**1. Introduction**

Rapid urbanization creates a strong need to optimize land use in densely populated cities. Attention is thus shifting from the very limited available space above ground to generation and increased use of underground spaces. Comparing to the above ground, underground is an unseen space. The trench for the building and maintenance of underground infrastructure costs a lot of money, as well as faces high risks. A prerequisite for including the underground in urban planning is the availability of sufficiently complete, accurate and up-to-date 3D maps of the underground. However, such maps are not yet widely available, if at all, and the required data acquisition is much more challenging than for spaces above ground.

With a population of more than five million living in an area of 720 square kilometres, Singapore has revealed a plan for placing infrastructure underground [1]. Currently, a data sharing platform, which is called GeoSpace, is maintained by the Singapore Land Authority (SLA) and used by government agencies (e.g., utility owners, land developers, and land owners) to establish a 2D map of Singapore underground including utility services. Figure 1 shows an example in the Marina Bay region of Singapore includes water supply, sewage, drainage, telecommunication and power grid networks. All the existing data are the 2D format. The 2D visualization with overlap pipelines has limitation to provide accurate and reliable information about underground utilities to various applications.

**Figure 1.** An example of utility data in the Marina Bay region of Singapore (source: Singapore Land Authority, 2018).

To observe the existing data, we zoom in to a corner of the Marina Bay region. Figure 2a presents five layers of different power grid networks. In the real world, the five different power grid networks may be located at the same place and different depths. However, these data have the same *x*, *y* value in the database, which makes them impossible to identify in the vertical space and distinguish them in 2D. All of the existing data are as-build data. We can not trust them to present the real situation of underground utility networks. From Figure 2b, the limited attributes are provided from the current database. Only the main water pipes have a diameter. Most of them have 2D geospatial information. In addition, data owners have more details of existing utility data. However, most of them are 2D data as well. Depending on the requirement of the application, some data owners try to collect 3D data. There are some issues during the data capture to usage. Without the utility survey standard, some of them only use the traditional survey method to get the 3D points data of pipelines and overlay on the existing data. Nobody can guarantee the quality of these data. Meanwhile, because of the limitation of the existing data model, it is difficult to integrate 3D data with the existing 2D data. Update cycles were observed to be infrequent and slow, which is once per six months. We not only need time information in the data model to maintain utility database frequently, but also should improve data governance procedures for updating. In general, some issues prevent these data from being sufficient for urban

planning, land administration, and on-site work. In fact, many existing databases, not only the ones in Singapore, contribute incompletely to the spatial understanding of the underground because of similar restrictions. In particular:

(**a**) An example of power grid data


(**b**) The attributes of existing utility data

**Figure 2.** The issues of existing utility data (Source: Singapore Land Authority, 2018).


Overall, the reliable and accurate 3D data of utility networks are sorely demanded. Therefore, the Singapore-ETH Centre together with the SLA and the Geomatics Department of the City of Zürich have started a related project under the name "Digital Underground" [2]. The initial goals of this project are to develop a road map, a data model and a concept for deriving a unified and complete 3D map of the relevant underground structures (in particular of utilities and spaces like corridors or tunnels). Collecting best practices for underground utility mapping is a special focus within the project. Figure 3 describes the workflow of data governance for 3D underground utility mapping. In the

data capture, different types of survey techniques (e.g., Ground Penetrating Radar (GPR), Gyro-based system) are explored and compared to find the optimal underground utility survey approach. After the data processing, the newly collected data should be integrated into the existing database aiming to improve the information of underground utility. As the backbone of the 3D underground utility map, the 3D consolidated database of underground utilities should be developed for data storage. This is a loop workflow. The data capture could improve and update the database. At the same time, the underground utility database should provide information to support data capture. In order to organize these four steps, we need two main components in the data governance. One is the framework to manage different roles and communication between them in data governance. The other is the underground utility data model, which is a conceptual model to describe the structure and content of geodata independent from the used hard- and software systems. It will provide the standard for the presentation of geometrical information, data quality management and various applications. This paper focuses on the design of the framework of data governance and underground utility data model. To ensure legal compliance, efficiency, and resilience of these utility networks, the reliable 3D underground utility data could shed light on their ownership and operation [3]. Then, the underground utility data can be used in various applications. To provide sufficiently and consistently accurate information about underground utilities, it is necessary to fill the gap between engineering practices and mapping disciplines. Meanwhile, we need to find the solution for how to use the existing data and integrate it with newly collected data.

**Figure 3.** Workflow of underground utility mapping.

Here, we focus on underground utilities, ignoring other underground structures that eventually need to be represented in the same 3D database as the utilities. This work aims at bridging the gap between underground utility surveying and data governance for land administration. Our proposal addresses the following:


Subsequently, we first introduce related works on 3D underground utility data acquisition and review the underground utility data governance for land administration in some countries or regions. In Section 3, we propose a framework to resolve the above issues about data governance and explain the design of a 3D underground utility data model. In Section 4, we briefly summarize a Singapore case study covering the work process from large scale GPR-based data acquisition to 3D visualization. We conclude with a summary and an outlook on future work.
