Nagaraja Rao Harshadeep
Title: Visualizing and Accessing the Digital Earth at your Fingertips
Decisions related to planning and operations made every day around the world need to be supported by the best data, information, and knowledge available. It seems like there is a new digital mousetrap invented every day to provide a new way to visualize some aspect of the world – from precipitation and landcover to poverty and trade. How do we go from the retail-level of “come to my portal and see my piece of the world” to a better way to leverage the tremendous advances in understanding each sector and emerging disruptive technologies including in-situ sensors, earth observation, big data, cloud computing, connectivity, AI, and interactive visualization to access this rapidly-changing digital world? This presentation will present an effort from the World Bank – the Spatial Agent App – to help showcase how this digital world can be made more accessible to all in a more ‘wholesale’ manner – bringing a world of global, regional, and local data and online computing to everyone’s fingertips. New ways of packaging digital earth data and knowledge using interactive e-books will also be showcased with a discussion of how you can partner to showcase your work.
Auroop R. Ganguly
Title: The Networked Digital Earth for Climate Change Impacts on Coupled Natural-Built-Human Systems
The earth’s climate system has been described through complex networks. The premise of describing climate as complex networks is that once networks are constructed from climate data the topology of these networks may yield new scientific understanding of fundamental climate drivers as well as novel predictive insights. The latter may be especially relevant for modeling phenomena like teleconnections which may be difficult to capture well through the current generation of models. Networks and machine learning, as well as other data-driven tools such as specialized statistical methods for extreme values and for nonlinear dynamics, can work in tandem with physical understanding and physics-based numerical models, to produce insights on climate change including the statistics of weather extremes and other climate induced stresses such as regional warming or water scarcity. The climate system in turn impacts the coupled natural-built infrastructure systems and defenses. Thus, riverine or coastal ecosystems, may offer natural buffers against floods and storm surges, while built systems such as dams, reservoirs, and levees, offer protection against floods or droughts. Vulnerable ecosystems including large food webs may lose significant biodiversity including threatened species. Cities and urbanized megalopolises may lose access to lifelines such as power, communications, transport and water or wastewater. The water-food-energy nexus as well as hazards and humanitarian logistics chains may get disrupted across communities and regions. Food networks in ecology and lifeline infrastructure networks such as the system-of-systems formed by communications-power-transport-water have been described through complex networks, as well as network optimization and information theoretic approaches. Data science and machine learning, as well as process models and network approaches, have been used to describe aspects of these coupled natural-built systems. The final impacted systems are micro and macro level human communities, interconnected urban areas and megalopolises, as well as human institutions and assets. Social networks and communities, as well as interconnected cities and institutions, may also be described through network sciences, combined with data science and process models such as agent-based systems. In addition to cascading flow of processes and information from climate or weather to natural-built infrastructures to human institutions and social networks, multiple feedback processes exist between the impacted social, informational, infrastructural and natural systems, as well as to the climate and weather or hydrologic systems. Globalization of economies as well as global environmental change imply that the systems are interconnected, and extremes or stresses percolate within and cascade across systems. The Networked Digital Earth concept may enable a comprehensive characterization of these deeply coupled natural-built-human systems. Multiscale networks or network-of-networks, spanning from local to global scales, may form a backbone of these system-of-systems. Specialized tools for predictive insights on trends and patterns, rare events and abrupt change, as well as cascading processes and uncertainly propagation may be necessary. Human-physics-data driven interfaces, including those based on recent developments in Artificial Intelligence, may enable a data to information to decisions approach. Proof of concept approaches from the literature and best practices, as well as emerging research, point to a future with new possibilities.
Title: Sensor Networks – Applications to Urban Floods
Sensors for monitoring the variables responsible for flooding in cities are helping to provide real-time information to the public in general and city managers in particular for reducing problems like traffic congestion, economic losses, and other inconveniences like reduced mobility. A similar attempt is being tried for Kolkata, the most important city in the eastern region of India, with the support of the city municipal corporation. This city, formerly known as Calcutta and which had once been the capital of the country under colonial dominion, has a combined drainage and sewerage network implemented initially around a century and a half back. However, with increasing urbanization, this system has been under stress in evacuating storm water completely, resulting in back-flow and consequent water logging of the streets. Sensors have been installed to monitor this condition and disseminate the information to the public through electronic billboards located at strategic point within the city. This presentation discusses some salient points about the project and how it is being used in managing the problem.
