Artificial Intelligence (AI) in Disaster Management

Syllabus: GS3/ Disaster Management

Context

• Recent disaster response initiatives in Nepal have highlighted the potential of Artificial Intelligence (AI) and crowdsourced information to enhance the speed and accuracy of disaster management.

Role of AI in Managing Disaster 

• Early warning and prediction: AI is able to process huge datasets of historical weather records and current meteorological observations to provide forecasts that are quicker and more specific to a particular location.
  • Google Flood Hub utilises weather and hydrological information to provide advance information on flood hazards.
• Hazard Mapping: AI is able to integrate satellite imagery, geographic data, meteorological information, maps and administrative databases to identify areas at risk from certain hazards.
• Response, Relief and Rescue: Disasters create an overwhelming amount of information from smartphones, social media, emergency calls, drones and government databases.
  • AI is able to manage this unstructured data and discover useful patterns far quicker than manual systems.
  • In the Nepal disaster, an AI-powered webpage helped link crowdsourced information on missing persons with official records of people killed or injured.
  • Thermal imaging drones can locate human heat signatures under rubble and guide rescue crews to probable victims.
• Post-Disaster Recovery: AI and satellite-based techniques can help detect isolated villages, destroyed infrastructure, blocked highways and suitable areas for emergency helicopter operations.
  • Authorities can use this information to prioritise the distribution of food, medicines, shelter and other relief goods.

Key Artificial Intelligence (AI) Applications

• GraphCast: An AI-driven weather prediction system that can quickly produce worldwide weather forecasts.
• Google Flood Hub: AI-powered flood forecasting and early-warning information.
• DisasterAWARE: Uses geographic and disaster information to enable hazard monitoring and risk assessment.
• SKAI: Uses satellite photography and artificial intelligence to help rapidly analyse disaster damage.

What are the Challenges? 

• AI models depend on large enough datasets that are accurate and dependable, which may not be available during disasters.
• Communities with limited access to smartphones, internet connectivity or digital services may be under-represented in AI-generated assessments.
• Social media and crowd-sourced platforms can generate significant amounts of erroneous, duplicate or misleading information.
• Artificial intelligence systems taught on one geographical area may not function similarly well in another area with varied terrain, climate and settlement trends.
• Information concerning missing persons, locations and groups affected by the incident raises questions about data protection.
• Mistakes made by AI may be serious if the results are utilised to make evacuation, rescue or relief decisions.

Way Ahead

• Integrate AI with existing disaster-management institutions rather than treating it as a substitute for human decision-making.
• Build India-centric hyperlocal datasets on weather, terrain, infrastructure, vulnerability and past disasters.
• Create automated systems in regional and local languages to facilitate accessibility during emergencies.
• Strengthen interoperability among IMD, ISRO, NDMA, state disaster-management authorities, telecom providers and local administrations.

Source: IE