20250905_3DGeoInfo

Transcript

1. 2025/09/05 20th International 3D GeoInfo Conference 2025 1 Analysis of Building 3D Attribute Coverage and Spatial Disparity of Editing Activities in OpenStreetMap Toshikazu SETO* A02 KOMAZAWA UNIVERSITY Visual Identity Guidelines Ϋʴ࿨จϩΰλΠϓ ,ϚʔΫʴ࿨จϩΰλΠϓͷ૊Έ߹Θͤ ͸ɺ ࠨͷछͰ͢ɻ ԣ جຊܗͱ͠ɺ ༏ઌతʹ࢖༻͠·͢ɻ ԣ ϫϯϙΠϯτͳͲɺ ʮԣʯ ͕഑ஔ͠ʹ͍͘ ৔߹ʹ࢖༻͠·͢ɻ ॎ ॎܕαΠϯͳͲɺ ࡉ௕͍഑ஔʹ࢖༻͠·͢ɻ ඞͣϚελʔσʔλΛ࢖༻͍ͯͩ͘͠͞ɻ ࠨهҎ֎ͷ૊Έ߹ΘͤΛ࡞੒͠ͳ͍Ͱ ͍ͩ͘͞ɻ ܗʣ ϙΠϯτ౳ʣ ॎ *: Associate Professor: Komazawa University Center for Spatial Information Science, the University of Tokyo Project PLATEAU Advisory Board tosseto@komazawa-u.ac.jp
2. 2025/09/05 20th International 3D GeoInfo Conference 2025 2 1. Background • Research Objectives and Significance • Data and Methodology • National Overview of 3D Attribute Data 2. Regional Distribution and Inequality 3. Editor Activity Concentration Analysis 4. Factors Behind Regional Inequality 5. Conclusions Outline GitHub: https://github.com/tossetolab/osm-3d-stat
3. 2025/09/05 20th International 3D GeoInfo Conference 2025 3 • Rising Demand in Multiple Sectors • Urban planning, disaster prevention, and smart city development • Government-led initiatives: Japan's "Project PLATEAU" • High-precision data development challenges: • Cost constraints • Time-intensive processes • VGI as Alternative Solution • Volunteered Geographic Information (VGI) gaining attention • Cost-effective approach to 3D data development • Community-driven data creation model Research Background (1) - Growing Demand for 3D Urban Models
4. 2025/09/05 20th International 3D GeoInfo Conference 2025 4 • OSM Overview (since 2004) • World's largest open-source geographic information platform • ~10 million registered users (February 2025) • ~510 million buildings worldwide • Open XML format under open license • Major 3D Attribute Types in OSM Key • height (building height) • building:levels (number of floors) • min_height (minimum height) • building:levels:underground (basement floors) • ele (elevation) • VGI Challenges • Regional inequality in data quality • Editor dependence issues ・import PLATEAU data (Furuhashi,2023) Research Background (2) - OpenStreetMap Characteristics and Challenges
5. 2025/09/05 20th International 3D GeoInfo Conference 2025 5 • Study Area and Data • Coverage: All of Japan (47 prefectures, 1,905 municipalities) • Dataset period: snapshot of February 1, 2025 • Source: Geofabrik dump data • Target 3D Attributes • height, • building:levels, • min_height, • ele • building:levels:underground Data and Methodology Measurement OSM Activities • Herfindahl-Hirschman Index (HHI) • where Si represents the distribute share (%) of editor i • Maximum value: 10,000 • >2,500: high concentration !!" = $ %! " # !$%
6. 2025/09/05 20th International 3D GeoInfo Conference 2025 6 N03-23_230101.shp japan-250201-internal.osm.pbf Input Data Source Phase1: Setup Database (import_osm_buildings.sh) 1. Initializing the database PostgreSQL + PostGIS 2. Creating schema admin_boundaries, editors, building_history 4. Data integration and area calculation spatial connection (building x administrative area) → building_history creation Phase2: Analyzing Data (building_analysis.py) 5. Performing statistical analysis City-level building statistics, building type analysis, 3D coverage, editor analysis, time series analysis, prefecture analysis Output .CSV