Reproducibility package for Reimagining Logistics and Connectivity from Space: Artificial Intelligence and Satellite Imagery Applications

Copernicus Sentinel-1 Ground Range Detected (GRD) SAR imagery

Raw Sentinel-1 scenes are not in the package; only project-processed derivatives (occupancy / queue-length estimates, variables Date; Occupancy) are included. Historical processing used the project's Sentinel Hub service and is not part of the package. Observations used: Figure 1, Figoil/Sarpi TIR park (park 14), 77 observation pairs, 19 April 2023 to 19 February 2024; Figure 14, 18 Georgian TIR parks, 907 observation pairs over the same period; Figure 3, Lebombo/Ressano Garcia border crossing, 445 observations, 3 January 2018 to 3 December 2023. Exact AOI polygons and full product IDs were not recovered. Files location: Figure1_WBG_R_closeout_package/data/raw/remote_sensing/sentinel1_processed_sar_estimates/14.xls (identical to Figure 14 .../14.xls); Figure14_WBG_R_closeout_package/data/processed/worldbank_georgia/1.xls to 26.xls (one per TIR park) and data/raw/kmls/1.kml to 26.kml (TIR park boundary polygons, geometry only); Figure3_WBG_PR_closeout_package/data/raw/prediction_ID_002_Lebombo_Sentinel1.zip (598 GeoTIFFs, 2016-2024, a broader set than the observations charted) and reference/source_uploads/prediction_ID_002_Lebombo_Sentinel1_results_timeSeries.zip (23 CSVs). Not included: Figure3_WBG_PR_closeout_package/data/report_source/Figure3.xlsx, the workbook read by the Figure 3 code, because it also contains PlanetScope-derived detections and Crickmay GPS data; and Figure3_WBG_PR_closeout_package/reference/arcgis/Figure3_SAR.ppkx (ArcGIS package with two Sentinel-1 rasters and Esri basemap references). Figure 3 therefore cannot be rerun from the public package. Figure 4: the original inputs were not part of the delivery, and Figure 4 is reconstructed from the published report (see the entry for the published report). A historical Figure4.xls (482 Date/Occupancy observations, 2 March 2015 to 16 March 2024) was recovered later; its link to the published figure has not been verified and it is not included in the package.

Data is publicly available and included in the reproducibility package.

Copernicus Sentinel Data Terms and Conditions

https://sentinels.copernicus.eu/documents/247904/690755/Sentinel_Data_Legal_Notice

https://dataspace.copernicus.eu

European Space Agency (ESA), Copernicus Programme. Sentinel-1 Ground Range Detected (GRD) SAR imagery [dataset], processed by the World Bank project team. Copernicus Data Space Ecosystem. https://dataspace.copernicus.eu. Access date: 2024.

Georgian Revenue Service Truck Parking Occupancy (TIR parks)

Ground-truth truck counts for Georgian TIR parks, collected by the project from the Georgian Revenue Service TIR-park service. Variables: truck counts by date and time for TIR parks 1 to 26 and 1001; park list with Customs Checkpoint, TIR Park, Capacity, Shapefile, id, scrapper_id, scrapper_subid. Files location: Figure1_WBG_R_closeout_package/data/raw/ground_truth/ ('ground truth TIR parks.xlsx', 'ground_truth_full_2min.csv', 'TIR Parks.csv'); Figure14_WBG_R_closeout_package/data/raw/ (ground_truth_TIR_parks.xlsx, TIR_Parks.csv). The data is not included in the package, so Figures 1 and 14 cannot be rerun from the public package. The live service cannot recreate the historical observations. Replicators can request the archived study extract and the site/date mapping from the authors, who will coordinate with the Georgian Revenue Service; access is not guaranteed. See data_hash_report.csv (included in the reproducibility package) for the location and SHA-256 hash of each excluded file. For more information about the data, contact the authors at ssahu8@worldbank.org.

Data access was granted directly to the study authors by the data owners/managers. It was obtained under a custom data license that does not allow for redistribution and it is not included in the reproducibility package.

https://www.rs.ge/TirPark-en?cat=1&tab=1

Georgian Revenue Service. Truck Parking Occupancy, TIR-park service [dataset]. https://www.rs.ge/TirPark-en?cat=1&tab=1. Collected by the World Bank project team, 2023-2024. Unpublished data.

PlanetScope Analytic Surface Reflectance imagery

Commercial imagery from Planet Labs PBC, and data derived from it, used for Figures 2, 12 and 13. Files location: Figure2_WBG_PR_closeout_support/ (560 *_3B_AnalyticMS_SR_harmonized_clip.tif scenes, Beitbridge); Figure2_WBG_PR_map_closeout_package and Figure2_WBG_PR_top_timeseries_closeout_package (derived detections, counts, annotations and scene manifest, including the processed files read by the Figure 2 code); Figure12_WBG_PR_closeout_package/arcgis_ppkx/; Figure13_WBG_report_correction_closeout_package/data/ (beta_v3 detector-score rasters, ground_truths and mask.png). The 42 beta_v3 composite rasters contain a single detector-score band and no reflectance bands. Raw imagery and derived files are not included, pending the applicable World Bank-Planet license terms. Figures 2, 12 and 13 therefore cannot be rerun from the public package. Replicators need their own Planet license; the scene list for Beitbridge (acquisitions from 15 December 2022 to 22 April 2023) is available from the authors. Original order IDs were not recovered. See data_hash_report.csv (included in the reproducibility package) for the location and SHA-256 hash of each excluded file. For more information about the data, contact the authors at ssahu8@worldbank.org.

Data was obtained under a commercial license that does not allow for redistribution, and it is not included in the reproducibility package.

Commercial License

https://www.planet.com/pricing/insights-platform/

https://www.planet.com

Planet Labs PBC. PlanetScope Analytic Surface Reflectance imagery, Beitbridge [dataset]. Unpublished data. https://www.planet.com. Access date: 2024.

Pléiades 1A/1B very high-resolution imagery (Airbus Defence and Space, via SkyWatch)

Very high-resolution scene of Incheon used for Figure 6 and Table 8, obtained through SkyWatch (EarthCache). Files location: Figure6_WBG_PR_closeout_VHR/SKYWATCH_PHR_PS_20231120T0215_ALL_6bff.tif (~3 GB, a mosaic of four tiles: Tile_0_0, Tile_0_1, Tile_1_0, Tile_1_1) and its sidecar files (.tif.ovr, .tif.aux.xml, .tif.xml, .tfw). Derived data, also excluded: truck points manually digitized from the scene (detections_v2, truck_detection_20240325_v1) and the Incheon building tables that use them (truck_density attribute: Incheon_train_with_feature.geojson, Incheon_test_with_feature.geojson, Incheon_warehouse_prediction), in Figure6_WBG_PR_closeout_package_v2/data/ and Table8_WBG_PR_closeout_package_CLOSED/data/. Figure 6 and Table 8 therefore cannot be rerun from the public package. No order ID was recovered. Replicators can request licensed access from SkyWatch using the scene name above. See data_hash_report.csv (included in the reproducibility package) for the location and SHA-256 hash of each excluded file. For more information about the data, contact the authors at ssahu8@worldbank.org.

Data was obtained under a commercial license that does not allow for redistribution, and it is not included in the reproducibility package.

Commercial License

https://skywatch.com/contact-sales/

https://skywatch.com/satellite-imagery/

Airbus Defence and Space. Pléiades very high-resolution imagery, Incheon [dataset]. Distributed by SkyWatch. Scene SKYWATCH_PHR_PS_20231120T0215_ALL_6bff. Access date: 2024.

Logistics Warehouse Information, Korea (MOLIT National Logistics Information Center)

Source of the warehouse_label training labels for Korea. The source file is 'Logistics warehouse information_230614.ko.en.xls' (snapshot of 14 June 2023), covering commercial warehouses larger than 1,000 square meters; 4,803 geocoded points were used for manual digitization of warehouse footprints. Files location: Table3_WBG_PR_closeout_package/data/korea_all_buildings_with_feature_updated.geojson (11,727 buildings; identical copy in Table7_WBG_PR_closeout_package/data/). These files are not included until the terms of use are confirmed, so the Korea results in Tables 3 and 7 cannot be rerun from the public package. See data_hash_report.csv (included in the reproducibility package) for the location and SHA-256 hash of each excluded file. For more information about the data, contact the authors at ssahu8@worldbank.org.

Data access was granted directly to the study authors by the data owners/managers. The terms of use for redistribution have not been confirmed, and it is not included in the reproducibility package.

Custom license

http://nlic.go.kr/nlic/WhsInfoWarehouseSch.action

Ministry of Land, Infrastructure and Transport (MOLIT), National Logistics Information Center. 2023. Logistics Warehouse Information [dataset]. http://nlic.go.kr/nlic/WhsInfoWarehouseSch.action. Snapshot of 14 June 2023.

Aktau Terminal Monthly Throughput

Monthly throughput data for a terminal of the Aktau seaport, Kazakhstan, provided to the World Bank in confidence in 2023. Files location: Figure11_Aktau_confidential_closeout_package/data/raw/AktauAfter12Months_other_CONFIDENTIAL_SOURCE.xlsx (Year, Month, Occupancy). Derived data, also excluded: Figure11_Aktau_closeout_package/data/processed/aktau_2022_public_indexed_series.csv (SAR occupancy and tonnage indices, January 2022 = 100). This file is a required input for the Figure 11 code, so Figure 11 cannot be rerun from the public package. No public access route exists; requests can be addressed to the authors. See data_hash_report.csv (included in the reproducibility package) for the location and SHA-256 hash of each excluded file. For more information about the data, contact the authors at ssahu8@worldbank.org.

Data access was granted directly to the study authors by the data owners/managers. It is confidential and it is not included in the reproducibility package.

Custom license

Aktau seaport terminal operator. 2023. Aktau Terminal Monthly Throughput [dataset]. Unpublished confidential data shared with the World Bank.

Microsoft Global ML Building Footprints

Building polygons used for attribute-based warehouse classification. Shape attributes (Shape_Leng, Shape_Area, long_ax, corners, squareness, shape_index, fractal_dimension, circular_compactness, square_compactness, elongation, convexity, ERI) were computed by the authors; OpenStreetMap, WorldPop and WSF attributes were added (see those entries). Files location: Figure7_WBG_PR_closeout_package/data/warsaw_warehouse_prediction.geojson (704 buildings); Table3_WBG_PR_closeout_package/data/DE_buildings_with_feature_updated.geojson (2,385) and lodz_buildings_w2w_with_feature.geojson (2,368); Table4_WBG_PR_closeout_package/data/{warsaw,lodz,katowice}_warehouse_prediction_bigger5000/ (shapefiles; 704, 74 and 47 records). The Korea building tables are described in the Korea warehouse labels entry. Not included: Figure8_ArcGIS_PR_closeout_package/evidence/Figure8.ppkx and its thumbnail. The ArcGIS Pro project contains four local layers derived from the Warsaw building table (2,470 building polygons, their centroids, DBSCAN clusters and a kernel-density raster) and references Esri basemaps (OpenStreetMap vector tiles, World Topographic Map, World Hillshade). Figure 8 was produced manually in ArcGIS Pro.

Data is publicly available and included in the reproducibility package.

Community Data License Agreement – Permissive – Version 2.0

https://cdla.dev/permissive-2-0/

https://github.com/microsoft/GlobalMLBuildingFootprints

Microsoft. Global ML Building Footprints [dataset]. https://github.com/microsoft/GlobalMLBuildingFootprints. Access date: 2024.

OpenStreetMap

Used to compute building attributes: highway_dis, closest_highway_length, highway_clossness_500, railway_dis and industrial_landuse. The building tables also contain airport_dis and seaport_dis, whose source is not documented in the package. Files location: attributes in the building tables listed under the Microsoft Global ML Building Footprints entry (Figure 7, Tables 3 and 4) and the Korea warehouse labels entry. Table5_OSM_datatypes_NA_closeout_package/data/table5_osm_datatypes_source.csv lists OSM data types only (no OSM extracts).

Data is publicly available and included in the reproducibility package.

Open Data Commons Open Database License (ODbL 1.0)

https://www.openstreetmap.org/copyright

https://www.openstreetmap.org

OpenStreetMap contributors. OpenStreetMap [dataset]. https://www.openstreetmap.org. Access date: 2024.

WorldPop population density (1 km, 2020)

Files location: Figure7_WBG_PR_closeout_package/data/warsaw_pd_2020_1km.tif (identical to the Table 4 copy); Table4_WBG_PR_closeout_package/data/ (warsaw_pd_2020_1km.tif, lodz_pd_2020_1km.tif, katowice_pd_2020_1km.tif, pol_pd_2020_1km.tif). Also the worldpop attribute in the building tables listed under the Microsoft Global ML Building Footprints entry.

Data is publicly available and included in the reproducibility package.

Creative Commons Attribution 4.0 International (CC BY 4.0)

https://creativecommons.org/licenses/by/4.0/

https://www.worldpop.org

WorldPop (School of Geography and Environmental Science, University of Southampton). Population density 2020, 1 km resolution [dataset]. https://www.worldpop.org. Access date: 2024.

World Settlement Footprint (WSF) Evolution and WSF3D (German Aerospace Center, DLR)

Files location: Figure7_WBG_PR_closeout_package/data/warsaw_WSF_EVO_3857.tif (clipped and reprojected WSF Evolution). WSF3D building height is in the wsf3d attribute of the building tables listed under the Microsoft Global ML Building Footprints entry; WSF Evolution is in the wsf_evo attribute of the Warsaw and Katowice building tables. In Figure 7, buildings are grouped by WSF settlement-appearance year; the groups describe currently mapped buildings, not historical warehouse inventories.

Data is publicly available and included in the reproducibility package.

Creative Commons Attribution 4.0 International (CC BY 4.0)

https://creativecommons.org/licenses/by/4.0/

https://geoservice.dlr.de/web/maps/eoc:wsfevolution

German Aerospace Center (DLR). World Settlement Footprint (WSF) Evolution and WSF3D [dataset]. https://geoservice.dlr.de. Access date: 2024.

U.S. Surface Transportation Board (STB) Rail Service Data

File locations: Figure10_Chicago_closeout_package/data/raw/chicago_clearing_monthly_source_extract.xlsx (variables: Date; Satellite detector: number of railcars; Ground truth: Number of railcars; Error) and data/processed/chicago_clearing_monthly_series.csv (month, sar_estimated_railcars, ground_truth_railcars, relative_error_sar_denominator). Ground truth is the STB Chicago Clearing yard inventory; the satellite series is the project's SAR detector output.

Data is publicly available and included in the reproducibility package.

Public Domain Information

https://www.stb.gov/reports-data/rail-service-data/

U.S. Surface Transportation Board. Rail Service Data: Chicago Clearing seven-day average yard inventory [dataset]. https://www.stb.gov/reports-data/rail-service-data/. Access date: 2024.

Reimagining Logistics and Connectivity from Space

Values transcribed or digitized from the published report; no third-party data. Files location: Figure4_PR_reverse_engineered_closeout_package/reference/ (3 report-page crops; the code regenerates data/figure4_digitized_from_report_page26.csv from them); Figure5_attribute_classification_NA_closeout_package/data/figure5_{nodes,edges,stages}.csv; Figure15_OHR_pipeline_closeout_package/data/figure15_pipeline_steps.csv; Figure16_DA_NA_final_nooverlap_closeout_package/data/figure16_{nodes,edges,panels}.csv; Table1_EO_AI_applications_NA_closeout_package/data/table1_applications_source.csv; Table2_trade_node_indicators_NA_closeout_package/data/table2_indicators_source.csv; Table5_OSM_datatypes_NA_closeout_package/data/table5_osm_datatypes_source.csv; Figure13_WBG_report_correction_closeout_package/reference/figure13_published_transcription.csv. Figure4_PR_reverse_engineered_closeout_package/data/figure4_laxmannath_candidate_aoi_not_original.geojson is a reconstructed area of interest, not the original.

Data is publicly available and included in the reproducibility package.

Creative Commons Attribution 3.0 IGO (CC BY 3.0 IGO)

https://creativecommons.org/licenses/by/3.0/igo/

Sahu, Saslavsky, and Schultz. 2024. Reimagining Logistics and Connectivity from Space. Washington, DC: World Bank.