Nothing Special   »   [go: up one dir, main page]
Next Issue
Volume 11, October
Previous Issue
Volume 11, August
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
6.9
2.8
Journals
IJGI
Volume 11
Issue 9
share announcement

ISPRS Int. J. Geo-Inf., Volume 11, Issue 9 (September 2022) – 38 articles

Cover Story (view full-size image): Gridded population datasets model the population at a relatively high spatial and temporal granularity by reallocating official population data from irregular administrative units to regular grids (e.g., 1 km grid cells). Such population data are vital for understanding human–environmental relationships and responding to many socioeconomic and environmental problems. We analyzed one very broadly used gridded population layer (GHS-POP) to assess its capacity to capture the distribution of population counts in several urban areas, spread across the major world regions. This analysis was performed to assess its suitability for global population modeling. View this paper
  • Issues are regarded as officially published after their release is announced to the table of contents alert mailing list.
  • You may sign up for e-mail alerts to receive table of contents of newly released issues.
  • PDF is the official format for papers published in both, html and pdf forms. To view the papers in pdf format, click on the "PDF Full-text" link, and use the free Adobe Reader to open them.
Order results
Result details
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
29 pages, 17805 KiB  
Article
Aquifer and Land Subsidence Interaction Assessment Using Sentinel-1 Data and DInSAR Technique
by Fatemeh Rafiei, Saeid Gharechelou, Saeed Golian and Brian Alan Johnson
ISPRS Int. J. Geo-Inf. 2022, 11(9), 495; https://doi.org/10.3390/ijgi11090495 - 19 Sep 2022
Cited by 12 | Viewed by 3133
Abstract
Climate change and overpopulation have led to an increase in water demands worldwide. As a result, land subsidence due to groundwater extraction and water level decline is causing damage to communities in arid and semiarid regions. The agricultural plain of Samalghan in Iran [...] Read more.
Climate change and overpopulation have led to an increase in water demands worldwide. As a result, land subsidence due to groundwater extraction and water level decline is causing damage to communities in arid and semiarid regions. The agricultural plain of Samalghan in Iran has recently experienced wide areas of land subsidence, which is hypothesized to be caused by groundwater overexploitation. This hypothesis was assessed by estimating the amount of subsidence that occurred in the Samalghan plain using DInSAR based on an analysis of 25 Sentinel-1 descending SAR images over 6 years. To assess the influence of water level changes on this phenomenon, groundwater level maps were produced, and their relationship with land subsidence was evaluated. Results showed that one major cause of the subsidence in the Samalghan plain was groundwater overexploitation, with the highest average land subsidence occurring in 2019 (34 cm) and the lowest in 2015 and 2018 (18 cm). Twelve Sentinel-1 ascending images were used for relative validation of the DInSAR processing. The correlation value varied from 0.69 to 0.89 (an acceptable range). Finally, the aquifer behavior was studied, and changes in cultivation patterns and optimal utilization of groundwater resources were suggested as practical strategies to control the current situation. Full article
(This article belongs to the Special Issue Geo-Information for Watershed Processes)
►▼ Show Figures

Figure 1

Figure 1
<p>Location map of study area in North Khorasan Province of Iran.</p> Full article ">Figure 2
<p>Percentage of different groundwater consumption sectors in Samalghan plain.</p> Full article ">Figure 3
<p>The location of drilling wells for the investigation of well logs (<b>a</b>); (<b>b</b>) the log of wells in AA’; (<b>c</b>) the log of wells in BB’ direction; and (<b>d</b>) the log of wells in CC’ section.</p> Full article ">Figure 3 Cont.
<p>The location of drilling wells for the investigation of well logs (<b>a</b>); (<b>b</b>) the log of wells in AA’; (<b>c</b>) the log of wells in BB’ direction; and (<b>d</b>) the log of wells in CC’ section.</p> Full article ">Figure 4
<p>Conventional DInSAR workflow in SNAP.</p> Full article ">Figure 5
<p>Variations in average groundwater level in Samalghan plain.</p> Full article ">Figure 6
<p>Groundwater level change map between 2008 and 2018.</p> Full article ">Figure 7
<p>Cumulative annual vertical displacement maps for 2015–2020 using descending data.</p> Full article ">Figure 8
<p>The trend of annual and cumulative land deformation in Ashkhane and Chamanbid cities.</p> Full article ">Figure 9
<p>Annual cumulative displacement maps for the ascending and descending data in (<b>a</b>) 2020, (<b>b</b>) 2019, and (<b>c</b>) 2018.</p> Full article ">Figure 10
<p>Evidence of land subsidence in the study area as cracks and holes (Photographs were taken at Ashkhane on 25 February, 2021 by Rafiei, F.).</p> Full article ">Figure 11
<p>(<b>a</b>) Change in water level and position of the piezometric wells overlaid with subsidence in the region from 2015 to 2020, with purple lines as section. (<b>b</b>) Sections in the aquifer showing the groundwater level drop and displacement plots.</p> Full article ">Figure 12
<p>Temporal evolution of deformation (InSAR estimations) (orange line), groundwater level change (blue line) at (<b>a</b>) well 1, (<b>b</b>) well 14, (<b>c</b>) well 15, and (<b>d</b>) well 16, and the groundwater level change with a one-year lag in (<b>b</b>,<b>c</b>) (gray-dashed line).</p> Full article ">Figure 13
<p>The value of the interpolated storage coefficient of the Samalghan aquifer, using the ordinary Kriging method.</p> Full article ">Figure 14
<p>Buffer areas with different radius values and locations around well fields in the study area on the displacement maps.</p> Full article ">Figure 15
<p>Graph of average surface changes at a distance of 250 m from observation wells (green and orange-dashed rectangles representing recovery and extraction periods, respectively).</p> Full article ">Figure 16
<p>Stress–strain analysis for wells W3 (<b>a</b>), W4 (<b>b</b>), W8 (<b>c</b>), W 12 (<b>d</b>), W13 (<b>e</b>), and W17 (<b>f</b>).</p> Full article ">Figure 17
<p>Geological map of the Samalghan plain [<a href="#B42-ijgi-11-00495" class="html-bibr">42</a>].</p> Full article ">Figure 18
<p>Soft soil thickness map of the Samalghan plain (<b>a</b>); the line shows the section used for investigating the soft soil section along the southwest–northeast direction (<b>b</b>).</p> Full article ">
25 pages, 38383 KiB  
Article
Reverse Difference Network for Highlighting Small Objects in Aerial Images
by Huan Ni, Jocelyn Chanussot, Xiaonan Niu, Hong Tang and Haiyan Guan
ISPRS Int. J. Geo-Inf. 2022, 11(9), 494; https://doi.org/10.3390/ijgi11090494 - 18 Sep 2022
Cited by 2 | Viewed by 1899
Abstract
The large-scale variation issue in high-resolution aerial images significantly lowers the accuracy of segmenting small objects. For a deep-learning-based semantic segmentation model, the main reason is that the deeper layers generate high-level semantics over considerably large receptive fields, thus improving the accuracy for [...] Read more.
The large-scale variation issue in high-resolution aerial images significantly lowers the accuracy of segmenting small objects. For a deep-learning-based semantic segmentation model, the main reason is that the deeper layers generate high-level semantics over considerably large receptive fields, thus improving the accuracy for large objects but ignoring small objects. Although the low-level features extracted by shallow layers contain small-object information, large-object information has predominant effects. When the model, using low-level features, is trained, the large objects push the small objects aside. This observation motivates us to propose a novel reverse difference mechanism (RDM). The RDM eliminates the predominant effects of large objects and highlights small objects from low-level features. Based on the RDM, a novel semantic segmentation method called the reverse difference network (RDNet) is designed. In the RDNet, a detailed stream is proposed to produce small-object semantics by enhancing the output of RDM. A contextual stream for generating high-level semantics is designed by fully accumulating contextual information to ensure the accuracy of the segmentation of large objects. Both high-level and small-object semantics are concatenated when the RDNet performs predictions. Thus, both small- and large-object information is depicted well. Two semantic segmentation benchmarks containing vital small objects are used to fully evaluate the performance of the RDNet. Compared with existing methods that exhibit good performance in segmenting small objects, the RDNet has lower computational complexity and achieves 3.9–18.9% higher accuracy in segmenting small objects. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>The scale variations in aerial images, (<b>a</b>,<b>b</b>) are the image patches in UAVid, and (<b>c</b>,<b>d</b>) are the image patches in Aeroscapes. Pedestrians, bikes, and drones with small sizes are marked by red rectangles.</p> Full article ">Figure 2
<p>The saliency maps of the low-level features and the high-level features (semantics). The low-level features (ResNet18 [<a href="#B11-ijgi-11-00494" class="html-bibr">11</a>]) and the high-level features (ResNet18) are extracted by the first and fourth inner layers of ResNet18, respectively. The low-level features (BiSeNetV2) and the high-level semantics (BiSeNetV2) are extracted by the detail and semantic branches of BiSeNetV2, respectively.</p> Full article ">Figure 3
<p>The architecture of reverse difference network. Down and Up are the interpolation operators for down-sampling and up-sampling.</p> Full article ">Figure 4
<p>The saliency maps of the features extracted by the inner layers in ResNet18. (<b>a</b>) is the input image patch, (<b>b</b>–<b>e</b>) are the saliency maps of the <math display="inline"><semantics> <msubsup> <mi>f</mi> <mn>1</mn> <mi>b</mi> </msubsup> </semantics></math>, <math display="inline"><semantics> <msubsup> <mi>f</mi> <mn>2</mn> <mi>b</mi> </msubsup> </semantics></math>, <math display="inline"><semantics> <msubsup> <mi>f</mi> <mn>3</mn> <mi>b</mi> </msubsup> </semantics></math>, and <math display="inline"><semantics> <msubsup> <mi>f</mi> <mn>4</mn> <mi>b</mi> </msubsup> </semantics></math> extracted by <math display="inline"><semantics> <mrow> <mi>L</mi> <msub> <mi>Y</mi> <mn>1</mn> </msub> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>L</mi> <msub> <mi>Y</mi> <mn>2</mn> </msub> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>L</mi> <msub> <mi>Y</mi> <mn>3</mn> </msub> </mrow> </semantics></math>, and <math display="inline"><semantics> <mrow> <mi>L</mi> <msub> <mi>Y</mi> <mn>4</mn> </msub> </mrow> </semantics></math> in ResNet18, respectively.</p> Full article ">Figure 5
<p>The proposed reverse difference mechanism. The <math display="inline"><semantics> <mrow> <mi>S</mi> <mo>(</mo> <mo>·</mo> <mo>)</mo> </mrow> </semantics></math> is the sigmoid function.</p> Full article ">Figure 6
<p>The feature alignment as Equation (<a href="#FD6-ijgi-11-00494" class="html-disp-formula">6</a>) in the cosine alignment.</p> Full article ">Figure 7
<p>The detailed stream (DS). The input of DS is the difference features <math display="inline"><semantics> <msup> <mi>f</mi> <mi>d</mi> </msup> </semantics></math> extracted by RDM, and it generates small-object semantics <math display="inline"><semantics> <msup> <mi>f</mi> <mi>s</mi> </msup> </semantics></math>. Conv, convolution layer; DW Conv, depthwise convolution layer; APool, adaptive average pooling; Sigmoid and ReLU, activation functions; Expand, expanding based on duplication.</p> Full article ">Figure 8
<p>The pyramid structure of the contextual stream (CS). The input of CS is <math display="inline"><semantics> <msubsup> <mi>f</mi> <mn>4</mn> <mi>b</mi> </msubsup> </semantics></math> extracted by the <math display="inline"><semantics> <mrow> <mi>L</mi> <msub> <mi>Y</mi> <mn>4</mn> </msub> </mrow> </semantics></math> in the backbone network, and the high-level semantics <math display="inline"><semantics> <msup> <mi>f</mi> <mi>h</mi> </msup> </semantics></math> are generated. APool, adaptive average pooling; Conv, convolution layer; DW Conv, depth-wise convolution layer; Up, up-sampling; Concat, concatenation; PAM, position attention module [<a href="#B26-ijgi-11-00494" class="html-bibr">26</a>]; CAM, channel attention module [<a href="#B26-ijgi-11-00494" class="html-bibr">26</a>].</p> Full article ">Figure 9
<p>The diverse landscapes in UAVid, (<b>a</b>) downtown area, (<b>b</b>) villa area, and (<b>c</b>) outskirt.</p> Full article ">Figure 10
<p>The diverse landscapes in Aeroscapes, (<b>a</b>) countryside, (<b>b</b>) playground, (<b>c</b>) farmland, (<b>d</b>) downtown area, (<b>e</b>) road, (<b>f</b>) animal zoo, (<b>g</b>) human settlement, and (<b>h</b>) seascape.</p> Full article ">Figure 11
<p>The visual comparisons for the UAVid dataset.</p> Full article ">Figure 12
<p>Visual comparisons for the Aeroscapes dataset.</p> Full article ">Figure 13
<p>The saliency map of the features produced by each module in RDNet.</p> Full article ">
20 pages, 3173 KiB  
Article
MSEN-GRP: A Geographic Relations Prediction Model Based on Multi-Layer Similarity Enhanced Networks for Geographic Relations Completion
by Zongcai Huang, Peiyuan Qiu, Li Yu and Feng Lu
ISPRS Int. J. Geo-Inf. 2022, 11(9), 493; https://doi.org/10.3390/ijgi11090493 - 17 Sep 2022
Cited by 3 | Viewed by 1936
Abstract
Geographic relation completion contributes greatly to improving the quality of large-scale geographic knowledge graphs (GeoKGs). However, the internal features of a GeoKG used in large-scale GeoKGs embedding are often limited by the weak connectivity between geographic entities (geo-entities). If there is no proper [...] Read more.
Geographic relation completion contributes greatly to improving the quality of large-scale geographic knowledge graphs (GeoKGs). However, the internal features of a GeoKG used in large-scale GeoKGs embedding are often limited by the weak connectivity between geographic entities (geo-entities). If there is no proper choice in the method of external semantic enhancement, this will often interfere with the representation and learning of the KG. Therefore, we here propose a geographic relation (geo-relation) prediction model based on multi-layer similarity enhanced networks for geo-relations completion (MSEN-GRP). The MSEN-GRP comprises three parts: enhancer, encoder, and decoder. The enhancer constructs semantic, spatial, structural, and attribute-similarity networks for geo-entities, which can explicitly and effectively enhance the implicit semantic associations between existing geo-entities. The encoder can obtain the long path relation dependency characteristics of geo-entities using a mixed-path sampling strategy and can support different optimization schemes for external semantic enhancement. Geo-relations prediction experiments show that the mean reciprocal ranking of this method is significantly higher than those of the traditional TransE DisMult and methods, and Hits@10 is improved by up to 57.57%. Furthermore, the spatial-similarity network has the most significant enhancement effect on geo-relations prediction. The proposed method provides a new way to perform relation completion in sparse GeoKGs. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>An example visualization of a GeoKG.</p> Full article ">Figure 2
<p>Framework of MSEN-GRP.</p> Full article ">Figure 3
<p>Lexical similarity calculation schemes.</p> Full article ">Figure 4
<p>Visual space distance model.</p> Full article ">Figure 5
<p>Method of constructing the structural similarity network.</p> Full article ">Figure 6
<p>Representation of the encoder.</p> Full article ">Figure 7
<p>Number of all relationship types appearing in the GeoDBpedia21 dataset.</p> Full article ">Figure 8
<p>Distribution of entities with different degrees.</p> Full article ">Figure 9
<p>Graphical visualization of the GeoDBpedia21 dataset.</p> Full article ">
18 pages, 5606 KiB  
Article
Sensing Tourist Distributions and Their Sentiment Variations Using Social Media: Evidence from 5A Scenic Areas in China
by Jingbo Wang, Yu Xia and Yuting Wu
ISPRS Int. J. Geo-Inf. 2022, 11(9), 492; https://doi.org/10.3390/ijgi11090492 - 17 Sep 2022
Cited by 4 | Viewed by 3825
Abstract
The distribution and sentiment characteristics of tourists directly reflect the state of tourism development, and are an important reference for tourists to choose scenic areas. Sensing the tourist distributions and their sentiment variations can provide decision support for the development planning of scenic [...] Read more.
The distribution and sentiment characteristics of tourists directly reflect the state of tourism development, and are an important reference for tourists to choose scenic areas. Sensing the tourist distributions and their sentiment variations can provide decision support for the development planning of scenic areas. In this study, we crawled tourist social media data to explore tourist distribution characteristics and the patterns of tourist sentiment variations. First, we used web crawlers to obtain social media data (tourist comment data) and the location data of China’s 5A scenic areas from the Ctrip tourism platform. Second, SnowNLP (Simplified Chinese Text Processing) was optimized and used to classify the sentiment of tourists’ comments and calculate the sentiment value. Finally, we mined the distribution characteristics of tourists in 5A scenic areas and the spatio-temporal variations in tourists’ sentiments. The results show that: (1) There is a negative correlation between the number of tourists to China’s 5A scenic areas and tourist sentiment: the number of tourists is highest in October and lowest in March, while tourist sentiment is highest in March and lowest in October. (2) The spatio-temporal distribution of tourists has obvious aggregation: temporally mainly in July, August and October, spatially mainly in the Yangtze River Delta city cluster, Beijing-Tianjin-Hebei city cluster, and Guanzhong Plain city cluster. (3) Tourist sentiment cold/hot spots vary significantly by city clusters: the Yangtze River Delta city cluster is always a sentiment hot spot; the northern city cluster has more sentiment cold spots; the central city cluster varies significantly during the year; the southwestern city cluster has more sentiment hot spots. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>Distribution of 5A scenic areas in China.</p> Full article ">Figure 2
<p>Research framework.</p> Full article ">Figure 3
<p>Sentiment classification process.</p> Full article ">Figure 4
<p>Monthly variation in the number of tourist comments (<b>a</b>) and sentiment values (<b>b</b>).</p> Full article ">Figure 5
<p>Spatial information on the distribution of scenic areas (<b>a</b>) and tourists (<b>b</b>) (kernel density and standard deviation ellipse).</p> Full article ">Figure 6
<p>Distribution of trends in tourist sentiment (<b>a</b>) and local abnormalities in sentiment (<b>b</b>).</p> Full article ">Figure 7
<p>Distribution of monthly variations in tourists’ sentimental cold spots and hot spots.</p> Full article ">
20 pages, 7139 KiB  
Article
Design and Application of Multi-Dimensional Visualization System for Large-Scale Ocean Data
by Teng Lv, Jun Fu and Bao Li
ISPRS Int. J. Geo-Inf. 2022, 11(9), 491; https://doi.org/10.3390/ijgi11090491 - 16 Sep 2022
Cited by 4 | Viewed by 2929
Abstract
With the constant deepening of research on marine environment simulation and information expression, there are higher and higher requirements for the sense of the reality of ocean data visualization results and the real-time interaction in the visualization process. Aiming at the challenges of [...] Read more.
With the constant deepening of research on marine environment simulation and information expression, there are higher and higher requirements for the sense of the reality of ocean data visualization results and the real-time interaction in the visualization process. Aiming at the challenges of 3D interactive key technology and GPU-based visualization algorithm technology, we developed a visualization system for large-scale 3D marine environmental data. The system realizes submarine terrain rendering, contour line visualization, isosurface visualization, section visualization, volume visualization and flow field visualization. In order to manage and express the data in the system, we developed a data management module, which can effectively integrate a large number of marine environmental data and manage them effectively. We developed a series of data analysis functions for the system, such as point query and line query, local analysis and multi-screen collaboration, etc. These functions can effectively improve the data analysis efficiency of users and meet the data analysis needs in multiple scenarios. The marine environmental data visualization system developed in this paper can efficiently and intuitively simulate and display the nature and changing process of marine water environmental factors. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>System framework.</p> Full article ">Figure 2
<p>Data management module framework.</p> Full article ">Figure 3
<p>Flow chart for constructing seafloor topography DEM data.</p> Full article ">Figure 4
<p>Framework for submarine 3D terrain simulation.</p> Full article ">Figure 5
<p>Submarine terrain visualization.</p> Full article ">Figure 6
<p>Two equivalent points.</p> Full article ">Figure 7
<p>Four equivalent points. (<b>a</b>,<b>b</b>) are two possibilities, respectively.</p> Full article ">Figure 8
<p>Contour line visualization. (<b>a</b>,<b>b</b>) are contour distributions at different depths.</p> Full article ">Figure 9
<p>Isosurface visualization. (<b>a</b>) is the debugging interface, (<b>b</b>) is the area display.</p> Full article ">Figure 10
<p>Section visualization.</p> Full article ">Figure 11
<p>Pre-integration piecewise linear sampling principle.</p> Full article ">Figure 12
<p>Pre-integrated volume visualization implementation steps.</p> Full article ">Figure 13
<p>Volume visualization.</p> Full article ">Figure 14
<p>Ocean current visualization.</p> Full article ">Figure 15
<p>Point query (<b>a</b>) and line query (<b>b</b>) renderings.</p> Full article ">Figure 16
<p>Local analysis renderings ((<b>a</b>) is surface rendering, (<b>b</b>) is volume rendering).</p> Full article ">Figure 16 Cont.
<p>Local analysis renderings ((<b>a</b>) is surface rendering, (<b>b</b>) is volume rendering).</p> Full article ">Figure 17
<p>Multi-screen display effect. (<b>a</b>) is the overview of the multi-screen display, (<b>b</b>) is the multi-screen display in the submarine scene.</p> Full article ">Figure 17 Cont.
<p>Multi-screen display effect. (<b>a</b>) is the overview of the multi-screen display, (<b>b</b>) is the multi-screen display in the submarine scene.</p> Full article ">
25 pages, 8585 KiB  
Article
Climate Justice in the City: Mapping Heat-Related Risk for Climate Change Mitigation of the Urban and Peri-Urban Area of Padua (Italy)
by Valeria Todeschi, Salvatore Eugenio Pappalardo, Carlo Zanetti, Francesca Peroni and Massimo De Marchi
ISPRS Int. J. Geo-Inf. 2022, 11(9), 490; https://doi.org/10.3390/ijgi11090490 - 16 Sep 2022
Cited by 13 | Viewed by 4340
Abstract
The mitigation of urban heat islands (UHIs) is crucial for promoting the sustainable development of urban areas. Geographic information systems (GISs) together with satellite-derived data are powerful tools for investigating the spatiotemporal distribution of UHIs. Depending on the availability of data and the [...] Read more.
The mitigation of urban heat islands (UHIs) is crucial for promoting the sustainable development of urban areas. Geographic information systems (GISs) together with satellite-derived data are powerful tools for investigating the spatiotemporal distribution of UHIs. Depending on the availability of data and the geographic scale of the analysis, different methodologies can be adopted. Here, we show a complete open source GIS-based methodology based on satellite-driven data for investigating and mapping the impact of the UHI on the heat-related elderly risk (HERI) in the Functional Urban Area of Padua. Thermal anomalies in the territory were mapped by modelling satellite data from Sentinel-3. After a socio-demographic analysis, the HERI was mapped according to five levels of risk. The highest vulnerability levels were localised within the urban area and in three municipalities near Padua, which represent about 20% of the entire territory investigated. In these municipalities, a percentage of elderly people over 20%, a thermal anomaly over 2.4 °C, and a HERI over 0.65 were found. Based on these outputs, it is possible to define nature-based solutions for reducing the UHI phenomenon and promote a sustainable development of cities. Stakeholders can use the results of these investigations to define climate and environmental policies. Full article
(This article belongs to the Special Issue Climate Adaptation: The Role of Geospatial Data in Sustainable Infrastructures)
►▼ Show Figures