city_building_stats.csv building_type_stats.csv 3d_coverage.csv editor_stats.csv area_stats.csv monthly_activity.csv prefecture_analysis.csv 3b. Processing OSM data osm2pgsql 3a. Processing Boundaries ogr2ogr / calculated area admin_boundaries planet_osm_polygon file Name item summary 3d_coverage.csv (9 items) prefcode, total_buildings, height_count, levels_count, min_height_count, underground_count, ele_count, height_coverage, levels_coverage prefecture_analysis.csv (14 items) prefcode, building_count, editor_count, gini_coefficient, top_editor, top_editor_share, top5_editor_share, top10pct_editor_share, hhi, height_count, height_coverage_rate, levels_count, levels_coverage_rate, first_edit_date, last_edit_date editor_stats.csv (11 items) uid, first_edit, last_edit, total_edits, building_edits, active_prefectures, building+height, building_types, height_edits, levels_edits, height_edit_rate
7. 2025/09/05 20th International 3D GeoInfo Conference 2025 7 • Summary • Total buildings: 24,183,256 • 3D attribute coverage rates (national average): • height: 1,125,395 buildings (4.7%) • levels: 1,104,648 buildings (4.6%) • Other attributes: 0.002-4.6% • Building Characteristics • Average area: 288.5 m² (all buildings), 1,260.3 m² (with 3D attributes) • Key finding: 3D attributes concentrated in larger buildings Overview of 3D Attribute OSM Data in JAPAN
8. 2025/09/05 20th International 3D GeoInfo Conference 2025 8 Number of buildings by type and 3D attributes Rank building_type count height levels underground ele avg_height avg_levels avg_area_sq 1 yes 21,019,547 388,685 254,883 56,518 382,107 6.2 7.28 132.63 2 house 1,704,721 529,163 487,023 53,975 527,392 7.08 1.93 88.81 3 detached 349,641 264 57,466 8 118 7.45 2.07 87.92 4 apartments 277,299 57,692 126,101 2,420 56,399 12.13 4.09 302.8 5 residential 183,740 31,137 30,226 2,089 30,827 8.12 2.39 122.4 6 greenhouse 126,243 546 9,868 0 67 1.91 1 392.86 7 roof 85,623 748 4,890 22 465 8.27 1.07 103.92 8 industrial 68,468 27,063 22,986 2,821 26,959 7.27 1.56 1320.7 9 retail 68,067 16,629 22,946 1,708 16,179 8.07 2.05 836.75 10 school 39,365 1,872 4,210 50 1,725 13.17 2.82 991.39 11 farm_auxiliary 36,934 21 123 0 22 6.86 1.46 140.74 12 commercial 36,058 17,577 21,028 1,057 17,079 10.68 3.34 411.57 13 warehouse 27,504 14,988 14,918 689 14,753 5.56 1.23 661.29 14 public 26,431 22,012 16,536 2,217 21,982 7.54 1.71 344.66 15 garage 20,831 65 6,805 7 20 4.34 1.06 74.01 Analysis by Building Type Unknown use buildings (building=yes) • 21,019,547 buildings (86.9% of total) • height attribute: 388,685 buildings (1.85%) • levels attribute: 254,883 buildings (1.21%) Residential buildings (house) • 1,704,721 buildings (7.05% of total) • height attribute: 529,163 buildings (2.19%) • levels attribute: 487,023 buildings (2.01%)
9. 2025/09/05 20th International 3D GeoInfo Conference 2025 9 Regional Distribution and inequality (1) • Top Pref. by Building Count • Hokkaido: 2,688,753 buildings (11.12%) • Aichi: 1,694,860 buildings (7.01%) • Osaka: 1,651,393 buildings (6.83%) • Top Pref. by 3D Attr. • Nagano: 46.2% • Aichi: 46.2% • Ibaraki: 27.2% • Osaka: 21.7% • Significant inequality • Hokkaido: Only 0.7% coverage despite highest building count
