Artificial Intelligence (AI)
Since its launch in November 2022, ChatGPT has sparked a significant impact on Artificial Intelligence’s presence in in the public domain, evoking curiosity, interest, and awe. The subject has led to urgent calls by industry leaders for establishing an ethical framework for AI, as well as regulating its development, usage, and deployment. We undoubtedly stand at the threshold of a new era, where generative AI capable of creating text, images, music, and video upon user command represents merely the visible tip of a very large iceberg. Beneath lies a realm, only partially known to subject matter experts.
AI’s applications span a vast array of industries, from robotics to healthcare, finance to autonomous driving, speech recognition to industrial automation. The ultimate ambition of AI lies in creating systems that can match or surpass human capabilities in specific tasks, unlocking new possibilities and offering efficient solutions to intricate problems.
One area where AI can make a positive impact is disaster management. This article provides a review of the state of the art based on published papers as well as conclusions by the authors.
Climate change and extreme events
In tandem with the rapid pace of technological advancement, we find ourselves navigating an era profoundly impacted by the often-catastrophic consequences of climate change. Unprecedented extreme weather events, ranging from heatwaves and floods to droughts and intense storms, serve as visible evidence of the far-reaching effects stemming from the long-term disruption of average temperature, precipitation, wind patterns, and other climatic parameters. These events are becoming increasingly frequent and severe across the globe.
Ironically, the challenge we face when addressing such events lies not in the scarcity of data but in our ability to swiftly process and interrelate these vast datasets, while developing predictive models and short- and very short-term action plans.
In this context, AI emerges as a powerful tool with the potential to revolutionize the speed and way we gather and analyze data. By transforming our data processing methodologies, it enables us to derive the “intelligence” required for rapid and informed decision-making.
Disaster Management and AI
Disaster Management, defined as a crucial set of strategies and actions aimed at handling extreme events, holds the key to mitigating negative impacts by focusing on prevention, response, and recovery efforts. Focused on the protection of human lives and material assets, such as buildings and infrastructure, it involves coordinating resources, executing emergency and resilience plans, and fostering collaboration among diverse entities, including governments, international organizations, local communities, and individuals. The complexity of this environment demands real-time understanding, correlation, and processing of data from various sources, as well as the coordination of large-scale response plans and the dissemination of information to the public.
Despite the wealth of available data and sensor technologies, a fundamental challenge hinders our readiness to respond effectively: the sheer volume of data overwhelms our capacity to derive value from it. Addressing this obstacle requires significant progress and innovative efforts. Leveraging AI technologies and systems theoretically allows for more efficient governance of emergency situations, facilitated by accurate real-time data analysis, machine learning, and the automation of critical operations. As a powerful ally for local and global security, AI holds the promise of transforming how we approach Disaster Management.
The use of AI systems in the preparatory phase
The initial phase of Disaster Management, known as the “before” phase, plays a crucial role in implementing prevention and preparedness measures to mitigate the potential impact of disasters. Prevention involves a systematic approach that identifies risks, assesses vulnerabilities, and evaluates the likelihood of events evolving. The goal is to implement preventive measures that can effectively minimize the consequences of such events.
While we won’t delve into the specific aspects of prevention and preparedness, let’s explore the systems that can aid during this stage. Advanced tools like LSTM (Long Short-Term Memory) and BiLSTM (Bilateral Long Short Memory) offer real-time monitoring of risks and enable forecasting of changes. These tools prove particularly valuable when dealing with natural, environmental, and human-originated events. By leveraging data sources such as satellite imagery and topographic maps, these systems support comprehensive hazard monitoring efforts.
Through the utilization of these cutting-edge technologies, Disaster Management teams can gain valuable insights, enabling them to make informed decisions and take timely actions to safeguard communities and critical assets. As the “before” phase sets the groundwork for disaster resilience, the integration of AI-powered tools enhances our ability to proactively respond to potential threats and minimize the impact of disasters.