Figure 1

Figure 1
<p>Functional Urban Area of Padua (530,322 inhabitants), including 31 municipalities (Veneto Region, NE Italy).</p> Full article ">Figure 2
<p>Flowchart of the GIS-based methodology.</p> Full article ">Figure 3
<p>Map of the land surface temperature in the Functional Urban Area of Padua (NE Italy) based on four Sentinel-3 scenes (100 m raster resolution output).</p> Full article ">Figure 4
<p>NDVI map (median value) geovisualising green areas in the Functional Urban Area of Padua (100 m raster resolution output).</p> Full article ">Figure 5
<p>Map of the percentage of green surfaces with a 1 km<sup>2</sup> grid size in the Functional Urban Area of Padua in Italy using the Corinne Land Cover (CLC) in 2018.</p> Full article ">Figure 6
<p>Map of urban thermal anomalies in the Functional Urban Area of Padua in Italy during the summers of 2018–2021 (100 m raster resolution output).</p> Full article ">Figure 7
<p>Map of the heat-related elderly risk index (<span class="html-italic">HERI</span>) in the Functional Urban Area of Padua in Italy during the summers of 2018–2021 (normalised values from 0 to 1).</p> Full article ">Figure A1
<p>Map of the LSTs in the Functional Urban Area of Padua in Italy based on four Sentinel-3 scenes (1 km raster resolution output).</p> Full article ">Figure A2
<p>Map of LSTs at the municipal level in the Functional Urban Area of Padua in Italy during the summers of 2018–2021 (June–August).</p> Full article ">Figure A3
<p>NDVI map (median value) geovisualising green areas in the Functional Urban Area of Padua (1 km raster resolution output).</p> Full article ">Figure A4
<p>Map of urban thermal anomalies in the Functional Urban Area of Padua in Italy during the summers of 2018–2021 (1 km raster resolution output).</p> Full article ">Figure A5
<p>Map of urban thermal anomalies at the municipal level in the Functional Urban Area of Padua in Italy during the summers of 2018–2021 (June–August).</p> Full article ">Figure A6
<p>Map of the exposure (E) parameter in the Functional Urban Area of Padua in Italy during the summers of 2018–2021 (normalised values from 0 to 1).</p> Full article ">Figure A7
<p>Map of the vulnerability (V) parameter in the Functional Urban Area of Padua in Italy during the summers of 2018–2021 (normalised values from 0 to 1).</p> Full article ">Figure A8
<p>Map of the hazard (H) parameter in the Functional Urban Area of Padua in Italy during the summers of 2018–2021 (normalised values from 0 to 1).</p> Full article ">
20 pages, 3286 KiB  
Article
Generalization of Linear and Area Features Incorporating a Shape Measure
by Natalia Blana and Lysandros Tsoulos
ISPRS Int. J. Geo-Inf. 2022, 11(9), 489; https://doi.org/10.3390/ijgi11090489 - 16 Sep 2022
Cited by 2 | Viewed by 1510
Abstract
This article elaborates on the quality issue in cartographic generalization of linear and area features focusing on the assessment of shape preservation. Assessing shape similarity in generalization is still a topic where further research is required. In the study presented here, shape description [...] Read more.
This article elaborates on the quality issue in cartographic generalization of linear and area features focusing on the assessment of shape preservation. Assessing shape similarity in generalization is still a topic where further research is required. In the study presented here, shape description and matching techniques are investigated and analyzed, a procedure for choosing generalization parameters suitable for line and area features depiction is described and a quality model is developed for the assessment and verification of the generalization results. Based on the procedure developed, cartographers will be confident that the generalization of linear and area features is appropriate for a specific scale of portrayal fulfilling on the same time a basic requirement in generalization, that of shape preservation. The results of the procedure developed are based on the processing and successful generalization of a large number of different line and area features that is supported by a software environment developed in Python programming language. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>Typology of shape description and representation techniques [<a href="#B19-ijgi-11-00489" class="html-bibr">19</a>].</p> Full article ">Figure 2
<p>Angles in cumulative angular function (redrawn) [<a href="#B30-ijgi-11-00489" class="html-bibr">30</a>].</p> Full article ">Figure 3
<p>Angles in the complex-valued exponential function of the total curvature (redrawn) [<a href="#B32-ijgi-11-00489" class="html-bibr">32</a>].</p> Full article ">Figure 4
<p>Selection of the generalized features suitable for portrayal: Select the one among the representatives that corresponds to the maximum tolerance value (operation-horizontal accuracy evaluation [<a href="#B38-ijgi-11-00489" class="html-bibr">38</a>]).</p> Full article ">Figure 5
<p>Line features generalized with the Douglas–Peucker simplification algorithm (red) and bend simplification algorithm (blue). Original lines are depicted in black (scale 1:500,000).</p> Full article ">Figure 6
<p>Line features generalized with the Douglas–Peucker simplification algorithm (red) and bend simplification algorithm (blue). Original lines are depicted in black (scale 1:1,000,000).</p> Full article ">Figure 7
<p>Polygon features simplified with the Douglas–Peucker simplification algorithm (red) and bend simplification algorithm (blue). Original polygons are depicted in black (scale 1: 500,000).</p> Full article ">Figure 8
<p>Polygon features generalized with the Douglas–Peucker simplification algorithm (red) and bend simplification algorithm (blue). Original polygons are depicted in black (scale 1: 1,000,000).</p> Full article ">Figure 9
<p>Initial road network and built areas (no generalization applied) at scale 1: 250,000.</p> Full article ">Figure 10
<p>Initial road network and built areas (no generalization is applied) at scale 1: 500,000.</p> Full article ">Figure 11
<p>Road network and built areas generalized with Douglas–Peucker simplification algorithm at scale 1: 500,000.</p> Full article ">Figure 12
<p>Section of the area enlarged to highlight the differences between before and after generalization.</p> Full article ">
19 pages, 5871 KiB  
Article
PM2SFCA: Spatial Access to Urban Parks, Based on Park Perceptions and Multi-Travel Modes. A Case Study in Beijing
by Shijia Luo, Heping Jiang, Disheng Yi, Ruihua Liu, Jiahui Qin, Yusi Liu and Jing Zhang
ISPRS Int. J. Geo-Inf. 2022, 11(9), 488; https://doi.org/10.3390/ijgi11090488 - 15 Sep 2022
Cited by 8 | Viewed by 2346
Abstract
Assessing park accessibility plays an essential role in providing rational recreational services for residents in a city. The perceptions and comments of residents are also important nonspatial factors for accessibility. However, there are few accessibility studies that are combined with public perceptions. Addressing [...] Read more.
Assessing park accessibility plays an essential role in providing rational recreational services for residents in a city. The perceptions and comments of residents are also important nonspatial factors for accessibility. However, there are few accessibility studies that are combined with public perceptions. Addressing this deficit, this study proposes a perception-based, multi-travel mode, two-step floating catchment area (PM2SFCA) method to calculate park accessibility. First, we quantified the selection probability of residents to parks by integrating the Huff model and the people’s perceptions towards parks. Next, under four travel modes (walking, biking, driving and public transport), we combined the Huff model and the two-step floating catchment area method to compute park accessibility. Furthermore, the Gini coefficient and the Pearson correlation coefficient were used to illustrate the proposed method compared with the traditional E2SFCA method. Based on the above, taking the area of Beijing within the Fifth Ring Road as a study area, this paper facilitated the accessibility computation. The results indicated that the spatial distribution patterns of accessibility differed greatly under the four travel modes. Even under the same travel mode, there was an uneven accessibility distribution. Areas with high accessibility were mainly concentrated in the north, and some marginal areas also presented higher accessibility to parks. The comparative analysis results suggest that our proposed method for accessibility measurements alleviates the underestimation and overestimation of accessibility values obtained by a traditional method such as the center and edge of the study area. The research explores a new research perspective for measuring park accessibility. Furthermore, this study offers better guidance for policymakers trying to optimize park spatial distribution issues. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>The framework of this research.</p> Full article ">Figure 2
<p>The study area in Beijing, China.</p> Full article ">Figure 3
<p>The research unit, parks and TAZ population.</p> Full article ">Figure 4
<p>An example of selection probability with park perceptions and distance decay.</p> Full article ">Figure 5
<p>The sentiment scores and frequencies of urban parks’ online reviews.</p> Full article ">Figure 6
<p>The area and public perceptions of parks.</p> Full article ">Figure 7
<p>The accessibility results of walking.</p> Full article ">Figure 8
<p>The accessibility results of biking.</p> Full article ">Figure 9
<p>The accessibility results of driving.</p> Full article ">Figure 10
<p>The accessibility results of public transport.</p> Full article ">Figure 11
<p>Local indication of spatial association (LISA) cluster map of accessibility index, under four travel modes.</p> Full article ">Figure 12
<p>The difference under four travel modes by the PM2SFCA method and E2SFCA method.</p> Full article ">Figure 13
<p>Scatter plot among the accessibility values by the PM2SFCA and E2SFCA methods ((<b>a</b>) biking mode; (<b>b</b>) driving mode; and (<b>c</b>) public transport mode).</p> Full article ">Figure 14
<p>The Gini index and Lorenz curve of park accessibility.</p> Full article ">
14 pages, 3490 KiB  
Article
Road Intersection Recognition via Combining Classification Model and Clustering Algorithm Based on GPS Data
by Yizhi Liu, Rutian Qing, Yijiang Zhao and Zhuhua Liao
ISPRS Int. J. Geo-Inf. 2022, 11(9), 487; https://doi.org/10.3390/ijgi11090487 - 14 Sep 2022
Cited by 11 | Viewed by 3481
Abstract
Road intersections are essential to road networks. How to precisely recognize road intersections based on GPS data is still challenging in intelligent transportation systems. Road intersection recognition involves detecting intersections and recognizing its scope. There are few works on intersections’ scope recognition. The [...] Read more.
Road intersections are essential to road networks. How to precisely recognize road intersections based on GPS data is still challenging in intelligent transportation systems. Road intersection recognition involves detecting intersections and recognizing its scope. There are few works on intersections’ scope recognition. The existing methods always focus on road intersection detection. It includes two parts: one is selecting turning points from GPS data and extracting their geometric features, another is clustering them into center coordinates of road intersections. However, the accuracy of road intersection detection still has improvement room due to two drawbacks: (1) Besides geometric features, spatial features explored from GPS data and the interactions among all features are also important to represent intersections’ semantics more accurately, and (2) How to capture the points around intersections for clustering has great impact on the accuracy of intersection detection. To solve the preceding problems, we propose a novel approach for road intersection recognition via combining a classification model and clustering algorithm based on GPS data, which involves detecting the center coordinate and computing the radius of the intersection. Firstly, we distil geometric features and spatial features from historical GPS points. These features are inputted into the Extreme Deep Factorization Machine (xDeepFM) model which is applied for capturing the GPS points nearby road intersections. Secondly, the preceding points are clustered into center coordinates of road intersections by the Density-Based Spatial Clustering of Applications with Noise algorithm (DBSCAN). Thirdly, we present a new method of radius computing by integrating Delaunay triangulation with circle shape structure. Experiments are carried out on the GPS data of Chengdu, China. Compared with some state-of-the-art methods, our approach achieves higher accuracy on road intersection recognition based on GPS data. The precision, recall, and f-measure of our proposed center coordinates detection method are respectively 99.0%, 92.7%, and 95.8% when the matching area’s radius is 30 m. Moreover, the error of the proposed radius calculation method is less than 26.5%. Full article
(This article belongs to the Special Issue Urban Geospatial Analytics Based on Crowdsourced Data)
►▼ Show Figures

