Ganguly & Colleagues Frame the Networked Digital Earth Concept Through the First International Conference on the Topic

The first International Conference on the Networked Digital Earth (ICNDE 2018), jointly organized by Northeastern University, Boston, MA, and the Indian Institute of Technology (IIT) Kharagpur, India, was held during March 7-9, 2018, at Kharagpur, India. Twenty-five invited speakers and four invited panelists were drawn from the United States and India, as well as Singapore, Europe, and Indonesia, and included representatives from academia, research laboratories, private industry, government agencies, and world bodies. The attendees included about one hundred and twenty interdisciplinary undergraduate and graduate students from nationally top-ranked institutes in India, as well as graduate students from Northeastern. Opening and mid-conference speeches were delivered by the Deputy Director and the Director of IIT Kharagpur, while representatives from the Indian Ministry of Water Resources and the Prime Minister's office in India, as well as the World Bank, shared perspectives. The conference covered general principles and theory of the Networked Digital Earth concept, with a focus on network science, geospatial data sciences, machine learning and AI, while emphasizing the domains of water sustainability and climate change, urban science and engineering, public health and ecology, as well as cyber and physical security and the social media. The attendees from Northeastern Civil and Environmental Engineering (CEE) included PhD students and ICNDE 2018 student organizers Shashank Konduri and Udit Bhatia, as well as CEE Professor and ICDNE 2018 organizer Auroop Ganguly. The attendees from Northeastern College of Computer and Information Science included PhD student and ICNDE 2018 student organizer Tanay Mehta as well as Professor and ICNDE 2018 organizer Ravi Sundaram. The third primary ICNDE 2018 organizer was Professor Niloy Ganguly from the Computer Science and Engineering department of IIT, Kharagpur, who also spearheaded the local organizing committee comprising interdisciplinary faculty, staff, and students at IIT. The conference was funded by both the organizing institutes, specifically, Northeastern University and IIT Kharagpur, as well as by Microsoft India. From Northeastern University, the bulk of the funding was provided by the Civil and Environmental Engineering department followed by the Tier 1 Seed Funds of the Office of the Provost. The local host institute happens to be the oldest of the Indian Institutes of Technology, and the one with the largest and most interdisciplinary campus. 

The Networked Digital Earth concept, originally proposed by CEE professor Auroop Ganguly in 2017 with interdisciplinary colleagues at Northeastern University, attempts to bring together two interrelated areas: The Next-generation Digital Earth, and the concept of Globally Networked Risks. The Digital Earth, originally envisioned by former US Vice President Al Gore in 1998, is a virtual representation of our planet, which enables scientists and ordinary citizens to easily browse through the possible histories, current state, and the plausible futures, of the Earth. The advent of virtual globes such as the Google Earth, as well as geoportals and geographic data libraries, may have taken steps in that direction. However, a 2012 perspective article, entitled "Next-generation Digital Earth", in the Proceedings of the National Academy of Sciences (PNAS) by Michael Goodchild and colleagues pointed to three unrealized goals of current GIS systems vis-a-vis the original Digital Earth vision. First, there are not much historical or forward simulation capabilities. Second, not all data are treated equal owing to restrictive conditions or ease of use limitations. Third, challenges persist in the communication of non-visual information such as spatial variability of demographics or biodiversity, and in the depiction of uncertainties. The PNAS article goes on to develop a framework for the next-generation digital earth. A perspective article by Dirk Helbing, published in the journal Nature, was entitled: "Globally networked risks and how to respond". Engineering systems, human or social institutions, and natural systems, have become deeply interconnected from community and urban to regional and global scales. While the interconnections provide mutual support and strength, these can also result in cascading failures and risks. The ability to leverage the strengths derived from systemic interdependence, while reducing the possibility of massive interconnected failures, is a crucial requirement in our networked world. Graphical methods, including but not limited to network optimization and complex network science (the latter pioneered among others by Albert-Laszlo Barabasi of Northeastern), offer one set of approaches to deal with these networked risks. Such methods attempt to develop theories and solutions for an interconnected world, with applications to engineered systems such as the electric grid, natural systems such as ecology, and human systems such as urban communities, regions and institutions, as well as the social media. Even the earth's climate system has been described in part via complex networks (e.g., see Climate as Complex Networks), while such have been also used for engineered systems (e.g., see the recent textbook: Critical Infrastructures Resilience). The World Economic Forum in their 2018 Global Risks Report discusses interconnected risks and risk-trends in terms of global networks. One illustration of the challenge inherent in the Networked Digital Earth concept may be described in the context of Machine Learning (which in turn is the form of Artificial Intelligence where machines learn from data). The ability to deal with geospatial and temporal data, such as those in a digital earth, requires machine learning approaches to be cognizant of what may be called proximity-based dependence structures. Geographers know this as the first law of geography, while time series and spatial statisticians know they need to deal with temporal or spatial autocorrelation. Thus, textbooks and articles have been written on areas like Geographic Data Mining. However, in a globally networked world, the proximity may instead be based on network distance, rather than geographic distance alone, leading to additional complexity. Thus, climate or hydrologic patterns can have long-range dependence and long-memory, while communities formed from cell phone calls may be based on human relations where physical proximity is not the sole or even the most important determinant. Interdependent natural, engineered, and human systems on the planet Earth may also exhibit complexities owing to nonlinear dynamics (including extreme sensitivity to initial conditions or chaos, i.e., the so-called butterfly effect), low frequency variability (including so-called pink noise), and cascading uncertainties and failure propagation, all of which could potentially cause massive disruptions with disproportionate impacts. The ability to characterize, prepare for, and ideally mitigate, such disruptive events is a desired goal of the Networked Digital Earth, and would require physics-data-human based models and decision support systems. Real world cases may be developed in societal priorities, such as the global risks highlighted by the World Economic Forum: extreme weather and natural hazards, water crises and climate change, urban engineering and infrastructures, as well as public health and security, among many others. The first ICNDE conference has helped crystallize many of these themes, including the general principles and a few case studies. The immediate next step is to develop an edited volume covering key aspects of the Networked Digital Earth, which is currently being discussed with a publisher.

Related Faculty: Auroop R. Ganguly

Related Departments:Civil & Environmental Engineering