Building an Earth Observations Data Cube with Remote Sensing
Mr CHEN Shuang
As one of the most efficient tools for surveying the Earth surface, satellite remote sensing has provided major advances in understanding the Earth system and its dynamics. With the recent technological development, the number of operating satellites and the available remote sensing data are undergoing an explosive growth.
However, there’s still a lack of the ability to efficiently manage, assimilate and utilize such Big data. To help address these challenges, our research aims at building a global daily seamless data cube (SDC) with the latest technologies such as high performance computing, machine learning and virtual constellations. The SDC is a new paradigm aiming to realize the full potential of EO data by reducing the barriers caused by these Big data challenges and providing access to large spatio-temporal data in an analysis-ready form.
The production process of the SDC mainly consists of 4 parts: data correction, missing data reconstruction, data fusion, and standardized data reorganization. The data get systematically processed and reorganized in an analysis-ready form to allow analysis with a minimum of additional user effort. By integrating theory and methods from traditional remote sensing with artificial intelligence, we attempt to enhance the generalizability and flexibility of the reconstruction process. To find an efficient and economical way to store such huge amounts of data, we propose a novel compression approach for Earth Observations data via compressive sensing, which achieves a compression ratio >10 with a sufficiently small reconstruction error.