Figure 1

Figure 1
<p>The framework of the proposed method.</p> Full article ">Figure 2
<p>The point B’s turning angle θ. The points A, B, and C are GPS points recorded in time order.</p> Full article ">Figure 3
<p>Turning distances in two situations: (<b>a</b>) Turning distance when a taxi changes direction; (<b>b</b>) Turning distance when a taxi goes straight. The points A, B, and C are GPS points recorded in time order.</p> Full article ">Figure 4
<p>Element values of eight neighborhoods.</p> Full article ">Figure 5
<p>The feature matrix constructed by geometric features, spatial features, and labels.</p> Full article ">Figure 6
<p>The structure of the xDeepFM model.</p> Full article ">Figure 7
<p>The radius of a road intersection: (<b>a</b>) X shape road intersections; (<b>b</b>) T shape road intersections.</p> Full article ">Figure 8
<p>Integrating Delaunay triangulation algorithm with the circle shape: (<b>a</b>) Before deleting outlier points; (<b>b</b>) After deleting outlier points.</p> Full article ">Figure 9
<p>The road intersections detected by the proposed method in the experimental area.</p> Full article ">Figure 10
<p>Radius computing of a road intersection.</p> Full article ">Figure 11
<p>Performance comparison between Tang’s method [<a href="#B2-ijgi-11-00487" class="html-bibr">2</a>] and our method of radius computing and center coordinate detection: (<b>a</b>) The average error of computing radius of intersections; (<b>b</b>) The accuracy of detecting center coordinate of intersections.</p> Full article ">Figure 12
<p>Performance comparison of three typical cluster algorithms.</p> Full article ">Figure 13
<p>Performance comparison of different feature matrixes.</p> Full article ">
18 pages, 6193 KiB  
Article
Exploring the Inter-Monthly Dynamic Patterns of Chinese Urban Spatial Interaction Networks Based on Baidu Migration Data
by Heping Jiang, Shijia Luo, Jiahui Qin, Ruihua Liu, Disheng Yi, Yusi Liu and Jing Zhang
ISPRS Int. J. Geo-Inf. 2022, 11(9), 486; https://doi.org/10.3390/ijgi11090486 - 14 Sep 2022
Cited by 6 | Viewed by 2609
Abstract
The rapid development of the economy promotes the increasing of interactions between cities and forms complex networks. Many scholars have explored the structural characteristics of urban spatial interaction networks in China and have conducted spatio-temporal analyzes. However, scholars have mainly focused on the [...] Read more.
The rapid development of the economy promotes the increasing of interactions between cities and forms complex networks. Many scholars have explored the structural characteristics of urban spatial interaction networks in China and have conducted spatio-temporal analyzes. However, scholars have mainly focused on the perspective of static networks and have not understood the dynamic spatial interaction patterns of Chinese cities. Therefore, this paper proposes a research framework to explore the urban dynamic spatial interaction patterns. Firstly, we establish a dynamic urban spatial interaction network according to monthly migration data. Then, the dynamic community detection algorithm, combined with the Louvain and Jaccard matching method, is used to obtain urban communities and their dynamic events. We construct event vectors for each urban community and use hierarchical clustering to cluster event vectors to obtain different types of spatial interaction patterns. Finally, we divide the urban dynamic interaction into three urban spatial interaction modes: fixed spatial interaction pattern, long-term spatial interaction pattern, and short-term spatial interaction pattern. According to the results, we find that the cities in well-developed areas (eastern China) and under-developed areas (northwestern China) mostly show fixed spatial interaction patterns and long-term spatial interaction patterns, while the cities in moderately developed areas (central and western China) often show short-term spatial interaction patterns. The research results and conclusions of this paper reveal the inter-monthly urban spatial interaction patterns in China, provide theoretical support for the policy making and development planning of urban agglomeration construction, and contribute to the coordinated development of national and regional cities. Full article
(This article belongs to the Special Issue Urban Geospatial Analytics Based on Crowdsourced Data)
►▼ Show Figures

Figure 1

Figure 1
<p>Study area. Administrative divisions of provinces and cities in China.</p> Full article ">Figure 2
<p>The technical route diagram.</p> Full article ">Figure 3
<p>The process of the Louvain algorithm. (<b>a</b>) is the initial setup of the network, assigning all nodes as separate communities, (<b>b</b>) is the result of local modularity optimization, where different colors mean different communities, and (<b>c</b>) is the result of folding communities into new nodes, forming a new network, where the new network contains edges not only between nodes but also within nodes.</p> Full article ">Figure 4
<p>The urban communities of each month. Each sub-plot represents the information of the urban communities in that month.</p> Full article ">Figure 5
<p>The distribution of Jaccard scores.</p> Full article ">Figure 6
<p>The numbers of five dynamic events in each period.</p> Full article ">Figure 7
<p>The dendrogram of the hierarchical clustering.</p> Full article ">Figure 8
<p>The silhouette coefficient under the different numbers of clusters.</p> Full article ">Figure 9
<p>Number of times urban communities existed.</p> Full article ">Figure 10
<p>The change times in urban community affiliation for each city.</p> Full article ">
24 pages, 8258 KiB  
Article
Exploring Spatial Features of Population Activities and Functional Facilities in Rail Transit Station Realm Based on Real-Time Positioning Data: A Case of Xi’an Metro Line 2
by Di Wang, Bart Dewancker, Yaqiong Duan and Meng Zhao
ISPRS Int. J. Geo-Inf. 2022, 11(9), 485; https://doi.org/10.3390/ijgi11090485 - 14 Sep 2022
Cited by 4 | Viewed by 2916
Abstract
The rail transit station realm is an important urban spatial node that carries various behavioral activities and multiple functions. In order to accurately identify the spatial and temporal distribution of population activities and functional facilities in the rail transit station realm and understand [...] Read more.
The rail transit station realm is an important urban spatial node that carries various behavioral activities and multiple functions. In order to accurately identify the spatial and temporal distribution of population activities and functional facilities in the rail transit station realm and understand the dynamic influence relationship between them, this paper takes four different types of stations of Xi’an Metro Line 2 as the research object, using real-time positioning data to represent population activities and points of interest (POIs) to represent functional facilities. An analytical framework combining the spatial point pattern identification technique and ordinary least squares (OLS) regression model is proposed. The results show that (1) there is spatial and temporal heterogeneity in the population activities in the rail transit station realm; the density distribution of population activities in different time periods shows the characteristic of clustering within 500 m of the station, regardless of working days or off days; (2) the distribution of shopping service POI, catering service POI, and living service POI in different station realms shows the feature of clustering around the stations; (3) the catering POI, living POI, shopping POI and transportation POI have positive attraction to population activities in different time periods; the constructed OLS model can basically explain the influence relationship between various functional facilities and population activities in all time periods. The conclusions can help city managers understand the spatial and temporal distribution and intrinsic mechanisms of population activities and functional facilities from a microscopic perspective and provide an effective decision-making basis for optimizing the allocation of functional resources in the station realm. Full article
(This article belongs to the Special Issue Urban Geospatial Analytics Based on Crowdsourced Data)
►▼ Show Figures

Figure 1

Figure 1
<p>One week passenger flow statistics chart of Xi’an Metro. Source: Xi’an Metro Official Weibo Account, <a href="https://weibo.com/xianditie" target="_blank">https://weibo.com/xianditie</a> (accessed on 18 to 24 October 2021).</p> Full article ">Figure 2
<p>The location of four research stations on Xi’an Metro Line 2.</p> Full article ">Figure 3
<p>The framework of the research process.</p> Full article ">Figure 4
<p>Pedestrian walking time (<b>a</b>) and walking speed (<b>b</b>) in BDJ station realm.</p> Full article ">Figure 5
<p>The scope adjustment process of the BDJ station realm: (<b>a</b>) before adjustment; (<b>b</b>) after adjustment.</p> Full article ">Figure 6
<p>Scope of the four research station realms: (<b>a</b>) XZZX Station, (<b>b</b>) LSY Station, (<b>c</b>) BDJ Station, (<b>d</b>) WYJ Station.</p> Full article ">Figure 7
<p>Data collection area and partial data. (<b>a</b>) Data collection area of Metro Line 2; (<b>b</b>) Point data distribution of population activities.</p> Full article ">Figure 8
<p>Statistics on the number of POIs for each research station realm.</p> Full article ">Figure 9
<p>Visualization of kernel density analysis results under different scale search radius: (<b>a</b>) 50 m radius, (<b>b</b>) 100 m radius, (<b>c</b>) 150 m radius, (<b>d</b>) 200 m radius, (<b>e</b>) more than 200 m radius.</p> Full article ">Figure 10
<p>Temporal and spatial distribution of human activities in each study station realm: (<b>a</b>) XZZX Station, (<b>b</b>) LSY Station, (<b>c</b>) BDJ Station, (<b>d</b>) WYJ Station.</p> Full article ">Figure 10 Cont.
<p>Temporal and spatial distribution of human activities in each study station realm: (<b>a</b>) XZZX Station, (<b>b</b>) LSY Station, (<b>c</b>) BDJ Station, (<b>d</b>) WYJ Station.</p> Full article ">Figure 11
<p>Peak nuclear density of population activity within 500 m of each research station realm: (<b>a</b>) XZZX station realm; (<b>b</b>) LSY station realm; (<b>c</b>) BDJ station realm; (<b>d</b>) WYJ station realm.</p> Full article ">Figure 12
<p>The peak kernel density of population activities in the study station realm during working days and off days: (<b>a</b>) XZZX Station, (<b>b</b>) LSY Station, (<b>c</b>) BDJ Station, (<b>d</b>) WYJ Station.</p> Full article ">Figure 13
<p>Kernel density analysis of each functional POI in study station realm. The analysis results of the (<b>a</b>) XXZX Station; (<b>b</b>) LSY Station; (<b>c</b>) BDJ Station; (<b>d</b>) WYJ Station.</p> Full article ">Figure 13 Cont.
<p>Kernel density analysis of each functional POI in study station realm. The analysis results of the (<b>a</b>) XXZX Station; (<b>b</b>) LSY Station; (<b>c</b>) BDJ Station; (<b>d</b>) WYJ Station.</p> Full article ">
26 pages, 1327 KiB  
Article
Soft Integration of Geo-Tagged Data Sets in J-CO-QL+
by Paolo Fosci and Giuseppe Psaila
ISPRS Int. J. Geo-Inf. 2022, 11(9), 484; https://doi.org/10.3390/ijgi11090484 - 13 Sep 2022
Cited by 7 | Viewed by 1596
Abstract
The possibility offered by the current technology to collect and store data sets regarding public places located on the Earth globe is posing new challenges, as far as the integration of these data sets is concerned. Analysts usually need to perform such an [...] Read more.
The possibility offered by the current technology to collect and store data sets regarding public places located on the Earth globe is posing new challenges, as far as the integration of these data sets is concerned. Analysts usually need to perform such an integration from scratch, without performing complex and long preprocessing or data-cleaning tasks, as well as without performing training activities that require tedious and long labeling of data; furthermore, analysts now have to deal with the popular JSON format and with data sets stored within JSON document stores. This paper demonstrates that a methodology based on soft integration (i.e., data integration performed through soft computing and fuzzy sets) can now be effectively applied from scratch, through the J-CO Framework, which is a stand-alone tool devised to process JSON data sets stored within JSON document stores, possibly by performing soft querying on data sets. Specifically, the paper provides the following contributions: (1) It presents a soft-computing technique for integrating data sets describing public places, without any preliminary pre-processing, cleaning and training, which can be applied from scratch; (2) it presents current capabilities for soft integration of JSON data sets, provided by the J-CO Framework; (3) it demonstrates the effectiveness of the soft integration technique; (4) it shows how a stand-alone tool able to support soft computing (as the J-CO Framework) can be effective and efficient in performing data-integration tasks from scratch. Full article
(This article belongs to the Special Issue Artificial Intelligence for Multisource Geospatial Information)
►▼ Show Figures

