The confluence of large volumes of readily accessible, high-resolution satellite observations, increasingly sophisticated software and continued growth in the capability of high-performance computing (HPC) now enable the production of high-demand, timely, high-quality, global-scale, geospatial products. Recent, highly successful partnerships have produced transformative geospatial products including the ArcticDEM, Reference Elevation Model of Antarctica (REMA), and the NASA Goddard Sahil Treecount Project, demonstrate the high return on investment achieved from providing collaborative teams of geospatial scientists and software engineers with access to large volumes of high resolution, high-frequency data, and efficient high-performance computing and services. Going forward, these projects offer the potential to deliver transformative new capabilities in the realms of artificial intelligence and machine learning, as well as efficient traditional methods, to dramatically improve the automated extraction of information from imagery over large areas and at high resolution and through time. Potentially, this will open an entirely new capability for a high-resolution understanding of a changing earth in many dimensions.

Geospatial data science represents an emerging interdisciplinary and transdisciplinary field intersecting among three broad knowledge domains: geospatial sciences and technologies, mathematical and statistical sciences, and cyberinfrastructure and computational sciences. The core of this intersection encompasses the synergies and interactions between AI, big data and cyberGIS (that is, geographic information science and systems based on advanced cyberinfrastructure) with geospatial principles guiding discovery and innovation. This presentation explores the frontiers of geospatial data science in the context of data-driven discovery and innovation. Several fundamental research problems are addressed while a set of challenges and opportunities are identified for synergistically advancing geospatial discovery and innovation. The presentation discusses how such advances are transforming data science education and research.

Data science already exerts considerable influence over our everyday lives. As we move into the age of autonomous vehicles, smart cities, personalized medicine, and artificial intelligence, both data-driven algorithms and dynamical feedback loops will permeate every aspect of our lives. Therefore, there is a pressing need for a synthesis of data science and dynamical systems, including a clear-eyed assessment of both positive impacts and limitations. The research performed in iDS2 aims to lay the foundations for this synthesis. This overview talk will introduce the four research themes of the institute (Data modeling; Sampling and inference; Algorithm design; Decision-Making) and present the institute’s vision and ongoing projects that integrate ideas and techniques from control, optimization, statistics, and computer science to realize this vision.