Title: MANAGEMENT OF THE TRADITIONAL IRRIGATION SYSTEM TOWARD THE CLIMATE CHANGE: Case of Rice farming development in Bali Province-Indonesia
Indonesia is one of the vulnerable countries to climate change in the relation to rice farming. This might influence to rice production and food security in the country. The availability of irrigation water is significantly very important for rice productivity as a main staple food for the people in Asia. Irrigation management should be technically and socially needed to control rice field cultivation along the year. In case of Bali province in Indonesia, the irrigation management along the river and at the farming level is run by government and local farmers groups called subak as a traditional irrigation system. The objectives of this study are to describe traditional irrigation system in the irrigation and rice farming development, and to explain the mitigation efforts conducted by subak to solve the irrigation water problem.
The study pointed out that subak manages irrigation under the local wisdom called tri hita karana (three harmonious relationships among the farmers with the God, with the environment, and with the other farmers and outsider persons). It has internal rules to manage the water distribution and allocation, operation and maintenance of irrigation facilities, fund raising and ritual activities. Mitigation efforts done by subaks are changing of planting schedule, rotation of cropping pattern, selecting of crops, improving of agricultural technologies, and performing of ritual activities. These were conducted simultaneously by farmers as subak members based on the climate information provided by government. It is needed to make a digitalized map for the forest area which is being a water catchment area as a source of water for subak. Besides, government should provide periodical information about the water balance and the plan of water allocation for irrigation and non-irrigation and improve the environmental upstream area. Besides, it must be also provided digitalized information about the climate forecast, water availability along the river and each weir/dam.
Title: Networking the Digital Earth for Rescuing our Coastal Areas
With a majority and an increasing proportion of the world’s people living in coastal areas, the goods and services provided by coastal ecosystems are in high demand and increasingly involved in trade-offs: exploitation of one ecosystem good or service reduces another ecosystem good or service. For example, the global increase in shipping, which transports 90% of the world’s goods, but also poses an increased risk from nonindigenous species (NIS) from ballast discharge and hull fouling. In this talk, I will present our work on coupled human-natural systems that provides the potential of moving the coastal invasive species research from descriptive science to a predictive science capable of informing policy. I will also present a case study on projecting ballast water mediated species introductions into the Arctic and ship-mediated dispersal of non-native species within the Arctic.
Title: Botnetome: The botnets that a company keeps
In this work, we show that there exist correlations in botnet activity between entities (companies) which enable improved identification of their (industry) sector and size. In analogy with the microbiome, we name this phenomena the botnetome. Working with a unique and comprehensive data-set (over 78 million events over 400 days involving 200 different botnets and 6,000 entities classified into 23 sectors and 5 size categories) generated by a leading security ratings firm we demonstrate progress towards establishing the botnetome as a viable concept. Our main result is that using deep learning on the botnet traffic emanating from a host entity we can predict its sector (size) 4.6 (2) times better than random.
We also take preliminary steps towards a theory of the botnetome. Based on observations of botnet communication patterns and norms of entity behavior (for botnet detection and removal) we conceived of a novel generative model - PREP (Periodically Reducing Eradication Probability) - that captures botnet-entity dynamics, from first principles. The PREP model predicts a power-law for the distribution of infection durations. And surprisingly, we were able to find several examples in the data where the distribution of infection lifetimes follows a power-law. The power law discovery is potentially useful for devising eradication strategies.
Title: Exaggerations in health research news: From universities to sub-disciplines, to print and social media
In this paper, we consider a dataset comprising press releases about health research from different universities in the UK along with a corresponding set of news articles and study how the basic information published in the scientific journals get exaggerated as they get reported in these press releases or news articles. For the first time in this paper, we compare the extent of exaggeration across various (i) universities, (ii) news agencies, and (iii) disciplines. Some of the key observations from this analysis are – as high as 60% articles from certain news agencies are exaggerated; more than 50% of the press releases from the certain universities are exaggerated; articles in topics like lifestyle and childhood are heavily exaggerated. As an additional objective we study how exaggerated news spreads over an online social network like Twitter. The fraction of tweets sharing exaggerated news articles is higher than the ones sharing non-exaggerated tweets almost a year after the publication date of the news. In these late arriving tweets, there is a sharp drop in fraction of certain LIWC categories like death words, sexual, sad and negative emotion; similarly, there is an increase in the fraction of words in the categories like assent, feel, religion and anxiety. The LIWC analysis finally points to a remarkable observation – these late tweets are essentially laden in words from opinion and realize categories which indicates that, given sufficient time, the wisdom of the crowd is actually able to tell apart the exaggerated news.
Title: Visualizing Urban spaces from spaces: decoding the digital earth
Urban planning is heavily dependent on localised user preference and behaviour. Data collection techniques have always been an area of concern in such research. Urban data collection is primarily based on stated preference questionnaire survey methods or long-time localised observation reconnaissance. These methods are extremely cost intensive and generate dis-continuous spatiotemporal data streams. Thus, restricting preparation of climate responsive urban plans, for meeting the two-degree climate targets, to a more reactive than a proactive approach. With the onset of the digital era, human digital footprints can be readily used to capture the impact of urban processes on climatic systems. The digital earth products like remote sensing data on the urban built environment and personal information communication data from cell phones, can now be utilised to characterise the spatial distribution of urban built areas and decode the proxies of allied functions. This talk demonstrates two urban studies which uses digital earth data to capture human behaviour for low carbon urban planning. Remote-sensed digital information and ICT based data is used to generate information about urban systems. The first study elaborates how human choices, broadcasted through their digital footprint can be used for low carbon urban mobility planning. The second experiment reveals how digital earth can be used to extract information for urban heat stress mapping. Thus, achieving the dual benefit of contributing to two-degree climate change target and societal well-being.
Title: Network Resilience
Resilience, a system’s ability to adjust its activity to retain its basic functionality when errors, failures and environmental changes occur, is a defining property of many complex systems. Despite widespread consequences for human health, the economy and the environment, events leading to loss of resilience–from cascading failures in technological systems to mass extinctions in ecological networks–are rarely predictable and are often irreversible. These limitations are rooted in a theoretical gap: the current analytical framework of resilience is designed to treat low-dimensional models with a few interacting components, and is unsuitable for multi-dimensional systems consisting of a large number of components that interact through a complex network. In the talk, I introduce how we bridge this theoretical gap by developing a set of analytical tools with which to identify the natural control and state parameters of a multi-dimensional complex system, helping us derive effective one-dimensional dynamics that accurately predict the system’s resilience. The proposed analytical framework allows us systematically to separate the roles of the system’s dynamics and topology, collapsing the behaviour of different networks onto a single universal resilience function. The analytical results unveil the network characteristics that can enhance or diminish resilience, offering ways to prevent the collapse of ecological, biological or economic systems, and guiding the design of technological systems resilient to both internal failures and environmental changes.
Title: The challenge of data for Habitat III New Urban Agenda monitoring: the case of Indonesia
Habitat III New Urban Agenda has passed the first year of its implementation, and a “Monitoring Framework” has been developed by UN Habitat to assist national and local governments to collect, analyze, and validate data by providing the definitions, methods of computation, and metadata of indicators, including geo-spatial indicators. However, a big challenge remains. In the recent 9th World Urban Forum critical questions has been raised: how should goals and targets be monitored? Who should do the work? How could monitoring efforts be parts of policy feedbacks, debates and discussions at the multi-level of urban governance systems? And more importantly, how can new and innovative types of data complement government statistics, and how new computational tools can enhance better understanding of the interconnected “urban SDG,” i.e., Goal-11’s targets of Sustainable Development Goals related to the “urban dynamics” of natural, engineered, and social systems. Indonesia as one of the highest urbanizing countries in Asia has to face a big challenge to respond to those questions. Yet, the heavy reliance on conventional statistical data remains a problem to the implementation and monitoring of the national New Urban Agenda in the country.
Title: Detection of rare events
Many rare events have a huge impact on society. Some examples of rare events are extreme weather events such as heavy precipitation, traffic incidents, natural disasters such as flooding, volcanic eruption, forest fires and health outbreaks. It is important to identify such rare events so that timely initiatives to mitigate them can be taken. Being able to forecast such events early may help in preventing them or taking prior steps to contain their effects. With the increasing availability of multiple data sources, progress in IOTs and sensors, cloud availability and computational methods there are enormous opportunities of gathering and monitoring data from multiple sources for real time anomaly and rare event detection which may be used for proper management of emergency situations.
However there are inherent challenges in the computational process of detecting rare events. Firstly, when datasets are available they are highly imbalanced. This may result in challenges for machine learning methods based on standard cost functions and evaluation metrics in detecting them. Different strategies of sampling, class specific costs and cost sensitive learning and evaluation metrics are used for overcoming these challenges. Secondly, anomalous events may not have any precedence so that supervised learning methods cannot be applied and unsupervised frameworks for detecting anomalies are useful.
In this talk we will discuss some approaches to identifying rare events and anomalies. Since many incidents of interest are spatiotemporal in nature we will discuss methods that work with such data.
Title: Controling disease outbreaks in a connected world
The recent Ebola outbreak in west Africa and the Zika outbreak in South America are reminders of the age-old societal concerns about how diseases spread and how to control them. While significant strides have been made in the past century in medicine and public health, infectious diseases continue to be a very significant social burden, accounting for more than 13 million deaths a year. The risk of a large pandemic outbreak remains high, due to global trends of increased urbanization and travel, climate change and other issues. The public health policy process is very complex, and involves a number of steps: developing accurate mathematical models, integrating diverse datasets for disease surveillance, designing efficient simulations, epidemic forecasting and developing methods to evaluate public health interventions. The difficulty of getting incidence data, non-linearity of mathematical models and the number of factors at play in the spread of diseases make these very challenging problems, which are at the interface of AI, data mining, machine learning, high performance computing and theoretical computer science, as well as mathematics, economics and statistics. In this talk, we provide an overview of the state of the art of this very exciting and multi-disciplinary area.
Title: Water sector in India - challenges and prospects
By 2030, under an average economic growth scenario and if no efficiency gains are assumed, global water requirements would grow from 4,500 BCM to 6,900 BCM. This is a full 40 percent above current accessible, reliable supply. This global figure masks many local gaps; one-third of the population, concentrated in developing countries, is likely to live in basins where this deficit is larger than 50 percent.
Agriculture accounts for approximately 3,100 billion m3, or *71 percent of global water withdrawals today*, and without efficiency gains will increase to 4,500 billion m3 by 2030 (*a slight decline to 65 percent of global water withdrawals*). The water challenge is therefore closely tied to food provision and trade. Centers of agricultural demand, also where some of the poorest subsistence farmers live, are primarily in *India (projected withdrawals of 1,195 billion m3 in 2030), Sub-Saharan Africa (820 billion m3), and China (420 billion m3). *
*Industrial withdrawals account for 16 percent of today’s global demand, growing to a projected 22 percent in 2030*. The growth will come primarily from *China (where industrial water demand in 2030 is projected at 265 billion m3, driven mainly by power generation), which alone accounts for 40 percent of the additional industrial demand worldwide*.
Demand for water for *domestic use will decrease as a percentage of total from 14 percent today to 12 *percent in 2030, although it will grow in specific basins, especially in emerging markets.
In this scenario, what are the challenges and prospects for India?
Title: Application of data analyses and assimilation in delineating groundwater scenarios of India
By using a combination of ground-based in-situ groundwater level data, NASA satellite-based estimates of groundwater storage, numerical analyses and simulation of global models on groundwater storage changes, we delineated the long-term, decadal-scale groundwater trends over India. Our study shows that in situ groundwater level trends shows simultaneous occurrence of wells with increasing and decreasing water level between 2005 and 2013. However, parts of the Indus-Ganges- Brahmaputra basin in India mostly show reducing groundwater levels whereas parts of western and southern India show increasing trends. The Groundwater storage (GWS) anomaly shows strong spatial variability in the study region. Observed GWS data indicate renewal of GWS in western (B) and southern (E) zones at a rate of 1.06±0.03, and 0.31±0.02 km3/year. On the other hand, the northern (zone A) and eastern (D) zones have been subjected to rapid GWS depletion at a rate of 4.55±0.11 km3/year and 3.59±0.14 km3/year, respectively. Satellite-based estimates indicate rapid depletion in northern (zone A) and eastern (zone D) zones at a rate of -1.40±0.14 and -1.16±0.35 cm/year (-14.02±1.37 km3/year and -14.49±4.36 km3/year) in the study period, respectively.
Title: Next-Generation Challenges and Opportunities in Big Earth Data to Establish Dynamic Connectedness
We are in the midst of a new-era of information generation where a massive, unprecedented influx of data coming from expanding populations of formal and informal sensors present new challenges and opportunities for knowledge discovery. A new generation of non-traditional geo-locatable data sources remains a largely underutilized information resource. These data can be described as non-authoritative, unstructured, largely heterogeneous, of varying quality, and commonly undocumented, thus making them difficult to use. The volume, velocity, variety, and validity of traditional and non-traditional geospatial data sources present unique challenges and opportunities for data driven analytics to discover hidden patterns, identify data and semantic relationships, and fill multi-scale spatial and temporal knowledge gaps, all of which contribute to revealing dynamic network connectedness. The foundational development and fusion of these disparate data sources can realize substantial and unrealized benefits, particularly when coupled with physics-based numerical modeling, in many domains including environmental monitoring, disaster management, critical infrastructure management and resilience, sustainable development, the food-energy- water nexus, human mobility, and social response and interaction.
Title: Double Feature -- "Real-Time Census Using Social Media" AND "Bias in Perception of Truth in News Stories and their Implications for Fact Checking"
First, given the importance of demographic data for monitoring development, the lack of appropriate sources and indicators for measuring progress toward the achievement of targets–like the United Nations’ “2030 Agenda for Sustainable Development”–is a significant cause of uncertainty. Data innovation, like new digital traces from a variety of technologies, is seen as a significant opportunity to inform policy evaluation and to improve estimates and projections. In the first part of the talk, I will present a previously untapped data source for demographic research: Facebook’s advertising platform. I will show how this freely available source allows advertisers and researchers to query information about socio-demographic characteristics of Facebook users, aggregated at various levels of geographic granularity.
Second, a flurry of recent research has focussed on understanding and mitigating the threat of "fake news" stories spreading virally on social media sites like Facebook and Twitter. However, few studies focus on how users perceive truth in viral news stories. In the second part of the talk, I will present our recent attempt to quantify the extent to which users can implicitly recognize (perceive) the accurate truth-level of a news story (obtained from fact checking sites like Snopes and Politifact). Our findings have important implications for fact-checking. Specifically, I will argue that while the stories that are in need of being fact checked are the stories where users exhibit the largest perception biases, existing fact checking strategies wrongly prioritize fact checking stories based on their actual truth-level rather than perception biases.
Title: THE STUDY OF BAY OF BENGAL USING HIGH RESOLUTION REGIONAL OCEAN MODELING SYSTEM (ROMS)
Bay of Bengal (BOB) plays a very important role in Indian weather and climate. Rainfall during summer and winter monsoon are affected by the conditions prevailing in this bay. BOB also houses large number of tropical cyclones which causes severe destruction in the subcontinent. In order to predict and study the bay, A high-resolution (10 km x 10 km) multiscale ocean modeling system was developed for the Bay of Bengal (BOB) region. The Regional Ocean Modeling System (ROMS) was implemented and initialized with Levitus 1/4o climatological fields for short-term forecasting. The results from these climatology-based model simulations for three representative months (February, June and October) in three different seasons (winter, summer and autumn) are discussed and also validated. The multiscale features during February include an anticyclonic basin-scale gyre with a strong western boundary current (WBC) in the western basin, the formation of several shallow mesoscale eddies in the head of the Bay and a cyclonic sub-basin-scale Myanmar Gyre in the northeast. During June, no well-defined boundary current is simulated along the Indian coast; instead, alternating cyclonic and anticyclonic eddies appear along the east coast with cross-basin eastward flow to support a deep cyclonic Andaman Gyre. In October, a basin-scale cyclonic gyre with a continuous well-defined East India Coastal Current (EICC), weak inflow from the Malacca Strait to the Andaman Sea and advection of BOB water into the Arabian Sea via the Palk Strait are simulated well by the model. We also present an application and synoptic validation of the system for October 2008 in a Hindcast mode reasonably predict the ocean currents, temperature and salinity. The forecast error increases as the forecast time window increases, although the system has a reasonable predictability for up to seven to ten days.
Title: Fusing Urban Mobility & Social Media for Human-Centric ‘Smart City’ Insights
Cities are increasingly providing public access to a variety of transportation-related “big data”, such as the real-time location of taxis, crowdedness level of buses and arrival times of trains. In this talk, I will demonstrate the concept of socio-physical analytics, which fuses such transportation-derived urban mobility data with social media sensing (via platforms such as Twitter & Foursquare) for a variety of new human-centric smart city insights and services. I will primarily describe three threads of analytics-driven work. One combines urban mobility data with sensor data from personal mobile devices to obtain novel insights into commuting experiences, which can be reinforced via social media sensing. Another combines multi-modal transportation data (e.g., occupancy levels of buses, dropoff and pickup rates of taxis) with social media content to detect and localize urban micro-events. The final thread utilizes social media and urban mobility data to develop a multi-factor model that predicts the likely failure of individual retail businesses.
Title: Data Analytics of Water Supply in West Bengal with reference to alleviation of public health
Title: Geospatial modelling and algorithms for evacuation planning
Making efficient evacuation plans for large buildings and highly populated areas is an important task for city administrators. The plans require middling the buildings or areas as a capacitated network with well defined exit points and to analyse evacuation time in different scenarios. Such type of analyses helps us in both the redesign of the building/locality to improve evacuation times as well as for more optimal allocations of the spaces. We model the problems in for various type of situations and present heuristic algorithms which are extensions of well known capacity constrained route planner (CCRP) algorithm: (i) Creating additional exits with ladders for faster evacuation, (ii) providing rescuers to assist injured people. We also consider complexities of modelling of highly contested urban areas typical of Indian cities where one has to deal with small lanes, pedestrians, and vehicles of all kinds. Considering the larger problem of disaster management, we highlight the importance of standards and interoperability for coordination and control. We also consider unpredictability of human movements and their impact on sensitivity and efficacy of the disaster management plans using simulations. We have designed Evacuation Planner tool which is publicly available and can be used for making indoor network from floor images, computing evacuation plans and simulating evacuee behaviours.
Title: Post Disaster Management using Social Media
In recent times, humanitarian organizations increasingly rely on social media to search for information useful for disaster response. Useful information such as reports of urgent needs, missing or found people that, if processed timely, can be very effective for humanitarian organizations for their disaster response efforts. To effectively utilize microblogging sites during disaster events, it is necessary to not only extract the situational information from the large amounts of sentiments and opinions, but also to summarize the large amounts of situational information posted in real-time. During disasters in countries like India, a sizeable number of tweets are posted in local resource-poor languages besides the normal English-language tweets. For instance, in the Indian subcontinent, a large number of tweets are posted in Hindi / Devanagari, and some of the information contained in such non-English tweets are not available (or available at a later point of time) through English tweets. Our proposed methodology is developed based on the understanding of how several concepts evolve in Twitter during disaster. This understanding helps us achieve superior performance compared to the state-of-the-art tweet classifiers and summarization approaches on English tweets.
Title: Route Recommendations for Idle Taxi Drivers: Find Me the Shortest Route to a Customer!
We study the problem of route recommendation to idle taxi drivers such that the distance between the taxi and an anticipated customer request is minimized. Minimizing the distance to the next anticipated customer leads to more productivity for the taxi driver and less waiting time for the customer. To anticipate when and where future customer requests are likely to come from and accordingly recommend routes, we develop a route recommendation engine called MDM: Minimizing Distance through Monte Carlo Tree Search. In contrast to existing techniques, MDM employs a continuous learning platform where the underlying model to predict future customer requests is dynamically updated. Extensive experiments on real taxi data from New York and San Francisco reveal that MDM is up to 70% better than the state of the art and robust to anomalous events such as concerts, sporting events, etc.
Title: Evolving Risk Profiles and Resilience of Built-Natural Systems
Built and natural systems, such as lifeline infrastructure networks and ecological systems ,need to be robust to perturbations, as well as gracefully recover from damage, caused by extreme events. The lifeline networks are interdependent, thus cascading failures may further complicate the challenges. The extreme events or severe stresses can be more intense than the design perturbations, both owing to natural variability and because of change, or a combination of the two. In particular, the statistical attributes of weather and hydrometeorological extremes have shown signs of change across many regions of the world owing to a combination of issues such as greenhouse-gas emissions induced warming, regional changes in land use or irrigation patterns, and urbanization. These changes in hydrometeorological extremes may alter the fundamental basis of proactive design, as well as advance plans for robustness and recovery. Here we develop geospatial and temporal risk profiles of potential nonstationary hydrometeorological extremes and examine what these could mean for the resilience of lifeline networks and network-of-networks at scales ranging from urban communities to megaregions, or interconnected urbanized areas, as well as larger regions. Optimization and network science based strategies are developed for translating the risk profiles to actionable plans for enhancing robustness to shocks and pulses as well as for effectively and efficiently recovering from damage. Resilience of Indian Railways Network, US National Airspace, Boston's mass transit system and 39 ecosystems across the globe are discussed as case studies.
Title: Resilience and the coevolution of networks
The increasing complexity and intelligence of critical infrastructures, such as the transportation network or the financial network, enable us to make better decisions and reduce the associated risks with such decisions. However, these infrastructures themselves are at risk due to a variety of factors including extreme weather, climate change, cyber-attacks, extreme changes in the operating environment, etc. The risks are further compounded by the complex interdependence between these networks which leads to cascading failures – researchers have developed several theoretical models to explain how the interdependence of networks can amplify small failures that ping pong back and forth across networks causing larger and larger outages and ultimately leading to a total collapse.
At the same time what we see in practice is that networks are remarkably resilient. Though predictions of apocalyptic collapses are legion, the real world nevertheless seems to bounce back from outages avoiding catastrophic meltdowns. In this project, we explore how different modes of coevolution between networks can affect their dependence structure and in turn, determining their resilience or ability to bounce back from failures. Our preliminary results indicate that one seemingly counter-intuitive factor that could explain the resilience of real-world networks is that they are coupled just right: neither too tightly nor too loosely. In other words, networks whose coevolution structures are completely correlated or completely uncorrelated seem to show less resilience as compared to networks which are semi-correlated. We provide both analytical results and simulation-based analysis on real-world network to justify our claims.
Title: Robustness of Interconnected Ecological Networks
Plant-pollinator and trophic networks have been studied independently for their robustness to environmental change perturbations. These networks, however, also interact with each other and hence, are interconnected. The goal of this study is to assess the impact of this interconnectedness on the robustness of plant-pollinator interactions. Adopting a network of networks approach has been found to cause an earlier onset of pollinator extinction while making the collapse more gradual. These changes can be attributed to greater stress on the plant species due to the introduction of trophic interactions and also to the possibility of a feedback mechanism between the plant-pollinator and trophic networks leading to a cascade of species extinctions. The results of this study have major implications for global food security as extinction of pollinators would negatively impact crop production.