Figure 1

Figure 1
<p>The <span class="html-italic">J-CO</span> Framework.</p> Full article ">Figure 2
<p>Sample <math display="inline"><semantics> <mrow> <mi>C</mi> <mi>l</mi> <mi>o</mi> <mi>s</mi> <mi>e</mi> </mrow> </semantics></math> membership function taken from [<a href="#B6-ijgi-11-00484" class="html-bibr">6</a>].</p> Full article ">Figure 3
<p>Examples of documents representing place descriptors. (<b>a</b>) Example of document in the <tt>FacebookDescriptors</tt> collection. (<b>b</b>) Example of document in the <tt>GoogleDescriptors</tt> collection.</p> Full article ">Figure 4
<p>Membership functions for the fuzzy operators in Listing 1. (<b>a</b>) <tt>Close</tt>; (<b>b</b>) <tt>Similar</tt>; (<b>c</b>) <tt>WeightedAggregationBeta</tt>.</p> Full article ">Figure 5
<p>Example of document generated by the <tt>JOIN OF COLLECTIONS</tt> instruction on line 5 of the <span class="html-italic">J-CO-QL<math display="inline"><semantics> <msup> <mrow/> <mo>+</mo> </msup> </semantics></math></span> script, before the <tt>CASE clause</tt>.</p> Full article ">Figure 6
<p>Examples of documents generated by the <tt>JOIN OF COLLECTIONS</tt> instruction on line 5. (<b>a</b>) Example for Case A. (<b>b</b>) Example for Case B. (<b>c</b>) Example for Case C.</p> Full article ">Figure 7
<p>Examples of documents transformed by the <tt>FILTER</tt> instruction on line 6 before the <tt>BUILD</tt> section. (<b>a</b>) Example for Case A. (<b>b</b>) Example for Case B. (<b>c</b>) Example for Case C.</p> Full article ">Figure 8
<p>Examples of documents generated by the <tt>FILTER</tt> instruction on line 6. (<b>a</b>) Example for Case A. (<b>b</b>) Example for Case B. (<b>c</b>) Example for Case C.</p> Full article ">Figure 9
<p>Examples of documents during selection of <tt>BestPairs</tt> in Listing 4. (<b>a</b>) Example of document after <tt>GROUP</tt> instruction on line 9. (<b>b</b>) Example of document during <tt>EXPAND</tt> instruction on line 10 before the <tt>BUILD</tt> clause. (<b>c</b>) Example of document after <tt>EXPAND</tt> instruction on line 10.</p> Full article ">Figure 10
<p>Sensitivity analysis of precision, recall and accuracy.</p> Full article ">
30 pages, 10402 KiB  
Article
Research on the Dynamic Evolution of the Landscape Pattern in the Urban Fringe Area of Wuhan from 2000 to 2020
by Yan Long, Shiqi Luo, Xi Liu, Tianyue Luo and Xuejun Liu
ISPRS Int. J. Geo-Inf. 2022, 11(9), 483; https://doi.org/10.3390/ijgi11090483 - 13 Sep 2022
Cited by 11 | Viewed by 2366
Abstract
The urban fringe area is a discontinuous spatial phenomenon that refers to the urban-rural interlacing zone which is undergoing urbanization on the fringe of the core built-up area of a large city after the emergence of industrialization. Dynamic, ambiguous, and complex interlacing of [...] Read more.
The urban fringe area is a discontinuous spatial phenomenon that refers to the urban-rural interlacing zone which is undergoing urbanization on the fringe of the core built-up area of a large city after the emergence of industrialization. Dynamic, ambiguous, and complex interlacing of various types of lands make urban planners and managers fuzzy about the spatial scope of the urban fringe and it is difficult to control its evolution patterns scientifically. Based on remote sensing data from 2000 to 2020, the range of Wuhan’s urban fringe was extracted from the surface impermeability ratio mutation points, landscape flocculation, and population density. On this basis, the dynamic evolution characteristics of land-use and landscape patterns in the urban fringe area of Wuhan City were analyzed by using dynamic change and landscape pattern index analysis. The results show that: Wuhan City shows a clear “urban core area-urban fringe area-rural hinterland” circle structure, and the urban fringe area continuously extends to the rural hinterland. Moreover, most of the rural hinterland, in the process of moving to the urban core area, has gone through the process of the urban fringe. By comparison with other cities, it is found that the expansion of large cities is generally influenced by policies, topography, and traffic arteries, and gradually shifts from expansion to infill, with the urban core of Wuhan continuously extending and the urban fringe rapidly expanding from 2000 to 2010, and gradually entering a stable development state from 2010 to 2020. The future urban construction of Wuhan should pay attention to the influences of these characteristics on the implementation of urban territorial spatial planning. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>Research area.</p> Full article ">Figure 2
<p>Workflow diagram.</p> Full article ">Figure 3
<p>Spatial sampling point design process.</p> Full article ">Figure 4
<p>Partial data sequence test results.</p> Full article ">Figure 5
<p>Impervious surface ratio of Wuhan City 2000–2020 extracted urban fringe area range.</p> Full article ">Figure 6
<p>The spatial division of landscape flocculation in the urban fringe of Wuhan, 2000–2020.</p> Full article ">Figure 7
<p>Spatial distribution of urban fringe areas in Wuhan from 2000 to 2020.</p> Full article ">Figure 8
<p>Area distribution of Wuhan urban fringe areas along different directions of expansion, 2000–2020.</p> Full article ">Figure 9
<p>Area histogram of the geographical spatial structure within each administrative region of Wuhan, 2000–2020.</p> Full article ">
20 pages, 16310 KiB  
Article
SocialMedia2Traffic: Derivation of Traffic Information from Social Media Data
by Mohammed Zia, Johannes Fürle, Christina Ludwig, Sven Lautenbach, Stefan Gumbrich and Alexander Zipf
ISPRS Int. J. Geo-Inf. 2022, 11(9), 482; https://doi.org/10.3390/ijgi11090482 - 13 Sep 2022
Cited by 6 | Viewed by 2691
Abstract
Traffic prediction is a topic of increasing importance for research and applications in the domain of routing and navigation. Unfortunately, open data are rarely available for this purpose. To overcome this, the authors explored the possibility of using geo-tagged social media data (Twitter), [...] Read more.
Traffic prediction is a topic of increasing importance for research and applications in the domain of routing and navigation. Unfortunately, open data are rarely available for this purpose. To overcome this, the authors explored the possibility of using geo-tagged social media data (Twitter), land-use and land-cover point of interest data (from OpenStreetMap) and an adapted betweenness centrality measure as feature spaces to predict the traffic congestion of eleven world cities. The presented framework and workflow are termed as SocialMedia2Traffic. Traffic congestion was predicted at four tile spatial resolutions and compared with Uber Movement data. The overall precision of the forecast for highly traffic-congested regions was approximately 81%. Different data processing steps including ways to aggregate data points, different proxies and machine learning approaches were compared. The lack of a universal definition on a global scale to classify road segments by speed bins into different traffic congestion classes has been identified to be a major limitation of the transferability of the framework. Overall, SocialMedia2Traffic further improves the usability of the tested feature space for traffic prediction. A further benefit is the agnostic nature of the social media platform’s approach. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>Conceptual layers contributing to the whole SM2T infrastructure. They are stacked in increasing order of priority. The <span class="html-italic">Live Traffic layer</span> is generated using two identified Twitter proxies, along with the land-use land-cover POI and betweenness centrality (cf. main text). Note that on the left map only Twitter proxies are shown.</p> Full article ">Figure 2
<p>A sample street network in the city of Heidelberg, Germany, showing (<b>a</b>) how geo-tagged tweets were selected based on the buffer around the highways for the “User count on a road segment” proxy, and (<b>b</b>) how a geo-tagged tweet cluster in a public space was used for the “User count within a vicinity” proxy.</p> Full article ">Figure 3
<p>The figure shows four ways to aggregate land-use and land-cover POI data for a given tile. For (<b>a</b>), a simple counting of all POIs in an area of interest is performed. In (<b>b</b>) a pre-defined road buffer, per different highway type, is used to select only nearby POIs before counting. In (<b>c</b>) pre-defined weights, according to the importance of the infrastructure, are used while counting, and in (<b>d</b>) a pre-defined road buffer, per different highway type, in addition to weighting is used to select and prioritise only nearby POIs before counting.</p> Full article ">Figure 4
<p>Accuracy of prediction of three traffic classes identified by the betweenness centrality measure, with the Uber Movement classes used as ground truth (<b>left</b>). Colours indicate the percentage of tiles correctly classified by the model using the adapted betweenness centrality as predictor. Extent of over/underestimation in wrongly classified classes using this proxy (<b>right</b>). Colours indicate the share of tiles incorrectly classified by the model using adapted betweenness centrality as a predictor.</p> Full article ">Figure 5
<p>The plot compares the performance of two Twitter aggregation methods per tile: (i) <span class="html-italic">User count on a road segment</span> and (ii) <span class="html-italic">User count on a road segment</span> + <span class="html-italic">User count within a vicinity</span>. The colour indicates the difference in the percentage of correctly classified tiles between the predictions based on the two approaches. A positive value (blue) implies that the aggregation method using the combination of the two proxies is a better predictor.</p> Full article ">Figure 6
<p>Degree of class imbalance in HTC vs. no-HTC labels per each dataset based on the Uber data (static class definitions). All instances with at least one empty feature space have been discarded. The x-axis shows the relationship between the number of HTC to no-HTC tiles per city for the different tile sizes. A value of 50% indicates perfectly balanced data. “Merged” represents the combination of all eleven cities.</p> Full article ">Figure 7
<p>Visualising traffic congestion classes derived from all feature spaces using a quantile approach and validation (Uber) dataset for three cities (showing both edge case scenarios). The number of cells with predictions differs, as the predictors were not available for all tiles. The precision of the model predictions is presented in <a href="#ijgi-11-00482-f010" class="html-fig">Figure 10</a>.</p> Full article ">Figure 8
<p>Comparison of the precision of five classification algorithms for different tile sizes. The three plots represent different datasets based on how empty cells were handled using the dummy value. The variability in the boxplots is due to the different precisions for the individual cities.</p> Full article ">Figure 9
<p>Selecting the best-performing number of neighbours of the k-nearest neighbours classifier for different tile sizes. The red curve represents the combined cities. The different black colours represent the individual cities. The vertical blue line characterises the selected value of 30 nearest neighbours.</p> Full article ">Figure 10
<p>The performance of the k-nearest neighbours classifier (k = 30) for different tile sizes, different cities and a combined dataset. For each city, the model was trained using a 5-fold cross-validation approach using all cities but the selected city and validated against the latter. For the combined dataset, the model was trained using a 5-fold cross-validation approach and with precision calculated for the whole dataset.</p> Full article ">Figure A1
<p>SM2T architecture and interface.</p> Full article ">
20 pages, 5100 KiB  
Article
Exploring the Applicability of Self-Organizing Maps for Ecosystem Service Zoning of the Guangdong-Hong Kong-Macao Greater Bay Area
by Yingwei Yan, Yingbin Deng, Ji Yang, Yong Li, Xinyue Ye, Jianhui Xu and Yuyao Ye
ISPRS Int. J. Geo-Inf. 2022, 11(9), 481; https://doi.org/10.3390/ijgi11090481 - 13 Sep 2022
Cited by 4 | Viewed by 2228
Abstract
Sustainability is one of the major challenges in the 21st century for humanity. Spatial zoning of ecosystem services is proposed in this study as a solution to meet the demands for the sustainable use of ecosystem services. This study presented a workflow and [...] Read more.
Sustainability is one of the major challenges in the 21st century for humanity. Spatial zoning of ecosystem services is proposed in this study as a solution to meet the demands for the sustainable use of ecosystem services. This study presented a workflow and performed an exploratory analysis using self-organizing maps (SOM) for visualizing the spatial patterns of the ecosystem service value (ESV) of the Guangdong-Hong Kong-Macao Greater Bay Area (GBA). The zoning was performed based on 11 types of ecosystem services, resulting in 11 ecosystem service zones. Each of the zones derived has its unique characteristics in terms of the dominating ecosystem service types, ESV, land use/land cover patterns, and associated human activity levels. It is recommended that reasonable and effective utilization of the ecosystem services in the GBA should be based on its zonal characteristics rather than haphazard exploitations, which can contribute to the sustainable economy and environment of the region. The applicability of SOM for the GBA ecosystem service zoning has been demonstrated in this study. However, it should be stressed that the method and workflow presented in this study should mainly be used for supporting decision-making rather than used for deriving gold-standard zoning maps. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>(<b>a</b>) The GBA location. (<b>b</b>) The GBA mega-city region shown on the Esri World Hillshade base map. Note: The administrative boundary of the GBA was collected from the Resource and Environment Data Center of the Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences [<a href="#B45-ijgi-11-00481" class="html-bibr">45</a>].</p> Full article ">Figure 2
<p>The overall ESV of the GBA (spatial resolution: 1 km; temporal stamp: 2018). Note: quantile classification is adopted to classify the cell values, which leads to an effective visualization of the spatial variation in the data.</p> Full article ">Figure 3
<p>The GBA land use/land cover (spatial resolution: 30 m; temporal stamp: 2017).</p> Full article ">Figure 4
<p>Flowchart briefly depicting how SOM works.</p> Full article ">Figure 5
<p>Annual average radiance value of the night-time lights of the GBA (spatial resolution: 1 km; temporal stamp: 2016). Note: quantile classification is adopted to classify the cell values, which leads to an effective visualization of the spatial variation in the data.</p> Full article ">Figure 6
<p>The workflow leading to the ecosystem service zoning result.</p> Full article ">Figure 7
<p>Fan diagram exhibiting the patterns in the distribution of the input indicators in the individual neurons of the 5 × 5 SOM. The chain link between a pair of neighboring neurons indicates that the Pearson correlation between the two neurons is greater than 0.75 (highly correlated) and thus can be further clustered into the same zone.</p> Full article ">Figure 8
<p>(<b>a</b>) Number of data samples falling in each neuron of the 5 × 5 SOM; (<b>b</b>) ecosystem service zoning result of the 5 × 5 SOM (11 zones labeled using integer numbers).</p> Full article ">Figure 9
<p>(<b>a</b>) Fan diagram exhibiting the patterns in the distribution of the input indicators in the individual neurons of the 2 × 2 SOM; (<b>b</b>) number of data samples falling in the individual neurons of the 2 × 2 SOM; (<b>c</b>) fan diagram exhibiting the patterns in the distribution of the input indicators in the individual neurons of the 3 × 3 SOM; (<b>d</b>) number of data samples falling in the individual neurons of the 3 × 3 SOM; (<b>e</b>) fan diagram exhibiting the patterns in the distribution of the input indicators in the individual neurons of the 4 × 4 SOM; (<b>f</b>) number of data samples falling in the individual neurons of the 4 × 4 SOM; (<b>g</b>) fan diagram exhibiting the patterns in the distribution of the input indicators in the individual neurons of the 6 × 6 SOM; (<b>h</b>) number of data samples falling in the individual neurons of the 6 × 6 SOM (gray neurons indicate zero data sample).</p> Full article ">Figure 10
<p>The ecosystem service zoning result of the SOM projected onto the study area (spatial resolution: 1 km; temporal stamp: 2018).</p> Full article ">Figure 11
<p>LULC patterns across the ecosystem service zones.</p> Full article ">Figure 12
<p>Distribution of the 2016 annual average radiance value night-time lights across the ecosystem service zones.</p> Full article ">
20 pages, 7868 KiB  
Article
Extracting Skeleton Lines from Building Footprints by Integration of Vector and Raster Data
by Guoqing Chen and Haizhong Qian
ISPRS Int. J. Geo-Inf. 2022, 11(9), 480; https://doi.org/10.3390/ijgi11090480 - 10 Sep 2022
Cited by 7 | Viewed by 2601
Abstract
The extraction of skeleton lines of buildings is a key step in building spatial analysis, which is widely performed for building matching and updating. Several methods for vector data skeleton line extraction have been established, including the improved constrained Delaunay triangulation (CDT) and [...] Read more.
The extraction of skeleton lines of buildings is a key step in building spatial analysis, which is widely performed for building matching and updating. Several methods for vector data skeleton line extraction have been established, including the improved constrained Delaunay triangulation (CDT) and raster data skeleton line extraction methods, which are based on image processing technologies. However, none of the existing studies have attempted to combine these methods to extract the skeleton lines of buildings. This study aimed to develop a building skeleton line extraction method based on vector–raster data integration. The research object was buildings extracted from remote sensing images. First, vector–raster data mapping relationships were identified. Second, the buildings were triangulated using CDT. The extraction results of the Rosenfeld thin algorithm for raster data were then used to remove redundant triangles. Finally, the Shi–Tomasi corner detection algorithm was used to detect corners. The building skeleton lines were extracted by adjusting the connection method of the type three triangles in CDT. The experimental results demonstrate that the proposed method can effectively extract the skeleton lines of complex vector buildings. Moreover, the skeleton line extraction results included a few burrs and were robust against noise. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>CDT. (<b>a</b>) Before encryption; (<b>b</b>) after encryption.</p> Full article ">Figure 2
<p>Triangle classification. (<b>a</b>) Type one; (<b>b</b>) type two; (<b>c</b>) type three.</p> Full article ">Figure 3
<p>Skeleton line extraction based on CDT.</p> Full article ">Figure 4
<p>Skeleton line extraction results. (<b>a</b>) CDT; (<b>b</b>) Rosenfeld; (<b>c</b>) Overlap of CDT and Rosenfeld results.</p> Full article ">Figure 5
<p>Process flow of building skeleton line extraction.</p> Full article ">Figure 6
<p>Extraction of the triangle in which the skeleton line was located.</p> Full article ">Figure 7
<p>Corner detection results. (<b>a</b>) L-shaped building; (<b>b</b>) C-shaped building; (<b>c</b>) I-shaped building; (<b>d</b>) T-shaped building.</p> Full article ">Figure 8
<p>Detection of endpoints and crosspoints. (<b>a</b>) Rosenfeld skeleton line; (<b>b</b>) Crosspoint; (<b>c</b>) Endpoint.</p> Full article ">Figure 9
<p>Three type three triangles without endpoints, corners, or intersections. (<b>a</b>) Delaunay triangles classification results; (<b>b</b>) Overlay of Rosenfeld and CDT results; (<b>c</b>) Overlay of Rosenfeld, CDT and VRDI results.</p> Full article ">Figure 10
<p>Traditional CDT method for type three triangle connections with a crosspoint. (<b>a</b>) Building with crosspoint; (<b>b</b>) Result of traditional CDT method.</p> Full article ">Figure 11
<p>VRDI extraction results. (<b>a</b>) The MBR of building; (<b>b</b>) Result of VRDI method for crosspoint.</p> Full article ">Figure 12
<p>Type three with corner points. (<b>a</b>) C-shaped building with corner point; (<b>b</b>) Result of VRDI method for corner point.</p> Full article ">Figure 13
<p>Triangle with endpoints. (<b>a</b>) C-shaped building with endpoint; (<b>b</b>) Result of VRDI method for endpoint.</p> Full article ">Figure 14
<p>Skeleton line segmentation. (<b>a</b>) L-shaped building with one corner point; (<b>b</b>) C-shaped building with two corner points; (<b>c</b>) T-shaped building with one crosspoint.</p> Full article ">Figure 15
<p>Skeleton line segment DP simplification results with different threshod; (<b>a</b>) L-shaped building with a threshold of 1 m; (<b>b</b>) L-shaped building with a threshold of 3 m; (<b>c</b>) C-shaped building with a threshold of 1 m; (<b>d</b>) C-shaped building with a threshold of 3 m.</p> Full article ">Figure 16
<p>(<b>a</b>) High-resolution remote sensing images. The highlighted red box area in (<b>a</b>) corresponds to (<b>b</b>,<b>c</b>); (<b>b</b>) Initially extracted vector buildings; (<b>c</b>) Skeleton line extraction results based on the proposed VRDI approach as shown by the red solid line.</p> Full article ">Figure 17
<p>Skeleton line extraction results for T-shaped and F-shaped buildings. (<b>a</b>) High-resolution remote sensing images. The highlighted red box area in (<b>a</b>) corresponds to (<b>b</b>,<b>c</b>); (<b>b</b>) initially extracted vector buildings; (<b>c</b>) skeleton line extraction results based on the proposed VRDI approach as shown by the red solid line; (<b>d</b>) High-resolution remote sensing images. The highlighted red box area in (<b>d</b>) corresponds to (<b>e</b>,<b>f</b>); (<b>e</b>) initially extracted vector buildings; (<b>f</b>) skeleton line extraction results based on the proposed VRDI approach as shown by the red solid line.</p> Full article ">Figure 18
<p>Building skeleton lines without crosspoints.</p> Full article ">Figure 19
<p>Building skeleton lines with crosspoints.</p> Full article ">Figure 20
<p>MBR for buildings and skeleton lines.</p> Full article ">
23 pages, 19369 KiB  
Article
Exploring Landscape Composition Using 2D and 3D Open Urban Vectorial Data
by Frédéric Pedrinis, John Samuel, Manuel Appert, Florence Jacquinod and Gilles Gesquière
ISPRS Int. J. Geo-Inf. 2022, 11(9), 479; https://doi.org/10.3390/ijgi11090479 - 10 Sep 2022
Cited by 4 | Viewed by 2690
Abstract
Methods and tools for assessing the visual impact of objects such as high-rises are rarely used in planning, despite the increase in opportunities to develop automated visual assessments, now that 3D urban data are acquired and used by municipalities as well as made [...] Read more.
Methods and tools for assessing the visual impact of objects such as high-rises are rarely used in planning, despite the increase in opportunities to develop automated visual assessments, now that 3D urban data are acquired and used by municipalities as well as made available through open data portals. This paper presents a new method for assessing city visibility using a 3D model on a metropolitan scale. This method measures the view composition in terms of city objects visible from a given viewpoint and produces a georeferenced and semantically rich database of those visible objects in order to propose a thematic vision of the city and its urban landscape. As far as computational efficiency is concerned and considering the large amount of data needed, the method relies on a dedicated system of automatic data organization for analyzing visibility over vast areas (hundreds of square kilometers), offering various possibilities for uses on different scales. In terms of operational uses, as shown in our paper, the various results produced by the method (quantitative data, georeferenced databases and 3D schematic images) allow for a wide spectrum of applications in urban planning. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>View composition regarding city features.</p> Full article ">Figure 2
<p>Viewpoint from south of Lyon.</p> Full article ">Figure 3
<p>Three-dimensional visualization of buildings and associated documents (See more examples in [<a href="#B17-ijgi-11-00479" class="html-bibr">17</a>]).</p> Full article ">Figure 4
<p>Composition of the 3D view in four steps: Field of View Description (Step A), Intersecting Objects in the 3D Scene (Step B), Storing intersected objects (Step C) and Storing results in the database (Step D).</p> Full article ">Figure 5
<p>Discretization of the 3D space according to rays generated from the viewpoint of interest.</p> Full article ">Figure 6
<p>A 1 × 1 km tile of Lyon composed of four types of objects: buildings, terrain, roads and vegetation.</p> Full article ">Figure 7
<p>Three rays are generated from the viewpoint of different kinds of objects.</p> Full article ">Figure 8
<p>Organization of a model city using a regular grid.</p> Full article ">Figure 9
<p>(<b>Left</b>): Top view of 3D skyline for a given point or view (purple). (<b>Right</b>): Composition of this skyline.</p> Full article ">Figure 10
<p>View composition analysis. (<b>Left</b>): decomposition of the skyline according to the intersected object. (<b>Right</b>): the view composition.</p> Full article ">Figure 11
<p>Three-dimensional visualization of a tile (1 × 1 km) of the CityGML model of Lyon.</p> Full article ">Figure 12
<p>One-meter-resolution DEM used for a visibility analysis for the Lyon Metropolis: view from the whole DEM covering the city of Lyon (48 sq km) (<b>left</b>) and zoom on a specific part of the city (<b>right</b>).</p> Full article ">Figure 13
<p>Visual comparison between data describing the same location (Bellecour square): raster data (1-m DEM—<b>left</b>) and 3D vector data (<b>right</b>).</p> Full article ">Figure 14
<p>Three-dimensional visualizations to compare the results of 2.5D raster (<b>left</b>) and 3D vector (<b>right</b>) visibility analyses from the vantage point of the Fourvière Basilica. On the raster analysis, the visible areas are in green, while on the vector one, we only see the visible 3D points colored according to their type. The 3D model of Saint Jean’s Cathedral has been added to the bottom visualizations, which zoom in on the panoramic view of the top images.</p> Full article ">Figure 15
<p>Three-dimensional visualization using our tools, with each color corresponding to the CityGML category of the resulting 3D points (green: vegetation, grey: buildings, yellow: terrain, blue: water).</p> Full article ">Figure 16
<p>Raster analysis results regarding the visibility of the Fourvière Basilica (Bellecour Square). Green pixels indicate that the Basilica is seen, and red pixels that it is not seen. The results are displayed on an aerial image of the square; hence transparency is used.</p> Full article ">Figure 17
<p>Same as <a href="#ijgi-11-00479-f016" class="html-fig">Figure 16</a>, with the addition of the visibility analysis from our tool (in green, vegetation; in yellow, terrain; in red, roofs; in white, building’s walls).</p> Full article ">Figure 18
<p>Raster analysis results regarding the visibility of the Fourvière Basilica (Bellecour Square). Green pixels indicate areas where the Basilica can be seen, and red pixels indicate places from which it is not visible. The results are displayed on an aerial image with a little transparency. Building footprints are represented in black.</p> Full article ">Figure 19
<p>Buildings that have a facade from which the Fourvière Basilica can be seen are shown in white. Buildings from which the Basilica is only visible from the rooftop are excluded.</p> Full article ">Figure 20
<p>Visualization of vantage points (each point is a vantage point). On the (<b>left</b>), the number of landmarks seen from the vantage points (from red, five landmarks seen, to yellow, zero landmarks seen). On the (<b>right</b>), effort needed to access the vantage point on foot, in calories (from light blue (less than 20 calories) to dark blue (more than 70 calories).</p> Full article ">Figure 21
<p>Visualization of an imaginary high-rise project (on the right) in the existing business district of La Part-Dieu from the belvedere of Fourvière.</p> Full article ">
23 pages, 13838 KiB  
Article
Early Detection of Suspicious Behaviors for Safe Residence from Movement Trajectory Data
by Junyi Cheng, Xianfeng Zhang, Xiao Chen, Miao Ren, Jie Huang and Peng Luo
ISPRS Int. J. Geo-Inf. 2022, 11(9), 478; https://doi.org/10.3390/ijgi11090478 - 3 Sep 2022
Cited by 4 | Viewed by 3302
Abstract
Early detection of people’s suspicious behaviors can aid in the prevention of crimes and make the community safer. Existing methods that are focused on identifying abnormal behaviors from video surveillance that are based on computer vision, which are more suitable for detecting ongoing [...] Read more.
Early detection of people’s suspicious behaviors can aid in the prevention of crimes and make the community safer. Existing methods that are focused on identifying abnormal behaviors from video surveillance that are based on computer vision, which are more suitable for detecting ongoing behaviors. While criminals intend to avoid abnormal behaviors under surveillance, their suspicious behaviors prior to crimes will be unconsciously reflected in the trajectories. Herein, we characterize several suspicious behaviors from unusual movement patterns, unusual behaviors, and unusual gatherings of people, and analyze their features that are hidden in the trajectory data. Meanwhile, the algorithms for suspicious behavior detection are proposed based on the main features of the corresponding behavior, which employ spatiotemporal clustering, semantic annotation, outlier detection, and other methods. A practical trajectory dataset (i.e., TucityLife) containing more than 1000 suspicious behaviors was collected, and experiments were conducted to verify the effectiveness of the proposed method. The results indicate that the proposed method for suspicious behavior detection has a recall of 93.5% and a precision of 87.6%, demonstrating its excellent performance in identifying the possible offenders and potential target places. The proposed methods are valuable for preventing city crime and supporting the appropriate allocation of police resources. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>Flowchart of the proposed method for suspicious behavior detection from the trajectory data.</p> Full article ">Figure 2
<p>The schematic (<b>a</b>) and actual trajectory (<b>b</b>) of aimlessly wandering. Each curve in (<b>a</b>) corresponds to a large directional change.</p> Full article ">Figure 3
<p>Several cases of large-angle changes in the heading angle. (<b>a</b>) correct large-angle change, (<b>b</b>) false large-angle change that is caused by noise, and (<b>c</b>) false large-angle change that is caused by a short stop.</p> Full article ">Figure 4
<p>An example of a trajectory of frequent short stops. Each red circle indicates a short stop.</p> Full article ">Figure 5
<p>An example of a trajectory of loitering.</p> Full article ">Figure 6
<p>Schematic of an unusual route.</p> Full article ">Figure 7
<p>Schematic of crowd gathering. The gray circles show that people are gathering at the moment.</p> Full article ">Figure 8
<p>Comparison of the accuracy of different methods in suspicious behavior detection. The compared methods include the method that was proposed by Wu et al. [<a href="#B17-ijgi-11-00478" class="html-bibr">17</a>] for loitering detection, θ_WD [<a href="#B47-ijgi-11-00478" class="html-bibr">47</a>] for aimlessly wandering detection, and LOF [<a href="#B48-ijgi-11-00478" class="html-bibr">48</a>] for unusual route detection.</p> Full article ">Figure 9
<p>Accuracy of suspicious behavior detection with varying parameters from the TucityLife dataset. (<b>a</b>) aimlessly wandering; (<b>b</b>) loitering.</p> Full article ">Figure 10
<p>Examples of the trajectories of aimlessly wandering in the TucityLife dataset. Aimlessly wandering on the same road (<b>a</b>–<b>d</b>), on different roads (<b>e</b>–<b>h</b>), and in an area (<b>i</b>,<b>j</b>).</p> Full article ">Figure 11
<p>Examples of the trajectories of loitering that were extracted from the TucityLife dataset. (<b>a</b>) Loitering and stopping; (<b>b</b>) loitering; (<b>c</b>–<b>f</b>) loitering and leaving briefly; (<b>g</b>–<b>i</b>) multiple areas are examined simultaneously.</p> Full article ">Figure 12
<p>Examples of frequent short stops that were detected from the TucityLife dataset. (<b>a</b>–<b>e</b>) Several trajectories with multiple short stops (red circle) occurring within a short time.</p> Full article ">Figure 13
<p>Examples of the unusual routes that were detected from the TucityLife dataset. (<b>a</b>–<b>c</b>) Several trajectories of unusual (brown) and normal (black) routes which are mapped on 50 m × 50 m grids.</p> Full article ">
22 pages, 10085 KiB  
Article
Spatial Interaction Analysis of Shared Bicycles Mobility Regularity and Determinants: A Case Study of Six Main Districts, Beijing
by Lujin Hu, Zheng Wen, Jian Wang and Jing Hu
ISPRS Int. J. Geo-Inf. 2022, 11(9), 477; https://doi.org/10.3390/ijgi11090477 - 2 Sep 2022
Cited by 6 | Viewed by 2364
Abstract
Understanding the regularity and determinants of mobility is indispensable for the reasonable deployment of shared bicycles and urban planning. A spatial interaction network covering streets in Beijing’s six main districts, using bike sharing data, is constructed and analyzed. as Additionally, the exponential random [...] Read more.
Understanding the regularity and determinants of mobility is indispensable for the reasonable deployment of shared bicycles and urban planning. A spatial interaction network covering streets in Beijing’s six main districts, using bike sharing data, is constructed and analyzed. as Additionally, the exponential random graph model (ERGM) is used to interpret the influencing factors of the network structure and the mobility regularity. The characteristics of the spatial interaction network structure and temporal characteristics between weekdays and weekends show the following: the network structure on weekdays is obvious; the flow edge is always between adjacent blocks; the traffic flow frequently changes and clusters; the network structure on weekends is more complex, showing scattering and seldom changing; and there is a stronger interaction between blocks. Additionally, the predicted result of the ERGM shows that the influencing factors selected in this paper are positively correlated with the spatial interaction network. Among them, the three most important determinants are building density, housing prices and the number of residential areas. Additionally, the determinant of financial services shows greater effects on weekdays than weekends. Full article
(This article belongs to the Topic Advances in Sustainable Communities, Neighborhoods and 15-Minute Cities-Theory, Methods and Techniques)
►▼ Show Figures

