Low-Rank Adaptation of Geospatial Foundation Models for Wildfire Mapping Using Sentinel-2 Data

TL;DR

This paper demonstrates that Low-Rank Adaptation (LoRA) enables efficient and effective adaptation of geospatial foundation models for wildfire burned-area mapping using Sentinel-2 data, outperforming full fine-tuning across diverse ecological regions.

Contribution

It introduces the application of LoRA to geospatial foundation models, showing improved cross-domain generalization and efficiency for wildfire mapping tasks.

Findings

LoRA outperforms full fine-tuning in accuracy and efficiency.

Prithvi-v2 with LoRA achieves the highest accuracy.

LoRA updates less than 1% of parameters, enabling scalable adaptation.

Abstract

Wildfire burned-area mapping is essential for damage assessment, emissions modeling, and understanding fire-climate interactions across diverse ecological regions. Recent geospatial foundation models provide strong general-purpose representations for satellite imagery, yet there is still no clear understanding of how to efficiently adapt these models for downstream Earth observation tasks, particularly under geographic and temporal domain shift. This study evaluates three state-of-the-art Geospatial Foundation Models (GFMs) - Terramind, DINOv3, and Prithvi-v2 - for burned-area mapping across the United States and Canada using Sentinel-2 data. Leveraging 3,820 wildfire events from 2017-2023, we conduct spatial and temporal generalization tests across diverse biomes. We systematically compare full fine-tuning, decoder-only fine-tuning, and Low-Rank Adaptation (LoRA) for adapting each model. Across all experiments, LoRA provides the strongest cross-domain generalization while updating less than 1% of parameters, demonstrating a favorable trade-off between accuracy and efficiency. Prithvi-v2 with LoRA achieves the highest overall accuracy and the largest improvement compared to full fine-tuning. These findings indicate that geospatial foundation models, when adapted using lightweight parameter-efficient methods such as LoRA, offer a robust and scalable solution for large-scale burned-area mapping. Code is available at https://github.com/alishibli97/wildfire-lora-gfm.