Data has become one of the most valuable resources of the 21st century. Indeed, data can be considered the most important input when it comes to make informed decisions. The recent global pandemic crisis highlighted the vital role of data for reporting accurate case numbers and outbreaks, identifying the most vulnerable demographics, and understanding the most effective vaccines, to mention few. Data also plays a key role when it comes to sustainability. Data help business to make decisions that guide measurable and responsible practices; they give us the opportunity to better understand our demand for food, energy, water and land as the world population increases and changes in climate affects these resources; data can also support real-time monitoring of the environment, help consumers adopt more sustainable behavior, and create sustainable value chains.

A specific type of data that has gained importance and popularity among different user groups is satellite-based data. Satellite-based Earth Observation relies on the use of remote sensing techniques for collecting information on the physical, chemical, and biological systems of the planet Earth to monitor land, marine habitats as well as freshwater bodies such as (rivers, lakes, as well as groundwater) and atmosphere (EUSPA, 2021). EO data has become an invaluable resource for supporting decision-makers in all sectors of society, including responding to the UN Sustainable Development Goals (SDGs) and the Convention on Wetlands (Ramsar, 1971).

Technological advances and significant reductions in the costs of operation and maintenance of satellites together with free and open data policy has allowed the generation of a large amount of geospatial data that are freely available for a wide range of users - such as researchers, small businesses, NGOs, and society - to benefit from EO data (Christopherson et al., 2019). Every day, several exabytes of free and open geospatial data are generated, which means that everyday a huge amount of data from different sources, of different scales, with different variable types, and spatial and temporal resolutions must be collected, characterized, analyzed, stored, and disseminated (Guo et al., 2015; Gomes et al., 2020). For instance, Landsat-7 and Landsat-8, Sentinel-1, -2 and -3, and MODIS (Terra and Aqua units) produced about 5 petabytes of open data in 2019 (Soille et al., 2018). These big data sets often exceed the memory, storage and processing capacities of personal computers which limits users to take full advantage of the amount of data that is available for use (Gomes et al., 2020).

To overcome those challenges, new approaches and solutions have been developed to properly store, process, disseminate and analyze satellite imagery. In this context, Earth Observation Data Cubes stand out for storing and transferring large quantities of Earth observation imagery, facilitating data organization and management, and allowing pre-processing and fast access to summarized data (Kopp et al., 2019), (Copernicus, 2021). The world’s first EO data cube, The Australian Geoscience Data Cube (AGDC), was developed in 1979 in the context of the countries’ first participation in the Landsat program (Lewis et al, 2017). Several other initiatives and projects have been deployed since then to support national and regional policies in decision-making (e.g.: Swiss Data Cube, DE Africa) and to foster the use and impact of satellite data (e.g.: Open Data Cube).

What are Data Cubes?

A data cube is a multidimensional data infrastructure used to store data. It represents the data in terms of dimensions and measures: each dimension corresponds to an attribute or a set of attributes, and each cell stores the value of some aggregate measure such as count or sum (Han et al., 2012). Figure 2 presents the diagram of a general data cube.

The data cube in Figure 2 stores sales information on a multinational corporation, such as sales in all time, sales at all branches, sales at all locations, sales of all items. Besides storing data, users can also manipulate a cube to select specific data for the purpose of analysis or to visualize data from a different perspective. Figure 3 and Table 1 presents the 5 main data cube operations using a simple example use case for ease of understanding.

When it comes to geospatial data, data cubes can store a large amount of satellite data and are specialized on a region or a thematic application over a specified period. Earth Observation data cubes (EODC) store data that has been processed to serve a specific set of requirements and organized into a form that allows immediate analysis with a small additional user effort and interoperability both, through time and with other datasets (CEOS, n.d.). Earth Observation data pre-processed for users and ‘ready to use’ is called Analysis Ready Data (ARD). General data pre-processing includes atmospheric correction, cloud masking, orthorectification of images and re-gridding (EUMETSAT, 2022).

The two main dimensions of a EODC are latitude and longitude; other typical dimensions include: time, depth and altitude. Over a specific area, images are mosaicked together and then these mosaics are stacked on top of each other (Stefan Brand, 2021), see Figure 4. This way, the user receives a continuous, gridded data structure with at least three dimensions (latitude, longitude and time).

A point-in-time query for an EO data cube, or a slicing operation, returns an image as shown in Figure 5 (image selection). This operation can be used to compare attributes of different regions at the same period. In turn, a query for a cube at a particular location, or a rolling-up operation, returns a time series of values at that location (Figure 5 – time series selection). The cube structure simplifies data aggregation operations such as weekly, monthly, and annual analysis (Koop et. al., 2019). Dicing and drill-down operations can be useful to analyze precipitation data on basins and the many levels of subbasins, for instance. Overall, users can inspect changes to any area over any timespan covered by the ingested datasets. (Open-source) algorithms are then applied, allowing for analysis of particular data – including on vegetation, land use, water coverage and quality, and urban expansion (ITU Hub, 2020).

Several countries have been implementing EO data cubes to support national policy and decision-making. Data cubes are also being deployed at regional scales, which allows addressing transboundary issues relevant to regional policy and planning needs (Dhu et al, 2019). Examples of regional and national open data cubes (ODC) already implemented and/or under development are shown in Table 2.

Earth Observation data cube applications for water resource management

Some of the most requested Earth observation datasets are for water resource management applications (ITU Hub, 2020). Data Cubes can help solve water issues including monitoring ecosystem degradation over vast areas and long-time periods (by using time series) with data including water extent, water detection, water quality, coastal change and water detection using radar. Table 3 presents the main applications available to users to date.

Conclusion

A significant growth in open and free land imagery data has significantly increased data volumes in the past years. Data cubes emerge as a valuable approach to expand the use of satellite data. They allow the storage of large amount of pre-processed data in a simple way, reduce data preparation burden, increase data interoperability, provide a summarized view of data as well as quick and better analysis. Overall, cubes help improving the performance of data analysis, which increases the user’s capacity to explore the multiple benefits of Earth observation data. However, lack of knowledge and skills, infrastructure, and resources to access and explore the potential of the available space-based data remains a challenge for most developing countries. Therefore, together with the development of open data cubes there should also exist an effort to develop training material along with the products, services, and databases to enable users to develop the skills needed to use them. For instance, DE Africa offers an online training to help users work autonomously on the platform.

When it comes to water resources management, several products and resources (maps, mosaics, models, algorithms, notebooks) with important applications have been developed. However, few regions in the world are already benefiting from those products. It is expected that the further development of open data cubes will contribute to close this gap.