10. 2025/09/05 20th International 3D GeoInfo Conference 2025 10 Nagano Pref. ・Saku-city ・Chino-city ・Okaya-city Aichi Pref. ・Nagoya-city Ibaraki Pref. ・Hokota-city Hyogo Pref. ・Asago-city ・Mitsugi-city Fukuoka Pref. ・Ukiha-city • High-Coverage areas: • Nagano Prefecture: • Saku: 99.7% (87,006 bldgs.) • Chino: 99.9% (68,373 bldgs.) • Okaya: 99.9% (34,149 bldgs.) • Other high-coverage areas: • Hokota City, Ibaraki: 99.9% (62,823 buildings) • Common Characteristics • Limited existing OSM data • PLATEAU data import implementation Regional Distribution and inequality (2)
11. 2025/09/05 20th International 3D GeoInfo Conference 2025 11
12. 2025/09/05 20th International 3D GeoInfo Conference 2025 12 • Editor Numbers and Activity • Total editors: 24,394 (2008-2025) • Prefecture-level editor counts: • Range: 195 (Miyazaki) to 3,276 (Tokyo) • ~17-fold regional inequality • Top Editor Share • National average: 35.3% • Range: 5.5% (Oita) to 84.2% (Akita) • ~15-fold inequality Editor Activity Concentration Analysis (1) Percentage of OSM buildings edited in the top editors.
13. 2025/09/05 20th International 3D GeoInfo Conference 2025 13 • HHI: Si represents the market share (%) of editor i • National average: 2,075 • Distribution: • Low concentration (<1,500): 22 prefectures (46.8%) • High concentration (>2,500): 18 prefectures (38.3%) • Extreme Concentration Cases • 6 prefectures with HHI >4,000: Akita, Saga, Yamaguchi, Nara, Yamanashi, Tottori • Average editors: 376 • Top editor share: 61.8-84.2% Editor Activity Concentration Analysis (2) HHI analysis of editorial occupancy of OSM buildings !!" = $ %! " # !$%
14. 2025/09/05 20th International 3D GeoInfo Conference 2025 14 • Major Contributing Factors • Geographic concentration of editing communities • Varying practical needs for 3D building data • Impact of Project PLATEAU data imports • Sustainability Concerns • Risk of top editor departure in rural areas • VGI inherent characteristics: "voluntariness" and "heterogeneity" • "1% rule" more pronounced in geographically constrained regions Factors Behind Regional inequality (1)
15. 2025/09/05 20th International 3D GeoInfo Conference 2025 15 • Correlation Analysis • Building count vs. editor count: r=0.608 (p<0.05) • Height coverage vs. levels coverage: r=0.602 (p<0.05) • Editor count vs. levels coverage: r=0.328 (p<0.05) • Editor count vs. height coverage: No significant correlation (r=0.256) • Implications • Height data development depends more on editing processes and specific editor skills than editor numbers • Different skill requirements for different 3D attributes Factors Behind Regional inequality (2)
16. 2025/09/05 20th International 3D GeoInfo Conference 2025 16 Conclusion and Future Directions • Key Findings: • Low 3D attribute penetration: Only 4.7% of OSM bldgs in Japan • Significant regional inequality: Nagano 46.2% - Hokkaido 0.7% • Urban distributed vs. rural concentrated patterns • Future Works: • Comparative analysis with Overture Maps • Bottom-up VGI vs. top-down commercial data characteristics comparison • Machine learning-based 3D attribute estimation accuracy evaluation • Practical Challenges • Utilizing external OSM resources for 3D attribute enrichment • Developing community revitalization strategies • Optimizing data development prioritization
17. 2025/09/05 20th International 3D GeoInfo Conference 2025 17 Thank You ! @tosseto tosseto@komazawa-u.ac.jp https://speakerdeck.com/tosseto GitHub: https://github.com/tossetolab/osm-3d-stat