GUARD-net: A geostationary satellite-based network for near-real-time active fire detection under class imbalance

Date:2026-10-26 clicks:

DOI number:10.1016/j.rse.2026.115702

Affiliation of Author(s):东北林业大学

Teaching and Research Group:教育部森林生态系统可持续经营重点实验室

Journal:Remote Sensing of Environment (Q1)

Place of Publication:荷兰（Elsevier出版社总部，或填“美国/国际”）

Funded by:Grant No. 2023YFC3006900；Grant No. GZ20250112

Key Words:Wildfire detection; Geostationary satellite; Convolutional neural network; Class imbalance; Domain adaptation

Abstract:Deep learning has substantially advanced satellitebased wildfire monitoring; however, nearrealtime detection using geostationary satellite sensors remains challenged by severe class imbalance and domain shift across re￾gions and satellite platforms. This study presents GUARDNet, a geostationary satellitebased network for nearreal-time active fire detection under class imbalance that integrates spectral, spatial, and topographic informa￾tion within a convolutional framework. The model employs a bottleneck residual (BR) architecture together with focal loss (FL) to enhance sensitivity to rare fire pixels under highly imbalanced conditions. To enhance trans￾ferability, we introduce a heterogeneousfeatures weighted domain adaptation (HWDA) strategy, which aligns deep mixed features and shallow physical attributes (spectral/spatial/topographic) with a dynamic selfadaptive loss weighting scheme. Ablation experiments using Himawari-8/9 and Geostationary Operational Environmental Satellite16 (GOES16) data under controlled imbalance scenarios show consistent performance gains over baseline models. In baseline tests with a fire:non-fire ratio greater than 1:20, GUARD-Net attains F1 score > 0.94, Area Under the PrecisionRecall Curve (PRAUC) > 0.98, and Maplevel Image Classification Efficacy (MICE) > 0.93; in a controlled sensitivity analysis with a prescribed 0.1% false-positive rate among non-fire samples, the 1:1000 tests yielded F1 score > 0.77, PRAUC > 0.86, and MICE >0.71. Under the oracle checkpoint-selection protocol used for the six cross-domain transfer tasks, HWDA achieved unsupervised F1 score up to 0.955 (maintaining F1 = 0.863 with 10% unlabeled target data) and supervised F1 score up to 0.989 (reaching F1 score = 0.943 with 20% labeled target samples), with relative performance varying by transfer task and metric. Across 30 documented events from 2023, GUARD-Net detected 29 events; processing of the evalu￾ated regional scenes required 2.53.9 min on the tested workstation, while comparisons with operational fire products varied by region. The results indicate that topographic inputs, focal-loss training, and heterogeneous￾feature alignment can contribute to robustness across the evaluated terrain, classimbalance, and domainshift settings, supporting near-real-time wildfire detection from geostationary satellite observations.

Co-author:Jungho Im,Yinghui Zhao

First Author:Junjie Chen

Indexed by:Journal paper

Correspondence Author:Zhen Zhen*,Guang Yang*

Document Code:115702

Discipline:Interdisciplinary

Document Type:J

Volume:347

Page Number:115702

ISSN No.:0034-4257

Translation or Not:no

Date of Publication:2026-01-01

Included Journals:SCI

Pre One:LiFE-BA-Net: A Lightweight Network With Depth Feature Enhancement and Boundary-Aware in Fire-Front Segmentation Next One:Integrating state and process variables to assess the carbon balance of China's terrestrial ecosystems from 2014 to 2023: the CarBalTE framework