The use of advanced AI systems like GRUs (Gated Recurrent Units) has revolutionized disaster management by enabling real-time detection and prediction of changes, issuing timely alerts, and monitoring emergency situations related to natural and environmental disasters. These powerful tools draw insights from historical data, such as river levels, rainfall patterns, and vegetation types, to offer accurate forecasts and identify critical emergency scenarios.
The Integrated Flood Alert System (IFAM) stands as a comprehensive solution, leveraging a wide array of data sources. IFAM integrates information from weather stations, hydrological sensors, satellite data, weather models, river levels, rainfall measurements, water flow velocity, land topography, and soil type. Driven by machine learning algorithms, this system generates alerts encompassing weather forecasts, river levels, and flood-prone areas, effectively communicating this vital information to local authorities and civil defense agencies. During emergency situations, IFAM empowers management agencies to make well-informed decisions and optimize resource allocation.
In the ongoing efforts to combat the climate crisis, major tech players like Google have taken significant steps. Google’s Flood Hub, a flood forecasting tool, now provides 7-day advance forecasts in 80 countries, protecting a staggering 460 million people.
Artificial neural networks (ANNs) such as Convolutional Neural Network (CNN) are mainly used to analyze images (satellite or from drones) and provide early warning and data analysis in various emergency contexts, such as natural disasters, industrial, health and terrorist incidents.
Private companies are also actively contributing to the advancement of disaster management through the development of cutting-edge platforms and products that integrate physical data with machine learning. Companies like OneConcern provide software platforms that play pivotal roles in guiding decision-making and promoting collaboration during relief operations by providing detailed information about critical infrastructure, including hospitals and evacuation centers, supporting planning and resource management across diverse disaster scenarios, be it natural calamities, transportation incidents, industrial emergencies, health crises, or terrorist threats.
The integration of AI-driven systems and data analytics in disaster management is opening new frontiers of preparedness and response capabilities, greatly enhancing our ability to safeguard lives, infrastructure, and communities in the face of unforeseen challenges.
The use of AI systems in the response phase
The integration of NLP (Natural Language Processing) tools, a branch of AI focused on enabling computers to comprehend spoken language like humans, holds tremendous potential in enhancing emergency event management. Leveraging NLP, sentiment analysis can be conducted on messages, tweets, and social media posts to identify distress alerts, urgent needs, and other relevant information. Moreover, NLP facilitates the analysis of situation reports from various organizations involved in emergency management, including government agencies, non-governmental organizations, and law enforcement entities. These reports contain crucial details about the disaster’s scope, infrastructure impact, population needs, and other pertinent information. Through machine learning techniques, NLP automatically categorizes texts, providing valuable insights for decision-making. NLP’s multi-language capability also enables instantaneous translation, facilitating international emergency coordination.
In parallel, GIS (Geographic Information System) technology empowers real-time geospatial analysis, supporting evacuation planning, escape route identification, and coordination of transfers. By integrating critical infrastructure, hazards, and demographic data into a single map, GIS highlights areas of risk and vulnerability. These real-time maps enable data-driven decision-making and foster effective collaboration among all involved agencies in the emergency response. Forward-thinking companies are exploring AI integration into GIS systems to further enhance decision-making capabilities.
Artificial neural networks (ANNs), such as DQN (Deep Q Network, are trained to analyze environmental conditions, emergency events, potential actions, and the consequences of those actions. After training, ANNs can intelligently determine the best course of action based on the present situation. This capacity to make data-driven decisions proves invaluable in guiding emergency management operations.
The use AI systems in the recovery phase
During the recovery phase of disaster management, key data points such as the precise event location, disaster intensity and extent, demographic information, population count, and the presence of vulnerable groups become paramount. Effectively addressing the challenges of damage assessment and mapping of affected areas can be enhanced using AI-based tools.
One such tool involves image processing utilizing algorithmic techniques like Edge Detection. This enables the analysis of captured images to identify damaged structures and critical elements, facilitating the mapping of affected regions and establishing viable routes following natural, transportation, or industrial disasters.
