模块化深度学习机制用于可审计的次日野火蔓延预测

中文摘要 AI 辅助

次日野火预测要求模型的预测结果能够与其计算中所使用的假设和历史证据一同被评估。尽管深度学习能够从遥感数据中学习空间模式，但仅凭预测性能并不能确立物理保真度或运行可信度。本研究探讨了三种用于次日活跃火灾预测的模块化增强方法：基于风和坡度的条件注意力偏置、基于物理特征检索增强的输出校正，以及火灾条件化的双流门控。注意力偏置暴露了预设的方向偏好，而检索模块使用九维环境和火灾状态描述符选择历史图块，并对冻结模型的逻辑值应用学习到的校正。这些模块在Next Day Wildfire Spread基准上，使用五个骨干网络进行评估，并采用了分阶段消融、方向审计、检索扰动、校准度量和计算比较。带有全部三种增强的SwinUNETR模型的三次随机种子平均F1分数和精确率-召回率曲线下面积（AUC-PR）分别为0.4216和0.3673。随后，一个混合集成（两个增强架构和一个非增强架构）模型达到了0.4292和0.3790。不同架构的收益各异，且检索相关的AUC-PR提升并不总能转化为更高的F1分数。构建的风偏置与输入风高度一致，但其与观测到的次日火灾位移的一致性要弱得多，这区分了先验可检查性与预测物理保真度。本研究贡献了一个框架，用于在野火预测模型中暴露和评估选定的领域信息组件。综合来看，所呈现的结果表明，预测性能、运行可信度和计算实用性不必是相互竞争的目标。

英文摘要

Next-day wildfire prediction requires models whose forecasts can be evaluated alongside the assumptions and historical evidence used in their computation. Although deep learning can learn spatial patterns from remote-sensing data, predictive performance alone does not establish physical fidelity or operational trustworthiness. This study investigates three modular augmentations for next-day active-fire prediction: wind- and slope-conditioned attention biases, physics-feature retrieval-augmented output correction, and fire conditioned dual-stream gating. The attention biases expose prescribed directional preferences, while the retrieval module selects historical tiles using a nine-dimensional environmental and fire-state descriptor and applies a learned correction to a frozen model's logits. The modules are evaluated across five backbones on the Next Day Wildfire Spread benchmark, using staged ablations, directional audits, retrieval perturbations, calibration measures, and computational comparisons. The three-seed mean F1 score and area under the precision--recall curve (AUC-PR) of a SwinUNETR model with all three augmentations are 0.4216 and 0.3673. Then, a mixed ensemble (two augmented architectures and one non-augmented architecture) model achieves 0.4292 and 0.3790. Benefits vary across architectures, and retrieval-related improvements in AUC-PR do not consistently translate into higher F1. The constructed wind bias aligns closely with input wind, but its alignment with observed next-day fire displacement is much weaker, distinguishing prior inspectability from predictive physical fidelity. The study contributes a framework for exposing and evaluating selected domain-informed components within wildfire prediction models. Together, the results presented show that predictive performance, operational trustworthiness, and computational practicality need not be competing objectives.

发表机构

• Texas A&M University（德克萨斯A&M大学）
• University of Maryland, Baltimore County（马里兰大学巴尔的摩县分校）
• American University（美利坚大学）

机构由 AI 辅助整理，请以论文原文为准。