Figure 1

Figure 1
<p>(<b>a</b>) Research area in Beijing; (<b>b</b>) Research Area of the darkened place in (<b>a</b>).</p> Full article ">Figure 2
<p>Heat map of shared bicycle spatial distribution at 17:00 and 20:00 from 19 May 2021 to 22 May 2021: (<b>a</b>) 17:00 19 May 2021; (<b>b</b>) 20:00 19 May 2021; (<b>c</b>) 17:00 22 May 2021; (<b>d</b>) 20:00 22 May 2021.</p> Full article ">Figure 3
<p>Spatial distribution of data in the research area: (<b>a</b>) POI of transportation facilities; (<b>b</b>) POI of financial services; (<b>c</b>) POI of education institution; (<b>d</b>) POI of residential area; (<b>e</b>) POI of shopping and dining places; (<b>f</b>) Road network in Urban Six Districts of Beijing.</p> Full article ">Figure 4
<p>Travel network construction. The data used in this paper include shared bike trajectories and study areas. The OD flows are generated from the shared bicycle trajectory, and the streets in the study area are extracted as network nodes. The location of the start and end points of the OD stream are determined by the location of the block. A travel network is generated with streets as nodes.</p> Full article ">Figure 5
<p>Bike travel network structure: (<b>a</b>) Weekday (17 May); (<b>b</b>) Weekday (18 May); (<b>c</b>) Weekdays (19–21 May); (<b>d</b>) Weekends (22–23 May).</p> Full article ">Figure 5 Cont.
<p>Bike travel network structure: (<b>a</b>) Weekday (17 May); (<b>b</b>) Weekday (18 May); (<b>c</b>) Weekdays (19–21 May); (<b>d</b>) Weekends (22–23 May).</p> Full article ">Figure 6
<p>Analysis of travel network characteristics: (<b>a</b>) Analysis of in-degree and out-degree during weekday (19–21 May); (<b>b</b>) Analysis of in-degree and out-degree during weekend (22–23 May); (<b>c</b>) Analysis of node strength during weekday (19–21 May); (<b>d</b>) Analysis of node strength during weekend (22–23 May); (<b>e</b>) NFR (19–23 May).</p> Full article ">Figure 6 Cont.
<p>Analysis of travel network characteristics: (<b>a</b>) Analysis of in-degree and out-degree during weekday (19–21 May); (<b>b</b>) Analysis of in-degree and out-degree during weekend (22–23 May); (<b>c</b>) Analysis of node strength during weekday (19–21 May); (<b>d</b>) Analysis of node strength during weekend (22–23 May); (<b>e</b>) NFR (19–23 May).</p> Full article ">Figure 7
<p>ERGM estimation results: (<b>a</b>) Coefficient values of all the influencing factors on weekdays; (<b>b</b>) Comparison of coefficient values of all influencing factors on weekdays and weekends; (<b>c</b>) Coefficient values of influencing factors other than Price, Bden and Res.</p> Full article ">
31 pages, 6053 KiB  
Article
Spatio-Temporal Sentiment Mining of COVID-19 Arabic Social Media
by Tarek Elsaka, Imad Afyouni, Ibrahim Hashem and Zaher Al Aghbari
ISPRS Int. J. Geo-Inf. 2022, 11(9), 476; https://doi.org/10.3390/ijgi11090476 - 2 Sep 2022
Cited by 5 | Viewed by 2768
Abstract
Since the recent outbreak of COVID-19, many scientists have started working on distinct challenges related to mining the available large datasets from social media as an effective asset to understand people’s responses to the pandemic. This study presents a comprehensive social data mining [...] Read more.
Since the recent outbreak of COVID-19, many scientists have started working on distinct challenges related to mining the available large datasets from social media as an effective asset to understand people’s responses to the pandemic. This study presents a comprehensive social data mining approach to provide in-depth insights related to the COVID-19 pandemic and applied to the Arabic language. We first developed a technique to infer geospatial information from non-geotagged Arabic tweets. Secondly, a sentiment analysis mechanism at various levels of spatial granularities and separate topic scales is introduced. We applied sentiment-based classifications at various location resolutions (regions/countries) and separate topic abstraction levels (subtopics and main topics). In addition, a correlation-based analysis of Arabic tweets and the official health providers’ data will be presented. Moreover, we implemented several mechanisms of topic-based analysis using occurrence-based and statistical correlation approaches. Finally, we conducted a set of experiments and visualized our results based on a combined geo-social dataset, official health records, and lockdown data worldwide. Our results show that the total percentage of location-enabled tweets has increased from 2% to 46% (about 2.5M tweets). A positive correlation between top topics (lockdown and vaccine) and the COVID-19 new cases has also been recorded, while negative feelings of Arab Twitter users were generally raised during this pandemic, on topics related to lockdown, closure, and law enforcement. Full article
(This article belongs to the Special Issue Artificial Intelligence for Multisource Geospatial Information)
►▼ Show Figures

