianhou Palace.
Buildings 15 01142 g001
Figure 2. Western-style architectural representatives: (a) HSBC Building; (b) former site of the Astor Hotel building.
Figure 2. Western-style architectural representatives: (a) HSBC Building; (b) former site of the Astor Hotel building.
Buildings 15 01142 g002
Figure 3. Contemporary architectural representatives: (a) Tianjin Eye; (b) Jin Tower.
Figure 3. Contemporary architectural representatives: (a) Tianjin Eye; (b) Jin Tower.
Buildings 15 01142 g003
Figure 4. A dual-dimensional analytical framework for urban renewal based on multi-source data collaboration.
Figure 4. A dual-dimensional analytical framework for urban renewal based on multi-source data collaboration.
Buildings 15 01142 g004
Figure 5. Model optimization results: (a) The loss curve; (b) The mIoU curve.
Figure 5. Model optimization results: (a) The loss curve; (b) The mIoU curve.
Buildings 15 01142 g005
Figure 6. Rules for identifying types of plots involved in urban renewal activities in Tianjin from 2021 to 2024.
Figure 6. Rules for identifying types of plots involved in urban renewal activities in Tianjin from 2021 to 2024.
Buildings 15 01142 g006
Figure 7. The 2021–2024 Tianjin POI comprehensive kernel density analysis map: The highest density of the main core has increased by 99.7%, and the characteristics of “one core dominance, secondary core support, and multi-point coordination” have become more evident.
Figure 7. The 2021–2024 Tianjin POI comprehensive kernel density analysis map: The highest density of the main core has increased by 99.7%, and the characteristics of “one core dominance, secondary core support, and multi-point coordination” have become more evident.
Buildings 15 01142 g007
Figure 8. The 2021–2024 Tianjin kernel density analysis map of POIs with different functional types: Shows a clear trend of public space fission, improvement of public services, and industrial space restructuring.
Figure 8. The 2021–2024 Tianjin kernel density analysis map of POIs with different functional types: Shows a clear trend of public space fission, improvement of public services, and industrial space restructuring.
Buildings 15 01142 g008aBuildings 15 01142 g008b
Figure 9. Spatial texture and functional zoning of Tianjin central urban area: (a) Historical district; (b) Outside the historical district.
Figure 9. Spatial texture and functional zoning of Tianjin central urban area: (a) Historical district; (b) Outside the historical district.
Buildings 15 01142 g009
Figure 10. Detection of urban texture changes in Tianjin’s central urban area.
Figure 10. Detection of urban texture changes in Tianjin’s central urban area.
Buildings 15 01142 g010
Figure 11. Example of partial shadow occlusion in satellite images. (Image source: https://earth.google.com (accessed on 2 October 2024)).
Figure 11. Example of partial shadow occlusion in satellite images. (Image source: https://earth.google.com (accessed on 2 October 2024)).
Buildings 15 01142 g011
Table 1. Tianjin POI data statistics table.
Table 1. Tianjin POI data statistics table.
CategoryQuantityGrowth Rate
20212024
urban residential area11,43710,864−5.0%
industrial facilities55784262−23.6%
commercial facilities95,611353,776270.0%
business facilities46,60560,23829.3%
logistics and warehousing facilities80267233.8%
public service facilities49,14670,31843.1%
utilities51384521−12.0%
transportation facilities27,84742,72253.4%
green space17328162.4%
square and street space4812715464.4%
total242,096549,964127.2%
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Zhang, Y.; Chen, G.; Yang, X. Research on the Evolution Characteristics of Building Space in the Central Urban Area of Tianjin Based on Multi-Source Data Collaboration: 2021–2024. Buildings 2025, 15, 1142. https://doi.org/10.3390/buildings15071142

AMA Style

Zhang Y, Chen G, Yang X. Research on the Evolution Characteristics of Building Space in the Central Urban Area of Tianjin Based on Multi-Source Data Collaboration: 2021–2024. Buildings. 2025; 15(7):1142. https://doi.org/10.3390/buildings15071142

Chicago/Turabian Style

Zhang, Yicheng, Guorui Chen, and Xue Yang. 2025. "Research on the Evolution Characteristics of Building Space in the Central Urban Area of Tianjin Based on Multi-Source Data Collaboration: 2021–2024" Buildings 15, no. 7: 1142. https://doi.org/10.3390/buildings15071142

APA Style

Zhang, Y., Chen, G., & Yang, X. (2025). Research on the Evolution Characteristics of Building Space in the Central Urban Area of Tianjin Based on Multi-Source Data Collaboration: 2021–2024. Buildings, 15(7), 1142. https://doi.org/10.3390/buildings15071142

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop