Toshikazu Seto

This paper aims to clarify the state of development of highly accurate and open 3D city model data and its usage methods, which started in Japan in 2020, from two aspects: quantitative geospatial analysis using publicly available data, and qualitative evaluation analysis of 40 use cases using the data. As a background to this study, digital twins, which are virtual replicas of the physical urban built environment (Shahat et al., 2021), are gaining global attention with the development of geospatial information technology to understand current conditions and plan future scenarios in cities (Lei et al., 2022). This trend can be applied in areas related to a wide range of urban issues, such as urban development, disaster prevention, and environmental and energy simulation, and has the potential to be used for urban planning through an intuitive approach via various GIS tools. On the other hand, the geospatial information required by the digital twin also needs to be accompanied by three-dimensional shape information and many attribute information of building units. Data development and related research using CityGML (Kolbe et al., 2021), a representative standard specification, has mostly been carried out in European and US cities, and there have been few efforts in Asia (https://github.com/OloOcki/awesome-citygml). In Japan, urban planning has mainly been carried out using analogue methods such as paper maps and window services. However, as citizens' lifestyles and socio-economic systems are drastically changing due to the high interest in smart cities and the spread of COVID-19 infection, urban policies such as disaster prevention and urban development using digital technology are becoming increasingly important. The digital transformation of urban policies such as disaster prevention and urban development using digital technology has become an urgent issue. “Project PLATEAU (https://www.mlit.go.jp/plateau/)” is a project initiated in 2020 under the leadership of the Ministry of Land, Infrastructure, Transport and Tourism (MLIT) to develop a high-precision 1:2500-level 3D city model CityGML format in a unified manner and convert it into open data format via CKAN's data portal (CityGML, 3D Tiles, GeoJSON, MVT, ESRI Shapefile), to develop an open-source data viewer and to explore use cases. This study details the history of the "Project PLATEAU" initiative and discusses the relationship between openness and urban data commons. Many of the data specifications, converters and online viewers are closely related to FOSS4G. Next, data for 126 cities in Japan (about 19,000 square kilometers) developed as open data over a three-year period are regionally aggregated and then quantitatively compared with OSM building data in Japan. Trends such as coverage rates between cities and micro-regional analysis within Tokyo are then attempted. To analyze a large amount of data, this part was carried out using data converted to FlatGeobuf format. Some of the results of the data preparation analysis are as follows: The basic analysis of the cities covered by PLATEAU showed that the total number of buildings in LOD1 was about 15.7 million, with a population coverage of about 38.4%. These cities have shown an increasing trend in population over the last five years (an average of about +10,000 for the 126 cities). By comparison, the total number of OSM buildings in the country is about 12.7 million, generally widely distributed across the country's 1903 administrative districts (about 38,000 square kilometers). Therefore, only the cities maintained by PLATEAU provide data with a higher level of detail than OSM. However, the detailed LOD2 building data with roof shape is limited to about 480,000 buildings (about 300 square kilometers in 97 cities nationwide), which are high-rise buildings and landmarks in large cities. To identify more micro trends, we compared the accuracy of the building data for central Tokyo, which has the largest number of units in both datasets, in 2020, the year the PLATEAU data was created. The number of units in each building dataset is OSM (726,685 units) and PLATEAU (1,768,868 units). When PLATEAU is used as the base data, the coverage of OSM is about 40%. On the other hand, of the 3190 city blocks in central Tokyo, 502 (about 15.7%) were identified as having more OSM buildings than PLATEAU. As a factor contributing to this discrepancy, a historical analysis of the timestamps and versions of the building data (about 80,000 units) that exist only in OSM revealed that most of them were created more than two years before the PLATEAU data and have never been updated. Therefore, the PLATEAU data should be updated to keep the data fresh, even in areas where OSM data are already widely distributed, if only data older than 2020 are maintained. For example, open 3D urban model data for cities of various sizes have been released in Japan, and they are highly accurate and complementary to OSM data. In addition, these data have begun to be used in administrative practice, and a total of 44 applications in new areas such as citizen participation and entertainment (especially services using XR) have been identified. The evaluation of the exploitation methods is explained in the paper, but the cases related to smart cities and disaster prevention are particularly striking. The issues to be addressed in these efforts are the nationalization of the scope of maintenance, the organic merging with open data as represented by OSM, and the further GIS education in the field of urban planning. Finally, as data contributing to the reproducibility of this study, the data sources used in the analysis are themselves open data and thus readily available. Therefore, we plan to provide a download list of each data source and GIS data summarizing the tabulation results as open data on Github.

None

In this research, the OSM-Notes feature is mainly viewed as data that can be examined speedily from OSM data globally in terms of the content of the notes posted and the location of users.

Data from volunteered geographic information (VGI) and public surveying are essentially different; while a great quality and quantity of data is available for VGIs, discussions about data quality are scarce (Senaratne et al., 2017). As the use of VGI in business expands, attempts have been made to develop quality verification tools (cf. KeepRight and Validation OSM) with the understanding that, in addition to position accuracy, data quality equally depends upon the diversity and interaction of the number of users involved in data generation (Haklay et al, 2010). OSM-Notes is a capability for describing errors and discussing OSM data and fixme, although there are few research cases (Seto et al., 2017). This feature is new and was only added to OSM.org in April 2013. It allows users to specify and comment on any point on the map, and the history of comments can be accumulated and closed when the problem is solved. Unlike the fixme tag (allows contributors to mark objects and places that need further attention in the form of a "note to self" or request for additional mapping resources), this function does not directly associate with OSM data, but there is no need to have an OSM account. In this research, the OSM-Notes feature is mainly viewed as data that can be examined speedily from OSM data globally in terms of the content of the notes posted and the location of users. The purpose of the research is to analyze the context of OSM-Notes use through a “GIS” approach (Cope & Elwood, 2009). This analysis was conducted on data dumped from Planet OSM (https://planet.openstreetmap.org/) as April 2019. Since planet-notes.osn (about 782MB) is a special binary format, we used an enhanced parser able to separately output open (unresolved discussions) and closed (resolved discussions) based on osn2osm. After converting to .osm format data, the set was combined with Natural Earth's border data and population data and treated as spatial data. As a result, there were OSM-Notes postings in 237 countries: 415,433 open and 129,887 closed records. By counting the number of OSM-Notes postings by country based on whether they are open or closed, we determined that the majority were in the United States, Germany, and Russia. Moreover, it became clear that Japan, Canada, Korea, and Taiwan are the regions where there are a large number of submissions in urban areas (based on Natural Earth’s definition). In addition to these, many OSM users have posted from many countries, including Iraq, Ukraine, and Ecuador, and it is clear that active discussions are being held by many contributors. In these countries, mapping for humanitarian assistance is commonplace, as other reference resources are unavailable, so it is necessary to improve data quality through the use of OSM-Notes. OSM-Notes even has a function that allows non-OSM users to post, as is the case with 50% or more of the posts in Spain, Korea, etc. This data can be used to analyze urban trends and spatial features within a single country. For example, according to the analysis for Japan, OSM-Notes has many posts about the location of shops and POI (Point of Interest), and suggestions based on Maps.ME, Facebook, and Pokémon go. This is considered to be the main reason that anonymous posting is permitted. It is also worth noting that very few users post only a single note. Thus, by analyzing OSM-Notes, it is possible to grasp hot issues between users of OSM data. Overall, OSM mapping was found to be more common in developed countries with active mapping, but similarly, bug reports from non-OSM applications as developing countries that require discussion among OSM users, and as a new aspect. Moreover, how to accept it is an important issue in considering the process of constructing OSM data. However, because the functional relationship between OSM-Notes and the data on OSM cannot be specified directly, the nature of the feature is also difficult to grasp directly compared to the fixme tag.