Figure 1

Figure 1
<p>The Workflow of our methodology to analyze the social data.</p> Full article ">Figure 2
<p>Reprocess COVID-19 tweets dataset.</p> Full article ">Figure 3
<p>Percentage of the Geo-tagged tweets in the new COVID-19 Tweets Dataset after applying the Location Extraction algorithm.</p> Full article ">Figure 4
<p>Tweets and Hashtags in Arab and non-Arab Countries.</p> Full article ">Figure 5
<p>Monthly distribution of Top 10 Topics in the Tweets Dataset.</p> Full article ">Figure 6
<p>Top topics distributed globally.</p> Full article ">Figure 7
<p>Main topics in Arab and non-Arab countries.</p> Full article ">Figure 8
<p>SA of COVID-19 Arabic Tweets.</p> Full article ">Figure 9
<p>SA Classifiers’ performance applied on Arabic Datasets.</p> Full article ">Figure 10
<p>Correlation between SA and the Official Health Records.</p> Full article ">Figure 11
<p>Correlation between SA and the Official Health Records in Arab and Non-Arab Countries.</p> Full article ">Figure 12
<p>Correlation between lockdown and Official COVID-19 New Cases in Arab Countries.</p> Full article ">Figure 13
<p>Correlation between lockdown and Official COVID-19 New Cases in Non-Arab Countries.</p> Full article ">Figure 14
<p>Correlation between Lockdown and SA in Arab Countries.</p> Full article ">Figure 15
<p>Correlation between Lockdown and SA in Non-Arab Countries.</p> Full article ">Figure 16
<p>Correlation between main topics and SA in Arab countries.</p> Full article ">Figure 17
<p>Correlation between main topics and SA in non-Arab countries.</p> Full article ">Figure A1
<p>Monthly distribution of tweets and Hashtags.</p> Full article ">Figure A2
<p>Sample of Arabic tweets with English translation.</p> Full article ">Figure A3
<p>Monthly distribution of Geo and Non-geo tweets in Tweets Dataset.</p> Full article ">Figure A4
<p>Word cloud of the top topics in the COVID-19 Arabic Tweets Dataset.</p> Full article ">Figure A5
<p>Top 5 Topics in Arab and Non-Arab Countries.</p> Full article ">Figure A6
<p>Monthly Top Topics in Arab and Non-Arab Countries.</p> Full article ">Figure A7
<p>Top 5 Topics in Top 10 Countries.</p> Full article ">Figure A8
<p>Frequency-Occurrence of Some Main Topics in Arab Countries.</p> Full article ">Figure A9
<p>Frequency-Occurrence of Some Main Topics in Non-Arab Countries.</p> Full article ">Figure A10
<p>Top Sentiment (Positive/Negative/Neutral) in Arab Countries.</p> Full article ">Figure A11
<p>Correlation between SA and Official Health Records in some countries.</p> Full article ">Figure A12
<p>Distribution of Arabic Tweets.</p> Full article ">Figure A13
<p>Distribution of Arabic Hashtags.</p> Full article ">Figure A14
<p>Distribution of Users.</p> Full article ">Figure A15
<p>Sentiment Analysis of Arabic Tweets.</p> Full article ">Figure A16
<p>Word Cloud of Sentiment Analysis of Arabic Tweets over the world.</p> Full article ">Figure A17
<p>Distribution of COVID-19 Cases.</p> Full article ">Figure A18
<p>Covid-19 Lockdown Days.</p> Full article ">
15 pages, 3155 KiB  
Article
Identification of Road Network Intersection Types from Vehicle Telemetry Data Using a Convolutional Neural Network
by Abdelmajid Erramaline, Thierry Badard, Marie-Pier Côté, Thierry Duchesne and Olivier Mercier
ISPRS Int. J. Geo-Inf. 2022, 11(9), 475; https://doi.org/10.3390/ijgi11090475 - 31 Aug 2022
Cited by 1 | Viewed by 3317
Abstract
GPS trajectories collected from automotive telematics for insurance purposes go beyond being a collection of points on the map. They are in fact a powerful data source that we can use to extract map and road network properties. While the location of road [...] Read more.
GPS trajectories collected from automotive telematics for insurance purposes go beyond being a collection of points on the map. They are in fact a powerful data source that we can use to extract map and road network properties. While the location of road junctions is readily available, the information about the traffic control element regulating the intersection is typically unknown. However, this information would be helpful, e.g., for contextualizing a driver’s behavior. Our focus is to use a map-matched GPS OBD-dongle dataset provided by a Canadian insurance company to classify intersections into three classes according to the type of traffic control element present: traffic light, stop sign, or no sign. We design a convolutional neural network (CNN) for classifying intersections. The network takes as entries, for a defined number of trips, the speed and the acceleration profiles over each segment of one meter on a window around the intersection. Our method outperforms two other competing approaches, achieving 99% overall accuracy. Furthermore, our CNN model can infer the three classes even with as few as 25 trips. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>Window of 100 m wide for a driver’s trip in one of the two heading directions of the road segment.</p> Full article ">Figure 2
<p>Speed–acceleration density contour plot for a given heading direction at two STs (<b>left</b>), NSs (<b>middle</b>) and TLs (<b>right</b>).</p> Full article ">Figure 3
<p>Speed traces of trips over junction window at an ST (<b>left</b>), an NS (<b>middle</b>) and a TL (<b>right</b>).</p> Full article ">Figure 4
<p>Speed measurements (solid dots) and interpolated values (empty dots) for one trip at an ST (<b>left</b>) or at a TL (<b>right</b>).</p> Full article ">Figure 5
<p>An example of speed profiles of trips on the window at an NS (<b>top</b>) and an ST (<b>bottom</b>), before (<b>left</b>) and after (<b>right</b>) filtering trips and fixing GPS errors.</p> Full article ">Figure 6
<p>CNN model architecture for characterizing TCEs at an intersection.</p> Full article ">Figure 7
<p>Examples of kernel density images for the method of [<a href="#B17-ijgi-11-00475" class="html-bibr">17</a>].</p> Full article ">Figure 8
<p>Training (blue) and validation (orange) losses per epoch for <span class="html-italic">CNN_25</span> (<b>left</b>), <span class="html-italic">CNN_50</span> (<b>middle</b>) and <span class="html-italic">CNN_100</span> (<b>right</b>).</p> Full article ">Figure 9
<p>Confusion matrix for class ST illustrating <math display="inline"><semantics> <mrow> <mi>T</mi> <msub> <mi>P</mi> <mrow> <mi>S</mi> <mi>T</mi> </mrow> </msub> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>F</mi> <msub> <mi>P</mi> <mrow> <mi>S</mi> <mi>T</mi> </mrow> </msub> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>T</mi> <msub> <mi>N</mi> <mrow> <mi>S</mi> <mi>T</mi> </mrow> </msub> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <mi>F</mi> <msub> <mi>N</mi> <mrow> <mi>S</mi> <mi>T</mi> </mrow> </msub> </mrow> </semantics></math>.</p> Full article ">Figure 10
<p>Feature importance in <span class="html-italic">RF_25</span> (<b>left</b>), <span class="html-italic">RF_50</span> (<b>middle</b>) and <span class="html-italic">RF_100</span> (<b>right</b>).</p> Full article ">Figure 11
<p>Training (blue) and validation (orange) losses per epoch for <span class="html-italic">K2D_25</span> (<b>left</b>), <span class="html-italic">K2D_50</span> (<b>middle</b>) and <span class="html-italic">K2D_100</span> (<b>right</b>).</p> Full article ">
15 pages, 5070 KiB  
Article
Potential Ecological Distributions of Urban Adapters and Urban Exploiters for the Sustainability of the Urban Bird Network
by Nurul L. Winarni, Habiburrachman A. H. Fuad, Bhisma G. Anugra, Nabilla Nuril Kaunain, Shania Anisafitri, Mega Atria and Afiatry Putrika
ISPRS Int. J. Geo-Inf. 2022, 11(9), 474; https://doi.org/10.3390/ijgi11090474 - 31 Aug 2022
Cited by 2 | Viewed by 3857
Abstract
The bird community in urban areas indicates the species-specific adaptability to urban conditions such as the increase in man-made habitats. Urban adapters and urban exploiters, two groups that make up most of the urban birds, were assessed to determine their suitable habitat and [...] Read more.
The bird community in urban areas indicates the species-specific adaptability to urban conditions such as the increase in man-made habitats. Urban adapters and urban exploiters, two groups that make up most of the urban birds, were assessed to determine their suitable habitat and explain their distribution, as well as to determine the environmental predictors for the two bird groups assemblages in Depok, one of Jakarta’s satellite cities. We used the point-count method to survey the birds in three habitat types, green spaces, residentials, and roadside, and then we used Maximum Entropy (MaxEnt) to analyze the species distribution modeling. We also the predicted habitat distributions for the urban adapters and urban exploiters based on several environmental predictors. Our results suggest that both urban adapters and urban exploiters were abundant in residential areas. Eurasian tree sparrows (Passer montanus) and cave swiflets (Collocalia linchi) were the most common species in all three habitat types. On average, canopy cover was most extensive in green spaces followed by residential and roadside areas. Urban exploiters were likely to have a high suitability extent compared to urban adapters. The distributions of both groups were affected by the distance to perennial water, then by land function for the urban adapters, and distance to patches for the urban exploiters. The presence of urban adapters and urban exploiters in residential areas suggests that home gardens supported critical habitats when green spaces were unavailable. Full article
(This article belongs to the Special Issue Application of GIS for Biodiversity Research)
►▼ Show Figures

Figure 1

Figure 1
<p>Bird survey locations were spread around the Universitas Indonesia campus, Depok, Indonesia.</p> Full article ">Figure 2
<p>Map of environmental predictors with (<b>A</b>) landscape greenness, (<b>B</b>) land function with land function zone (1. water bodies, 2. roads, 3. business districts and industry, 4. offices and small businesses, 5. buffer area, 6. settlement and housing, 7. public service area, 8. agriculture and tourism, and 9. green open space), (<b>C</b>) temperature (°K) in the survey area, (<b>D</b>) distance to patches, (<b>E</b>) distance to water, and (<b>F</b>) elevation data within the survey area.</p> Full article ">Figure 3
<p>Observed canopy coverage per bird based on urban tolerance classification.</p> Full article ">Figure 4
<p>Habitat suitability of (<b>A</b>) urban adapters, (<b>B</b>) urban exploiters, (<b>C</b>) different values between urban adapters and urban exploiters to emphasize the greater range/niche/distribution of urban exploiter species in urban areas.</p> Full article ">Figure 5
<p>The ROC curve of urban adapter (<b>left</b>), and urban exploiter species (<b>right</b>).</p> Full article ">Figure 6
<p>Response of urban adapters to land function variable (<b>left</b>), the response of urban exploiters to land function variable (<b>right</b>) with (1) water body, (2) roadside, (3) industrial/warehousing, (4) office area/small business, (5) river buffer/catchments, (6) high-density settlement, (7) public facilities, (8) agriculture/tourism, (9) green open spaces.</p> Full article ">
25 pages, 18885 KiB  
Article
Mapping Climate Parameters over the Territory of Botswana Using GMT and Gridded Surface Data from TerraClimate
by Polina Lemenkova
ISPRS Int. J. Geo-Inf. 2022, 11(9), 473; https://doi.org/10.3390/ijgi11090473 - 31 Aug 2022
Cited by 17 | Viewed by 4006
Abstract
This articles presents a new series of maps showing the climate and environmental variability of Botswana. Situated in southern Africa, Botswana has an arid to semi-arid climate, which significantly varies in its different regions: Kalahari Desert, Makgadikgadi Pan and Okavango Delta. While desert [...] Read more.
This articles presents a new series of maps showing the climate and environmental variability of Botswana. Situated in southern Africa, Botswana has an arid to semi-arid climate, which significantly varies in its different regions: Kalahari Desert, Makgadikgadi Pan and Okavango Delta. While desert regions are prone to droughts and periods of extreme heat during the summer months, other regions experience heavy downpours, as well as episodic and unpredictable rains that affect agricultural activities. Such climatic variations affect social and economic aspects of life in Botswana. This study aimed to visualise the non-linear correlations between the topography and climate setting at the country’s scale. Variables included T °C min, T °C max, precipitation, soil moisture, evapotranspiration (PET and AET), downward surface shortwave radiation, vapour pressure and vapour pressure deficit (VPD), wind speed and Palmer Drought Severity Index (PDSI). The dataset was taken from the TerraClimate source and GEBCO for topographic mapping. The mapping approach included the use of Generic Mapping Tools (GMT), a console-based scripting toolset, which enables the use of a scripting method of automated mapping. Several GMT modules were used to derive a set of climate parameters for Botswana. The data were supplemented with the adjusted cartographic elements and inspected by the Geospatial Data Abstraction Library (GDAL). The PDSI in Botswana in 2018 shows stepwise variation with seven areas of drought: (1) −3.7 to −2.2. (extreme); (2) −2.2 to −0.8 (strong, southern Kalahari); (3) −0.8 to 0.7 (significant, central Kalahari; (4) 0.7 to 2.1 (moderate); (5) 2.1 to 3.5 (lesser); (6) 3.5 to 4.9 (low); (7) 4.9 to 6.4 (least). The VPD has a general trend towards the south-western region (Kalahari Desert, up to 3.3), while it is lower in the north-eastern region of Botswana (up to 1.4). Other values vary respectively, as demonstrated in the presented 12 maps of climate and environmental inventory in Botswana. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>Topographic map of Botswana. Mapping: GMT. Source: author.</p> Full article ">Figure 2
<p>T °C minimum (<math display="inline"><semantics> <msub> <mi>T</mi> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> </msub> </semantics></math>) in Botswana. Mapping: GMT. Source: author.</p> Full article ">Figure 3
<p>T °C maximum (<span class="html-italic">T<sub>max</sub></span>) in Botswana. Mapping: GMT. Source: author.</p> Full article ">Figure 4
<p>Precipitation in Botswana (2018). Mapping: GMT. Source: author.</p> Full article ">Figure 5
<p>Soil moisture in Botswana. Mapping: GMT. Source: author.</p> Full article ">Figure 6
<p>Potential evapotranspiration (PET). Data: WorldClim. Mapping: GMT. Source: author.</p> Full article ">Figure 7
<p>Actual evapotranspiration (AET) in Botswana. Mapping: GMT. Source: author.</p> Full article ">Figure 8
<p>Downward surface shortwave radiation. Mapping: GMT. Source: author.</p> Full article ">Figure 9
<p>Vapor pressure in Botswana (2018). Mapping: GMT. Source: author.</p> Full article ">Figure 10
<p>Vapor pressure deficit in Botswana. Mapping: GMT. Source: author.</p> Full article ">Figure 11
<p>Wind speed in Botswana. Mapping: GMT. Source: author.</p> Full article ">Figure 12
<p>PDSI in Botswana. Mapping: GMT. Source: author.</p> Full article ">
18 pages, 9364 KiB  
Review
HBIM Open Source: A Review
by Filippo Diara
ISPRS Int. J. Geo-Inf. 2022, 11(9), 472; https://doi.org/10.3390/ijgi11090472 - 31 Aug 2022
Cited by 19 | Viewed by 4057
Abstract
Historic Building Information Modelling (HBIM) methodology has revolutionized the entire cultural heritage documentation panorama since 2009. At the same time, the possibility of creating and managing HBIM projects by using open source solutions opened new research paths in 2016. Different reasons can drive [...] Read more.
Historic Building Information Modelling (HBIM) methodology has revolutionized the entire cultural heritage documentation panorama since 2009. At the same time, the possibility of creating and managing HBIM projects by using open source solutions opened new research paths in 2016. Different reasons can drive the utilisation of free and open source software (FOSS), however the accessibility of a tailor-made project should be the main purpose. After six years of research on open source HBIM, this paper will review the actual panorama of designed and operative programmes on informative models of historic architecture built with FOSS solutions. Different aspects will be analysed, from open source software setup to parametric modelling and from semantic dimension to data exchange and cloud accessibility. Then, the advantages and drawbacks of open source protocols will be highlighted. Lastly, the next updates, future scenarios and developments on open source HBIM will be estimated. Full article
(This article belongs to the Special Issue Heritage Building Information Modeling: Theory and Applications)
►▼ Show Figures

Figure 1

Figure 1
<p>HBIM projects designed and developed via open source instruments and programming languages: the Staffarda refectory project, the Domus Regia project and the ARK-BIM development.</p> Full article ">Figure 2
<p>The Staffarda refectory project: HBIM model designed from reality-based data and NURBS reconstruction. Stratigraphy and other historical information implemented for future analyses.</p> Full article ">Figure 3
<p>FreeCAD main implementations: from features developed by the community to specific features for HBIM.</p> Full article ">Figure 4
<p>Dynamo and the implementation of DynFreeCAD. The project is currently undergoing development tests and improvements.</p> Full article ">Figure 5
<p>HBIM_Library macro inside FreeCAD for evocating predefined custom parametric models. The column is composed of PART elements such as torus, cubes and cones. The macro is included inside the HBIM workbench.</p> Full article ">Figure 6
<p>The Domus Regia project: from 2D archaeological data to HBIM model and data exchange exclusively with open source solutions.</p> Full article ">Figure 7
<p>From NURBS to HBIM model via FreeCAD open source software: default tools and implemented workbenches for converting NURBS into parametric and informative objects.</p> Full article ">Figure 8
<p>Semantic data inclusion is essential for query processes. Default menu and DynamicData WB allow to enrich the parametric model with technical and historical information.</p> Full article ">Figure 9
<p>Stratigraphic units and analysis implemented inside the HBIM model of the refectory: stratigraphy and interpretation as parametric objects, database, semantic data and images.</p> Full article ">Figure 10
<p>BIMData platform: detail of the sample model inside the 3D viewer. On the right side the properties tree; on the left the BCF plugin for generating reviews.</p> Full article ">Figure 11
<p>From the BIMData platform to ARK-BIM development: default, implemented and unlocked features inside the new CDE.</p> Full article ">Figure 12
<p>The Domus Regia HBIM model inside the ARK-BIM environment. Additional plugins are on the left and right side of the window: details on picked stratigraphic units and “IfcColumn” isolated in the 3D environment.</p> Full article ">Figure 13
<p>Possibilities and advantages of using open source solutions applied to HBIM methodology: from initial reasons to final stages of the project.</p> Full article ">
9 pages, 1302 KiB  
Communication
A Forest of Forests: A Spatially Weighted and Computationally Efficient Formulation of Geographical Random Forests
by Stefanos Georganos and Stamatis Kalogirou
ISPRS Int. J. Geo-Inf. 2022, 11(9), 471; https://doi.org/10.3390/ijgi11090471 - 31 Aug 2022
Cited by 25 | Viewed by 7468
Abstract
The aim of this paper is to present developments of an advanced geospatial analytics algorithm that improves the prediction power of a random forest regression model while addressing the issue of spatial dependence commonly found in geographical data. We applied the methodology to [...] Read more.
The aim of this paper is to present developments of an advanced geospatial analytics algorithm that improves the prediction power of a random forest regression model while addressing the issue of spatial dependence commonly found in geographical data. We applied the methodology to a simple model of mean household income in the European Union regions to allow easy understanding and reproducibility of the analysis. The results are encouraging and suggest an improvement in the prediction power compared to previous techniques. The algorithm has been implemented in R and is available in the updated version of the SpatialML package in the CRAN repository. Full article
(This article belongs to the Special Issue Advances in Machine Learning and Statistical Analysis of Geographical Data)
►▼ Show Figures

Figure 1

Figure 1
<p>Spatial distribution of the independent and dependent variables used in this study at the NUTS2 level: (<b>A</b>) household income, (<b>B</b>) tertiary education, (<b>C</b>) unemployment rate and (<b>D</b>) employment in the tech sector for the year of 2016.</p> Full article ">Figure 2
<p>Bandwidth optimization for the GRF models. The out-of-bag (OOB) coefficient of determination peaked using 20 neighbors as a bandwidth value.</p> Full article ">Figure 3
<p>Bandwidth optimization for the GWR models. The AICc demonstrated a minimum using 23 neighbors as a bandwidth value.</p> Full article ">Figure 4
<p>The computational effect of parallelization with multiple threads in training GRF models with multiple bandwidths.</p> Full article ">
20 pages, 192872 KiB  
Article
Spatial Prediction of COVID-19 Pandemic Dynamics in the United States
by Çiğdem Ak, Alex D. Chitsazan, Mehmet Gönen, Ruth Etzioni and Aaron J. Grossberg
ISPRS Int. J. Geo-Inf. 2022, 11(9), 470; https://doi.org/10.3390/ijgi11090470 - 30 Aug 2022
Cited by 2 | Viewed by 2307
Abstract
The impact of COVID-19 across the United States (US) has been heterogeneous, with rapid spread and greater mortality in some areas compared with others. We used geographically-linked data to test the hypothesis that the risk for COVID-19 was defined by location and sought [...] Read more.
The impact of COVID-19 across the United States (US) has been heterogeneous, with rapid spread and greater mortality in some areas compared with others. We used geographically-linked data to test the hypothesis that the risk for COVID-19 was defined by location and sought to define which demographic features were most closely associated with elevated COVID-19 spread and mortality. We leveraged geographically-restricted social, economic, political, and demographic information from US counties to develop a computational framework using structured Gaussian process to predict county-level case and death counts during the pandemic’s initial and nationwide phases. After identifying the most predictive information sources by location, we applied an unsupervised clustering algorithm and topic modeling to identify groups of features most closely associated with COVID-19 spread. Our model successfully predicted COVID-19 case counts of unseen locations after examining case counts and demographic information of neighboring locations, with overall Pearson’s correlation coefficient and the proportion of variance explained as 0.96 and 0.84 during the initial phase and 0.95 and 0.87 during the nationwide phase, respectively. Aside from population metrics, presidential vote margin was the most consistently selected spatial feature in our COVID-19 prediction models. Urbanicity and 2020 presidential vote margins were more predictive than other demographic features. Models trained using death counts showed similar performance metrics. Topic modeling showed that counties with similar socioeconomic and demographic features tended to group together, and some of these feature sets were associated with COVID-19 dynamics. Clustering of counties based on these feature groups found by topic modeling revealed groups of counties that experienced markedly different COVID-19 spread. We conclude that topic modeling can be used to group similar features and identify counties with similar features in epidemiologic research. Full article
(This article belongs to the Collection Spatial Components of COVID-19 Pandemic)
►▼ Show Figures

Figure 1

Figure 1
<p>Spatial modeling of case dynamics during initial phase of pandemic. Blue shade indicates observed cases over first 30 days in counties used for model training (<b>a</b>) and testing (observed) (<b>b</b>), with predicted case counts in test counties shown in (<b>c</b>). Cases were aggregated over 30 days in each county in the maps. (<b>d</b>) The most predictive top 20 features selected overall by the algorithm for the initial phase. Purple-colored features are negatively correlated with case counts, and the orange-colored features are positively correlated with the case counts. (<b>e</b>) <math display="inline"><semantics> <mrow> <mi>P</mi> <mi>C</mi> <mi>C</mi> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <msup> <mi>R</mi> <mn>2</mn> </msup> </mrow> </semantics></math> values of all predictions.</p> Full article ">Figure 2
<p>Spatial modeling of case dynamics during nationwide phase of pandemic. Blue shade indicates observed cases over first 30 days in counties used for model training (<b>a</b>) and testing (observed) (<b>b</b>), with predicted case counts in test counties shown in (<b>c</b>). Cases were aggregated over the time period after 11 September 2020 until 21 March 2021 in each county in the maps. (<b>d</b>) The most predictive top 20 features selected overall by the algorithm for the nationwide phase. Purple-colored features are negatively correlated with case counts, and the orange-colored features are positively correlated with case counts. (<b>e</b>) <math display="inline"><semantics> <mrow> <mi>P</mi> <mi>C</mi> <mi>C</mi> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <msup> <mi>R</mi> <mn>2</mn> </msup> </mrow> </semantics></math> values of the predictive models on a state-by-state level.</p> Full article ">Figure 3
<p>Correlation matrix of the spatial features used in the <math display="inline"><semantics> <mrow> <mi>S</mi> <mi>G</mi> <mi>P</mi> </mrow> </semantics></math> model. Blue indicates positive correlation, red indicates negative correlation, and cross indicates no correlation. Shade indicates strength of correlation per scale shown at bottom of matrix.</p> Full article ">Figure 4
<p>Topic modeling identifies associations between sets of spatial features and COVID-19 dynamics. (<b>a</b>) Top 10 feature scores for features associated with topic 8. (<b>b</b>) Topic 8 scores for each county in the US. Legend of the map is the same as the topic score heatmaps given in (<b>c</b>,<b>e</b>). (<b>c</b>) Heatmap of each county z-scored topic score against the mean deaths during the nationwide phase, binned into quintiles. To highlight the relationships between topic scores and deaths, the heatmap is sorted by topic 8. (<b>d</b>) Boxplot of topic scores for each county across death quintiles for topic 8, showing positive correlation with death counts. (<b>e</b>) Heatmap of each county z-scored topic score against the mean deaths during the nationwide phase, binned into quintiles. In order to highlight the relationships between topic scores and deaths, the heatmap is sorted by topic 10. (<b>f</b>) Boxplot of topic scores for each county across death quintiles for topic 10, showing negative correlation with death counts. (<b>g</b>) Top 10 feature scores for features associated with topic 10. (<b>h</b>) Topic 10 scores for each county in the US. Legend of the map is the same as the topic score heatmaps given in (<b>c</b>,<b>e</b>).</p> Full article ">Figure 5
<p>Counties clustered using spatial topics show similar patterns in COVID-19 cases/death counts. Clustering by topics can identify high- and low-risk counties. (<b>a</b>) Geographical map of counties and their discrete cluster assignments when topic-county matrix inputted into Louvain clustering. (<b>b</b>) Mean topic score for each topic for each of the 9 clusters of counties. (<b>c</b>) Bar graph of the number of countries within each cluster that fall within each quintile bin of cases and deaths for the initial as well as nationwide phases of the pandemic.</p> Full article ">Figure A1
<p>Overview of our predictive computational framework structured Gaussian process regression (<math display="inline"><semantics> <mrow> <mi>S</mi> <mi>G</mi> <mi>P</mi> <mi>R</mi> </mrow> </semantics></math>).</p> Full article ">Figure A2
<p>(<b>a</b>)US-wide total 7-day moving average case counts per 100 thousand population and the dates we selected for analysis of early and late pandemic dynamics. Red lines are at 6 April 2020, 11 September 2020, and 21 March 2021. (<b>b</b>) 7-day moving average case counts of the first month with a case of each US county and US-wide cumulative case counts of the first month with a case.</p> Full article ">Figure A3
<p>(<b>a</b>) Selected predictive features for each state by the algorithm for the initial phase prediction of cases. (<b>b</b>) <math display="inline"><semantics> <mrow> <mi>P</mi> <mi>C</mi> <mi>C</mi> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <msup> <mi>R</mi> <mn>2</mn> </msup> </mrow> </semantics></math> values of the predictions reported as a box plot.</p> Full article ">Figure A4
<p>(<b>a</b>–<b>c</b>) On three different maps, we presented the data used for training our model for the initial phase prediction of deaths, the test data (i.e., holdback, observed) and our predictions for test locations. Deaths were aggregated over 30 days in each county in the maps. (<b>d</b>) Selected predictive features for each state. (<b>e</b>) <math display="inline"><semantics> <mrow> <mi>P</mi> <mi>C</mi> <mi>C</mi> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <msup> <mi>R</mi> <mn>2</mn> </msup> </mrow> </semantics></math> values of the predictions reported as a heatmap per state. One can identify the states falling far from the observed versus predicted line from the accuracy heatmap. (<b>f</b>) The most predictive top 20 features selected overall by the algorithm for the initial phase. Purple-colored features are negatively correlated with the death counts and the orange-colored features are positively correlated with the case counts. (<b>g</b>) <math display="inline"><semantics> <mrow> <mi>P</mi> <mi>C</mi> <mi>C</mi> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <msup> <mi>R</mi> <mn>2</mn> </msup> </mrow> </semantics></math> values of the predictions reported as a box plot.</p> Full article ">Figure A5
<p>(<b>a</b>) Selected predictive features for each state by the algorithm for the nationwide phase prediction of cases. (<b>b</b>) <math display="inline"><semantics> <mrow> <mi>P</mi> <mi>C</mi> <mi>C</mi> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <msup> <mi>R</mi> <mn>2</mn> </msup> </mrow> </semantics></math> values of the predictions reported as a box plot.</p> Full article ">Figure A6
<p>(<b>a</b>–<b>c</b>) On three different maps, we presented the data used for training our model for the nationwide phase prediction of deaths, the test data (i.e., holdback, observed) and our predictions for test locations. Deaths were aggregated over the time period after 11 September 2020 until 21 March 2021 in each county in the maps. (<b>d</b>) Selected predictive features for each state. (<b>e</b>) <math display="inline"><semantics> <mrow> <mi>P</mi> <mi>C</mi> <mi>C</mi> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <msup> <mi>R</mi> <mn>2</mn> </msup> </mrow> </semantics></math> values of the predictions reported as a heatmap per state. (<b>f</b>) The most predictive top 20 features selected overall by the algorithm for the nationwide phase. Purple-colored features are negatively correlated with the death counts and the orange-colored features are positively correlated with the case counts. (<b>g</b>) <math display="inline"><semantics> <mrow> <mi>P</mi> <mi>C</mi> <mi>C</mi> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <msup> <mi>R</mi> <mn>2</mn> </msup> </mrow> </semantics></math> values of the predictions reported as a box plot.</p> Full article ">Figure A7
<p>(<b>a</b>,<b>b</b>) The most predictive top 20 features selected overall by the algorithm for case and death count predictions from 11 September 2020 to 1 January 2021, respectively. Purple-colored features are negatively correlated with the case/death counts and the orange-colored features are positively correlated with the case/death counts, respectively.</p> Full article ">Figure A8
<p>Top 10 feature scores for spatial features associated with each topic.</p> Full article ">Figure A9
<p>Normalized topic scores for each county in the US. Legend of the map is the same as the as the topic score heatmaps given in (<a href="#ijgi-11-00470-f004" class="html-fig">Figure 4</a>c,e) where given a topic, counties with colors closer to orange have a high topic value and counties with colors closer to purple have low topic value.</p> Full article ">
38 pages, 25580 KiB  
Article
Heri-Graphs: A Dataset Creation Framework for Multi-Modal Machine Learning on Graphs of Heritage Values and Attributes with Social Media
by Nan Bai, Pirouz Nourian, Renqian Luo and Ana Pereira Roders
ISPRS Int. J. Geo-Inf. 2022, 11(9), 469; https://doi.org/10.3390/ijgi11090469 - 30 Aug 2022
Cited by 22 | Viewed by 4038
Abstract
Values (why to conserve) and Attributes (what to conserve) are essential concepts of cultural heritage. Recent studies have been using social media to map values and attributes conveyed by the public to cultural heritage. However, it is rare to connect heterogeneous modalities of [...] Read more.
Values (why to conserve) and Attributes (what to conserve) are essential concepts of cultural heritage. Recent studies have been using social media to map values and attributes conveyed by the public to cultural heritage. However, it is rare to connect heterogeneous modalities of images, texts, geo-locations, timestamps, and social network structures to mine the semantic and structural characteristics therein. This study presents a methodological framework for constructing such multi-modal datasets using posts and images on Flickr for graph-based machine learning (ML) tasks concerning heritage values and attributes. After data pre-processing using pre-trained ML models, the multi-modal information of visual contents and textual semantics are modelled as node features and labels, while their social relationships and spatio-temporal contexts are modelled as links in Multi-Graphs. The framework is tested in three cities containing UNESCO World Heritage properties—Amsterdam, Suzhou, and Venice— which yielded datasets with high consistency for semi-supervised learning tasks. The entire process is formally described with mathematical notations, ready to be applied in provisional tasks both as ML problems with technical relevance and as urban/heritage study questions with societal interests. This study could also benefit the understanding and mapping of heritage values and attributes for future research in global cases, aiming at inclusive heritage management practices. Moreover, the proposed framework could be summarized as creating attributed graphs from unstructured social media data sources, ready to be applied in a wide range of use cases. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>The framework to create multi-modal machine learning datasets as attributed graphs from unstructured data sources.</p> Full article ">Figure 2
<p>Data flow of the multi-modal feature generation process of one sample post in Venice, while graph construction requires all data points of the dataset. The original post owned by user <tt>17726320@N03</tt> is under CC BY-NC-SA 2.0 license.</p> Full article ">Figure 3
<p>The proportion of posts and sentences that are predicted and labeled as each heritage value (OUV selection criterion) as top-3 predictions by both BERT and ULMFiT. One typical sentence from each category is also given in the right part of the figure.</p> Full article ">Figure 4
<p>Typical image examples in each city labelled as each heritage attribute category (depicted scene) and bar plots of their proportions in the datasets (length of bright blue background bars represent 50%). Three examples with high confidence and one negative example with low confidence (in red frame) are given. All images are 150 × 150 px “thumbnails” flagged as “downloadable”.</p> Full article ">Figure 5
<p>The back-end geographical networks for three case studies, respectively, showing the graph structure, degree ranking distribution, and the ranking distribution of posts per geo-spatial node (on a logarithm scale) in Amsterdam, Suzhou, Venice, and Venice-XL. The sizes of nodes denote the number of nearby posts allocated to the nodes, and the colors of nodes illustrate the degree of the node on the graph. Each link connects two nodes reachable to each other within 20 min.</p> Full article ">Figure 6
<p>The rank-size plots of the degree distributions in the three cases of Amsterdam, Suzhou, and Venice, with regard to the temporal links, social links, spatial links, as well as the entire multi-graph.</p> Full article ">Figure A1
<p>The subgraphs of the multi-graphs in each case study city visualized using spring layout in NetworkX. The node size and colour reflect the degrees, and link thickness the edge weights.</p> Full article ">
25 pages, 6340 KiB  
Article
HBIM Meta-Modelling: 50 (and More) Shades of Grey
by Martina Attenni, Carlo Bianchini, Marika Griffo and Luca James Senatore
ISPRS Int. J. Geo-Inf. 2022, 11(9), 468; https://doi.org/10.3390/ijgi11090468 - 30 Aug 2022
Cited by 6 | Viewed by 1978
Abstract
The paper aims at investigating modelling strategies in HBIM context to identify at what extent the final use of the model might affects, or should affect, the modelling approach itself. Moreover, the discussion wants to shed light on the possibility of connecting in [...] Read more.
The paper aims at investigating modelling strategies in HBIM context to identify at what extent the final use of the model might affects, or should affect, the modelling approach itself. Moreover, the discussion wants to shed light on the possibility of connecting in just one digital environment several instances connected to the building. These aims will be discussed presenting and evaluating two different modelling approaches: the “black box” modelling and the “white box” model-ling. The two terms are partially borrowed from computer science to explain two types of testing. The “black box” testing is performed without any preliminary knowledge about the system functionality and internal components; on the contrary, the “white box” testing, implies a full knowledge of the system. These two approaches will be compared to two ways of conceiving a building information model. In conclusion, the paper will investigate the possibility to integrate in just one model, the grey box model, the two ones previously discussed. Full article
(This article belongs to the Special Issue Heritage Building Information Modeling: Theory and Applications)
►▼ Show Figures

