| Gunel Aliyeva, Turgay Farhadli, Ahmad Huseynli, Sahil Guluzada A Satellite-Based Wildfire Dataset for Azerbaijan |
|---|
| Abstract. Wildfires are a growing global concern. However, the availability of high-quality, processed satellite imagery for developing detection systems remains limited, especially in under- represented areas. This study presents a novel dataset developed by integrating NASA FIRMS fire alerts with Sentinel-2 satellite imagery, covering the Azerbaijan region. The dataset is openly accessible through a dedicated website that offers tools for visual exploration, filtering, and downloading, addressing the challenge of limited wildfire imagery data. A comprehensive data extraction and processing pipeline was implemented, including cloud and noise filtering, spatial and temporal alignment with fire alerts, contrast enhancement, and manual verification. The overall approach is designed to be scalable and easily extend- able to other regions. Special emphasis was placed on smoke plume visibility and geographic representation. The dataset contributes to the ongoing efforts in wildfire detection, fire risk mitigation, and data-driven environmental policy, in response to the increasing frequency and impact of human-caused fires influenced by changing climate conditions. Images are accessible through the project website (https://wildfireaze.github.io), and the complete source code is available in the GitHub repos- itory (https://github.com/gunelaliyevaa/wildfire-detection-using-satellite-imagery). |
| Keywords: Wildfire dataset, Satellite imagery, Remote sens- ing, Sentinel-2 |
 Download PDF |
| DOI: |
