Forecasting the future: AI in early warning systems

Google will present an overview of how they are using AI for early warning at a global scale, with AI applications from pillars 1-3. This will cover:

• A high level overview of product solutions throughout the Multi-Hazard Early Warning Systems value chain as a part of Research and Crisis Response initiatives.
• A deep dive on Flood Hub, which uses AI to make critical riverine flood forecasting accessible up to seven days in advance of a flood event. Google will present the hydrological and inundation models, and talk through the outputs of the model, particularly the ungauged forecasts. These riverine flood forecasts are available in over 90 countries, disseminated on Google Search, Maps, and Android notifications to help more people access flood information with plans.

This segment will delve into the early pilot applications of AI for Early Warnings, focusing on the conceptualization, technologies utilized, identified data supporting various use cases, and specific pilot projects underway. The following use case will be presented which are linked to Pillar 1 and 4. An AI-powered mapping tool, developed through a joint effort by United Nations Office for Disaster Risk Reduction (UNDRR), the Ethiopian government, Microsoft, Planet, and the Institute for Health Metrics and Evaluation (IHME). This innovative tool aims to identify populations in high-risk areas for natural hazards by integrating building density and hazard data.

Ecuador is among the first 30 countries in the EW4All cohort, and National Institute of Meteorology and Hydrology (INAMHI) is a key institution for its implementation. The speaker will focus particularly on the deployment of hydroinformatics platforms, such as the one supported by the AI-powered GEOGLOWS streamflow model under the initiative of the Group on Earth Observations (GEO) dedicated to global water sustainability, which also employs ECMWF reanalysis and forecast services. Additionally, he will present the potential of this tool in Ecuador, its current applications at the national, local, and community scales, and their connection to other EW4All pillars while collaborating with the Risk Management Secretariat. The speaker's intervention will also include an overview of the hydrometeorological monitoring network and reactivation plans aimed at improving GEOGLOWS and hydrometeorological services in Ecuador.

In this session, the use of AI to identify populations that have no ability to receive early warning notifications due to lack of electricity, and therefore have no access to cellular, broadband, or siren-based technologies, will be explored. A pilot with ITU will be announced to identify partner countries. This will include the introduction of a framework for quantifying and spatially understanding risk information, essential for effective dissemination to decision-makers and early warning system administrators. By applying this framework, it is possible to identify areas in need of intervention and tailor early warning strategies accordingly. The goal is to demonstrate AI capabilities using a demonstration from 3 Small Island Developing States and secure engagement from other nations, which is linked to Pillar 3.

Funded by the European Commission Horizon Europe Programme, the consortium of 29 partners aims to develop a standardised Decision Support Dissemination System (DSDS) for risk and vulnerability assessment, utilising AI and leveraging and enhancing existing capabilities, and leading to a fully integrated multi-hazard platform. The Mediterranean and pan-European forecast and Early Warning System against natural hazards project aims to build a product at TRL6-8 by end 2026. It is based on four paired twin pilot sites, that bridge areas with different climatic and geographic conditions but subject to similar hazardous events, to deliver innovative forecasting and impact assessment services, and support EW4All implementation in target countries.

Everbridge will discuss the use of AI for public safety and introduce the AI chatbot and EW4All application.

The space industry is undergoing a transformative phase driven by rapid advancements in sensor and digital technologies. This rapid evolution, characterised by unprecedented speed, is reshaping how we collect, analyse, and utilise data from space. The convergence of ICT breakthroughs, such as cloud computing and storage, alongside a sensing and connectivity revolution, has ushered in a new era of big data. This wealth of dynamic data presents both opportunities and challenges, particularly in leveraging Artificial Intelligence (AI) to extract meaningful insights. This presentation will showcase some of ESA's research and partnerships within the framework of the Civil Security from Space (CSS) programme, encouraging collaboration to harness the synergies of big data and AI for innovative space services, with a particular focus on rapid response and disaster management. https://connectivity.esa.int/civil-security-space

This session seeks to delve into strategies for unlocking the potential of AI-driven Early Warning Systems (EWS), not only through financial avenues but also through active participation and resource support. We will explore diverse pathways for engagement beyond monetary contributions, fostering a holistic discussion on how various stakeholders can contribute to advancing AI solutions for climate resilience. Alongside exploring funding opportunities, we will delve into ways to get involved, such as sharing expertise, providing technical resources, or offering institutional support. The session will highlight the Green Climate Fund's (GCF) project modalities as an example to stimulate discussion on investment possibilities, while also encouraging exploration of non-monetary forms of engagement to harness the full potential of AI in mitigating climate-related disasters.

During this interactive session, participants will engage in breakout group discussions led by EW4All Pillar leads, technical speakers, and representatives from countries and governments. The session will serve as a consultation and matchmaker activity, leveraging the presence of technical experts to address practical challenges and identify solutions. Breakout groups will explore how data, platforms, institutions and partnerships can help in EW4All implementation, and further explore how AI can address these gaps or answer related questions. The discussions will also delve into the necessary AI-based investments and capabilities required for the upcoming year. Participants are encouraged to think about scenarios requiring support from the EW4All initiative in terms of needs/gaps/opportunities for Pillar 1 (disaster risk knowledge), Pillar 2 (hazard forecasting), Pillar 3 (warning dissemination and communications), and Pillar 4 (preparedness and response) and especially inter-Pillar coordination and collaboration. The discussion will focus on identifying country-specific problems and exploring how EW4All can assist in scaling solutions.

This concluding session will reflect on insights and ideas shared throughout the workshop, emphasizing the importance of collaboration and innovation in leveraging AI for early warning systems. Key outcomes from the breakout group discussions will be summarized by the Pillar leads.