Figure 1

Figure 1
<p>(<b>a</b>) The building in Piazza Borghese in its current state. (<b>b</b>) Piazza Borghese as represented in a G. B. Falda drawing, 1943. (<b>c</b>) Piazza Borghese in a planimetric view of Rome, made by Giuseppe Vasi in 1676.</p> Full article ">Figure 2
<p>The building in Piazza Borghese, analysis, and classification of main architectural element.</p> Full article ">Figure 3
<p>(<b>a</b>) The building in Piazza Borghese, point cloud. (<b>b</b>) Planimetric view of the building ground floor, point cloud. (<b>c</b>) Planimetric view of the building first floor, point cloud.</p> Full article ">Figure 4
<p>The building in Piazza Borghese, “black box” model.</p> Full article ">Figure 5
<p>The building in Piazza Borghese, the elements of the “black box” model.</p> Full article ">Figure 6
<p>The Botany Institute.</p> Full article ">Figure 7
<p>The Botany Institute, views from numerical model derived from 3D integrated survey.</p> Full article ">Figure 8
<p>The Botany Institute, historical image of the internal space derived from bibliographic research.</p> Full article ">Figure 9
<p>The Botany Institute. Detailed view of constructive solutions for external surfaces (<b>right</b>) and their localization in the BIM model (<b>left</b>).</p> Full article ">Figure 10
<p>The Botany Institute. “White box” model.</p> Full article ">Figure 11
<p>“Black box”, “white box” and “grey box” approach.</p> Full article ">Figure 12
<p>Comparison between “black box” and “white box” models.</p> Full article ">
32 pages, 16733 KiB  
Article
Metric, Topological, and Syntactic Accessibility in Three-Dimensional Urban Networked Spaces: Modeling Options and Visualization
by Thi Hong Diep Dao and Jean-Claude Thill
ISPRS Int. J. Geo-Inf. 2022, 11(9), 467; https://doi.org/10.3390/ijgi11090467 - 29 Aug 2022
Cited by 1 | Viewed by 1987
Abstract
In this paper, we take the position that cities gain to be represented as three-dimensional spaces populated by scores of micro-scale-built spaces (buildings, rooms, passageways, squares, etc.). Effective algorithms that evaluate place-based accessibility in built structures while considering the indoor spaces’ complexity at [...] Read more.
In this paper, we take the position that cities gain to be represented as three-dimensional spaces populated by scores of micro-scale-built spaces (buildings, rooms, passageways, squares, etc.). Effective algorithms that evaluate place-based accessibility in built structures while considering the indoor spaces’ complexity at a fine granularity are essential for indoor–outdoor seamless urban planning, navigation, way findings, and supporting emergencies. We present a comprehensive set of spatial modeling options and visualizations of indoor accessibility for an entire built structure based on various notions of travel impedance. Notably, we consider the metric length of the paths and their cognitive complexities due to topologic, syntactic, or integrated intricacy within our approaches. Our work presents a comprehensive selection of indoor accessibility analysis with a detailed implemental discussion that can be applied as a solid foundation for smart city applications or seamless urban research and planning. The analysis and visualization techniques presented in this paper can be easily applied to analyze and visualize built interior geographic spaces to study accessibility differentials in cities with vast vertical expansion aimed at achieving (or at avoiding) specific accessibility outcomes. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>Angular value, or the change in travel direction, in degree between two traveling arcs.</p> Full article ">Figure 2
<p>A long corridor represented as a composition of arcs (<b>a</b>) and as a stroke (<b>b</b>).</p> Full article ">Figure 3
<p>Floor turn representation.</p> Full article ">Figure 4
<p>Syntactic turn definition.</p> Full article ">Figure 5
<p>Angular cognitive turn definition.</p> Full article ">Figure 6
<p>Metric-weighted angular turn definition.</p> Full article ">Figure 7
<p>Topological break-point definition.</p> Full article ">Figure 8
<p>Metric impedance (minutes) of traveling paths and accessibility estimation with α = 2.</p> Full article ">Figure 9
<p>Syntactic impedance of traveling paths and accessibility estimation with α = 2.</p> Full article ">Figure 10
<p>Angular impedance of traveling paths and accessibility estimation with α = 2.</p> Full article ">Figure 11
<p>Topological impedance of traveling paths and accessibility estimation with α = 2.</p> Full article ">Figure 12
<p>The metric-syntactic impedances of traveling paths and accessibility estimation with α = 2.</p> Full article ">Figure 13
<p>The metric-angular impedance of traveling paths and accessibility estimation with α = 2.</p> Full article ">Figure 14
<p>Accessibility estimation workflow.</p> Full article ">Figure 15
<p>2.5D model of the Ellicott complex and its dining locations.</p> Full article ">Figure 16
<p>Strokes identified on each floor of the test complex.</p> Full article ">Figure 17
<p>Metric, syntactic, and topological accessibility to dining facility 3: classification in quintiles.</p> Full article ">Figure 18
<p>Angular, metric-syntactic, and metric-angular accessibility to dining facility 3: classification in quintiles.</p> Full article ">Figure 19
<p>Floor 1 accessibility to dining facility 3 (P = point of analysis, highlighted for comparison among different approaches) based on different approaches: classification in quintiles.</p> Full article ">Figure 20
<p>Metric accessibility to dining facility 3 on different floors: classification in quintiles.</p> Full article ">Figure 21
<p>Syntactic accessibility to dining facility 3 on different floors: classification in quintiles.</p> Full article ">Figure 22
<p>Metric, syntactic, and topological accessibility to three dining facilities: classification in quintiles.</p> Full article ">Figure 23
<p>Angular, metric-syntactic, and metric-angular accessibility to three dining facilities: classification in quintiles.</p> Full article ">Figure A1
<p>Arial photograph of the Ellicott complex.</p> Full article ">Figure A2
<p>Floor 1 accessibility by different modeling approaches to three dining facilities: classification in quintiles.</p> Full article ">Figure A3
<p>Metric accessibility to three facilities for different floors: classification in quintiles.</p> Full article ">Figure A4
<p>Syntactic accessibility to three facilities for different floors: classification in quintiles.</p> Full article ">
20 pages, 3356 KiB  
Article
Commuting Analysis of the Budapest Metropolitan Area Using Mobile Network Data
by Gergő Pintér and Imre Felde
ISPRS Int. J. Geo-Inf. 2022, 11(9), 466; https://doi.org/10.3390/ijgi11090466 - 29 Aug 2022
Cited by 4 | Viewed by 5584
Abstract
The analysis of human movement patterns based on mobile network data makes it possible to examine a very large population cost-effectively and has led to several discoveries about human dynamics. However, the application of this data source is still not common practice. The [...] Read more.
The analysis of human movement patterns based on mobile network data makes it possible to examine a very large population cost-effectively and has led to several discoveries about human dynamics. However, the application of this data source is still not common practice. The goal of this study was to analyze the commuting tendencies of the Budapest Metropolitan Area using mobile network data as a case study and propose an automatized alternative approach to the current, questionnaire-based method, as commuting is predominantly analyzed by the census, which is performed only once in a decade in Hungary. To analyze commuting, the home and work locations of cell phone subscribers were determined based on their appearances during and outside working hours. The detected home locations of the subscribers were compared to census data at a settlement level. Then, the settlement and district level commuting tendencies were identified and compared to the findings of census-based sociological studies. It was found that the commuting analysis based on mobile network data strongly correlated with the census-based findings, even though home and work locations were estimated by statistical methods. All the examined aspects, including commuting from sectors of the agglomeration to the districts of Budapest and the age-group-based distribution of the commuters, showed that mobile network data could be an automatized, fast, cost-effective, and relatively accurate way of analyzing commuting, that could provide a powerful tool for sociologists interested in commuting. Full article
►▼ Show Figures

Figure 1

Figure 1
<p>The observation area (<b>a</b>) including Budapest (brown), and its agglomeration (green) in relation to Pest county, and the district groups (<b>b</b>) of Budapest, defined by the Hungarian Central Statistical Office (HCSO).</p> Full article ">Figure 2
<p>The SIM cards in the “April 2017” data set categorized by the number of CDRs. The <b>left</b> figure shows the number of CDRs and the <b>right</b> figure shows the number of SIM cards in each category. The SIM cards over a thousand records (17.7%) provide the majority (75.48%) of the activity.</p> Full article ">Figure 3
<p>SIM card distribution of the “April 2017” data set, by the number of active days. The <b>left</b> figure shows the number of CDRs and the <b>right</b> figure shows the number of SIM cards in each category. Around 45.13% of the SIM cards have activity on at least 14 different days.</p> Full article ">Figure 4
<p>The number of cell phone subscribers (<b>a</b>), and the ratio of all mobile network activity (<b>b</b>) by age category, when the subscriber age is known.</p> Full article ">Figure 5
<p>Workflow of the data processing.</p> Full article ">Figure 6
<p>Mixed cell centroid (<b>a</b>), Voronoi polygons using the original cells (<b>b</b>), and Voronoi polygons after merging the close cells (<b>c</b>).</p> Full article ">Figure 7
<p>The schema of obtained, “raw” data (<b>a</b>), and the normalized data tables (<b>b</b>).</p> Full article ">Figure 8
<p>Comparing the ground truth (the population registered by the HCSO [<a href="#B2-ijgi-11-00466" class="html-bibr">2</a>]) (<b>a</b>) and the detected population based on mobile network data (<b>b</b>). The district numbers of Budapest are also displayed.</p> Full article ">Figure 9
<p>Correlation between the population of the agglomeration and the 23 districts of Budapest based on HCSO (ground truth) and mobile network data. The markers represent the settlements or districts (in the case of Budapest). In <b>left</b> figure (<b>a</b>), all the SIM cards were used, in the <b>right</b> figure (<b>b</b>), SIM cards that certainly operated in nonphone devices were excluded.</p> Full article ">Figure 10
<p>Using kernel density plots (with Gaussian kernel) to display the typical working locations for three selected settlements and a district of Budapest, with (<b>a</b>,<b>c</b>,<b>e</b>) and without (<b>b</b>,<b>d</b>,<b>f</b>) local workers. The administrative boundaries of Budapest and the selected settlements are also displayed.</p> Full article ">Figure 11
<p>Connection by commuters between Budapest districts (numbered nodes) based on the home and work locations. A link represents how strong the commuting between two districts is. The weak links are omitted to improve visibility. The district nodes are colored by the district groups.</p> Full article ">Figure 12
<p>Comparison between the census-based (<b>a</b>) (Figure 1 in [<a href="#B9-ijgi-11-00466" class="html-bibr">9</a>]) and the CDR (<b>b</b>) commuting ratios for the districts of Budapest, from the same district, other parts of Budapest, the agglomeration or out of the agglomeration.</p> Full article ">Figure 13
<p>Commuting from the six sectors of the urban agglomeration, based on CDR evaluation.</p> Full article ">Figure 14
<p>Commuting to the seven district groups of Budapest from selected settlements of the agglomeration, comparing census (1990, 2001, and 2011) and mobile network data. Next to the legends, the location of the settlements in question is displayed on a map.</p> Full article ">Figure 15
<p>Distribution of the commuters by age categories and the sectors of the agglomeration (%). Comparison between microcensus data (Table 1 in [<a href="#B9-ijgi-11-00466" class="html-bibr">9</a>]) (<b>a</b>) and mobile network data (<b>b</b>).</p> Full article ">
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
 
Displaying articles 1-38
Previous Issue
Next Issue
Back to TopTop