Near-real time fires detection using satellite imagery in Sudan conflict

TL;DR

This study demonstrates near-real-time fire and burn-scar detection in the Sudan conflict using 4-band PlanetScope imagery and an unsupervised RaVAEn change-detection model. It adapts RaVAEn to four channels and validates it across five incidents, outperforming a pixel-based baseline, with 4-band inputs providing most gains and only marginal improvements from 8-band or time-series data. The work highlights practical considerations like spectral limits of commercial imagery and advocates interdisciplinary efforts to incorporate uncertainty and verification for conflict monitoring. Overall, the approach offers a scalable, rapid tool for conflict assessment and humanitarian response, complemented by recommendations for future uncertainty-aware data fusion and collaboration across remote sensing, peace research, and humanitarian communities.

Abstract

The challenges of ongoing war in Sudan highlight the need for rapid monitoring and analysis of such conflicts. Advances in deep learning and readily available satellite remote sensing imagery allow for near real-time monitoring. This paper uses 4-band imagery from Planet Labs with a deep learning model to show that fire damage in armed conflicts can be monitored with minimal delay. We demonstrate the effectiveness of our approach using five case studies in Sudan. We show that, compared to a baseline, the automated method captures the active fires and charred areas more accurately. Our results indicate that using 8-band imagery or time series of such imagery only result in marginal gains.

Figures (9)

• Figure 1: Map of attacks in Sudan.
• Figure 2: The depiction of study areas used for this research.
• Figure 3: Comparison of baseline with prediction in Jaranga.
• Figure 4: Comparison of baseline with prediction in El Fasher.
• Figure 5: Comparison of before and after input imagery with an active fire prediction in Gandahar Market with a smoke plume.