Satellite and AI monitoring humanitarian crises in the Gaza Strip during the early stage of Israeli–Palestinian conflict

Abstract

Extensive power shortages and airstrikes have plunged Gaza into a severe humanitarian crisis. Here, we introduce a rapid and efficient framework to evaluate the impacts of armed conflict and monitor potential humanitarian crises in the Gaza Strip based on multi-source satellite imageries and cutting-edge artificial intelligence (AI) techniques. Since October 9th, 2023, the complete blockade had led to widespread power outages across the entire Gaza Strip, particularly in the southern region. Airstrikes destroyed more than 20% of buildings, including nearly 60% of shops and hospitals, as well as 80% of water infrastructure. The affected vulnerable population exposure is further quantified at the block level, which further helps determine high-risk humanitarian crisis locations and aid in the precise allocation of humanitarian resources. In general, the proposed framework performs well in applying to complex environments during an armed conflict. It can also offer rapid and continuous monitoring of the Gaza Strip and can be broadly applied in armed conflicts or natural disasters. Our research is one of the early studies to apply AI foundation models to improve the efficiency of rapid post-disaster investigation.

Keywords

• Rapid post-disaster response
• AI foundation model
• deep learning
• very-high-resolution satellite imagery