Unmanned Aerial Vehicles (UAVs) equipped with photogrammetry capabilities play a vital role in condition monitoring and support for rescue operations. Drones equipped with thermal, and LiDAR (light detection and ranging) sensors capture crucial imagery that is then analyzed using specialized AI algorithms. These analyses aid in the detection of missing or trapped individuals in hard-to-reach areas, enhancing the effectiveness of rescue efforts.
AI-based tools such as the U-Net based Algorithm, utilizing Machine Learning algorithms, contribute to risk assessment, reconstruction prioritization, and the segmentation of affected areas. Leveraging labeled image datasets from various sources, like satellite or aerial imagery, these algorithms can accurately identify damaged zones and assist in resource planning during the recovery phase. Artificial neural networks (ANNs) and convolutional neural networks (CNNs) further enhance the process, enabling precise classification of damaged areas, comprehensive damage assessment, and strategic resource allocation, regardless of the disaster’s origin. These networks also facilitate mapping of residual resources, such as intact structures, water, or energy reserves, crucial information that optimizes coordination in recovery operations and accelerates the restoration of post-disaster activities.
Episteme, Techne, and Phronesis
Before drawing conclusions and to better understand the current situation, it may be useful to refer to the different forms of knowledge proposed by Aristotle: Episteme, Techne, and Phronesis. Episteme represents universal knowledge, explaining the “why” of things, which has largely shaped our understanding thus far. Techne, on the other hand, denotes technical knowledge that empowers us to know “how” to accomplish tasks. Thanks to the scientific breakthroughs achieved it has characterized the 20th century. However, looking ahead, Phronesis, or “practical wisdom”, becomes increasingly relevant. This form of practical knowledge entails making the right decisions at the right time, recognizing the dynamic nature of events that cannot always be addressed through rigid rules.
In the domain of disaster management, a field defined by complexity and uncertainty, Phronesis becomes crucial for making appropriate decisions and addressing emerging challenges. Successfully managing critical events necessitates flexibility and adaptive capabilities—qualities that Phronesis uniquely provides. Hence, it makes sense to develop approaches and solutions that blend codified knowledge with human experience, contemplating how to integrate AI to support and enhance decision-making processes[1].
The utilization of AI in Disaster Management presents both crucial prerequisites that become limitations and opportunities, depending on the context of application. One critical prerequisite is access to data, which plays a pivotal role. In more developed countries, the abundance of sensors and data allows for more effective implementation of prevention policies and plans. Conversely, many low-income regions face challenges in accessing reliable data, potentially limiting the application of AI-based solutions.
Ensuring agile data accessibility necessitates seamless integration between different systems and other IT infrastructures. Access to the Internet and stable Wi-Fi or satellite connections are vital elements for effective communication and data exchange between systems. This, however, remains a challenge in different parts of the world, posing a significant hurdle for the widespread implementation of AI in disaster management.
When considering AI-based systems in this context, it becomes fitting to abandon the logic of distinct phases (before, during, and after) and instead think in terms of a “continuous phase”. Critical situations demand dynamic and constant management, relying on the continuous use of AI tools to ensure the flow of information, analysis, and timely decisions necessary for effective event management.
Another critical aspect is data validation, as some data may be unreliable or “noisy”[2]. Operating on reliable and accurate data demands appropriate validation processes to identify and discard compromised data, ensuring high reliability and security during data processing and achieving consistent and effective results.
Dealing with such complexity necessitates not only AI expertise but also a transdisciplinary approach, breaking free from vertical silos of expertise and embracing cooperation between AI specialists and experts from other disciplines: environmental sciences, engineering, social sciences, crisis and disaster management, communications.
The internationally renowned Italian semiologist Umberto Eco pointed out that with any new technological horizon, the dichotomy between apocalyptic and integrated perspectives emerges[3]. It is in our view crucial, at least at this stage, to view AI as a collaborative tool rather than a substitute for human experience. AI should complement human capabilities, saving valuable time, providing timely and precise information, enabling faster and more accurate decision-making.
In conclusion, at a time when our planet is affected by increasingly violent and difficult to predict events linked to climate change AI can become a valuable ally in disaster management enabling advanced analysis, forecasting, decision support, and resource management. By incorporating AI as a complementary tool alongside human expertise, we can leverage the best of both worlds to enhance response efficiency and ultimately ensure the safety and well-being of affected communities.
About the authors
Chiara Munno, graduate in strategic communications
A graduate of the Faculty of Strategic Communication at IULM University in Milan, she is specializing in the study of communication and crisis management. During her academic career she spent 6 months as an intern at TT&A Advisors, a Swiss based crisis management advisory consultancy.
Irene Proto, crisis management and crisis communications consultant
With an academic background in Communication studies between Padua, Brussels, Turin and Sciences Po Grenoble she started her professional career in Paris at Heiderich Consultants. In 2020 she joined TT&A Advisors consulting in crisis management and crisis communications. She holds an Executive Master in Complexity Management at the Complexity Institute (Italy).
Patrick Trancu, crisis management advisor, CEO TT&A Advisors
TEDx speaker, Crisis Advisor, CBCI, editor and co-author of “The State in Crisis. Pandemic, chaos, and questions for the future” (Franco Angeli, 2021), a multidisciplinary analysis of the management of the first phase of the Covid19 crisis in Italy. He is CEO at Swiss based TT&A Advisors where he works with multinational companies and institutions to prepare and manage unexpected critical events.
Bibliography and sitography
Abid, S. K., Sulaiman, N., Chan, S. W., Nazir, U., Abid, M., Han, H., Vega-Muñoz, A., et al. (2021). Toward an Integrated Disaster Management Approach: How Artificial Intelligence Can Boost Disaster Managment. Sustainability.
Cortez, B., Carrera, B., Kim, Y.-J., & Jung, J.-Y. (2018). An architecture for emergency event prediction using LSTM recurrent neural networks. Expert Systems with Applications.
Elfahim, O., Laoula, E. M., Youssfi, M., Barakat, O., & Mestari, M. (2022). Deep Reinforcement Learning Approach for Emergency Response Management. 2022 International Conference on Intelligent Systems and Computer Vision (ISCV). Fez, Marocco: IEEE.
Erdelj, M., Król, M., & Natalizio, E. (2017). Wireless Sensor Networks and Multi-UAV systems for natural disaster management. Computer Networks.
European Commission. (2019, September 12). AI can help us fight climate change. But it has an energy problem, too. Tratto da Horizon. The EU Research & Innovation Magazine: https://ec.europa.eu/research-and-innovation/en/horizon-magazine/ai-can-help-us-fight-climate-change-it-has-energy-problem-too.
Fang, C., Fan, X., Zhong, H., Lombardo, L., Tanyas, H., & Wang, X. (2022). A Novel Historical Landslide Detection Approach Based on LiDAR and Lightweight Attention U-Net. Remote Sensing.
Gevaert, C. M., Carman, M., Rosman, B., Georgiadou, Y., & Sode, R. (2021). Fairness and accountability of AI in disaster risk management: Opportunities and challenges. CelPress Open Access.
Guha, S., Jana, R. K., & Sanyal, M. K. (2022). Artificial Neural Network approaches for disaster management: A literature review. Journal of Disaster Risk Reduction.
Ivić, M. (2019). Artificial Intelligence and Geospatial Analysis in Disaster Management. The International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences.
Kareem, K. Y., Seong, Y., Kim, K., & Jung, Y. (2022). A Case Study of Tidal Analysis Using Theory-Based Artificial Intelligence Techniques for Disaster Management in Taehwa River, South Korea. Water.
Kaur, M., Kaur, P. D., & Sood, S. K. (2022). ICT in disaster management context: a descriptive and critical review. Environmental Science and Pollution Research.
Kim, S. S., Shin, D. Y., Lim, E. T., Jung, Y. H., & Cho, S. B. (2022). Disaster Damage Investigation using Artificial Intelligence and Drone Mapping. The international Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences.
Liu, Y., Hany, M., Montgomery, K., & Candell, R. (2021). A Collaborative Robot Work-Cell Testbed for Industrial Wireless Communications. National Insitute of Standards and Technology.
Magid, E., Pashkin, A., Simakov, N., Abbyasov, B., Suthakorn, t., Svinin, M., & Matsuno, F. (2020). Artificial Intelligence Based Framework for Robotic Search and Rescue Operations Conducted Jointly by International Team. Proceedings of 14th International Conference on Electromechanics and Robotics “Zavalishin’s Readings”.
Meng, C. (2021). Natural Language Processing for urban research: A systematic review. CelPress.
Minevich, M. (2022, Luglio 8). How To Fight Climate Change Using AI. Tratto da Forbes: https://www.forbes.com/sites/markminevich/2022/07/08/how-to-fight-climate-change-using-ai/?sh=801b9c42a838.
Munawar, H. S. (2020). Flood Disaster Management Risks, Technologies, and Future Directions. In S. R. Muthukumaran Malarvel, Machine Vision Inspection Systems: Image Processing, Concepts, Methodologies and Applications.Londra: Scrivener Publishing LLC.
Munawar, H. S., Hammad, A. W., Waller, S. T., Thaheem, M. J., & Shrestha, A. (2021). An Integrated Approach for Post-Disaster Flood Management via the Use of Cutting-Edge Technologies and UAVs: A Review. Sustainability.
Nimmagadda, S., Sivakumar, S., Kumar, N., & Haritha, D. (2020). Predicting Airline Crash due to Birds Strike Using Machine Learning. 2020 7th International Conference on Smart Structures and Systems (ICSSS). Chennai, India: IEEE.
Shakeel, M., Itoyama, K., Nishida, K., & Nakadai, K. (2021). Detecting earthquakes: a novel deep learning-based approach for effective disaster response. Applied Intelligence.
Subhashini, R., Thomas, J. J., Sivasangari, A., Mohana, P., Vigneshwari, S., & Asha, P. (2022). Artificial intelligence–based intelligent geospatial analysis in disaster management Author links open overlay panel. Advances of Artificial Intelligence in a Green Energy Environment.
Sun, W., Bocchini, P., & Davison, B. D. (2020). Applications of artificial intelligence for disaster management. Natural Hazards.
Tan, L., Guo, J., Mohanarajah, S., & Zho, K. (2020). Can we detect trends in natural disaster management with artificial intelligence? A review of modelling practices. Natural Hazards.
United Nations Office for Disaster Risk Reduction. (2017, Maggio 22). Simulation using artificial intelligence in training senior crisis staff members and in validating contingency plans. Tratto da UNDRR: https://www.undrr.org/publication/simulation-using-artificial-intelligence-training-senior-crisis-staff-members-and.
Wheeler, B. J., & Karim, H. A. (2020). Deep Learning-Enabled Semantic Inference of Individual Building damage magnitude from Satellite Images. Algorithms.
Xu, Y., Hu, C., Wu, Q., Jian, S., Li, Z., Chen, Y.,Wang, S., et al. (2022). Research on particle swarm optimization in LSTM neural networks for rainfall-runoff simulation. Journal of Hydrology.
[1] A Cravera, Nell’era dell’AI scopriamo l’esigenza di nuovi modelli educativi, Il Sole 24 ore, July 4th, 2023.
[2] L.P. Faina Garcia, A. De Carvalho, A.C. Lorena, Noisy Data Set Identification, 628 – 638, 2013.
[3] U. Eco, Apocalittici e Integrati. Comunicazioni di massa e teorie della cultura di massa, Bompiani, 1964.
Authors: Chiara Munno, Irene Proto and Patrick Trancu
If you liked this article, you might also want to read this one.
