wildfire

AI4Fire 评估六大核心大模型在五项野火管理任务上的零样本表现，涵盖裸输入与任务特定 grounding 两种设置，发现 grounding 仅在补充信息直接包含答案时有效（如 SQL 工具将准确率从≤16%提升至≥88%），简单规则常胜过模型，且多个公开数据集存在标签缺陷可能虚高评分。

This paper systematically evaluates 15 machine learning models, including the TabPFN foundation model, for post-wildfire debris-flow prediction using USGS basin-scale data, finding TabPFN achieves the best performance (threat score 0.637) and that synthetic data augmentation improves most models.

U.S. senators are calling for a crackdown on wildfire prediction markets, arguing they create perverse incentives and monetize disasters. Polymarket defends its approach, while wildfire experts express disgust at the practice.

Wildfires in Spain forced the evacuation of NASA's Madrid Deep Space Network complex, temporarily halting operations and leaving only one of three 70-meter antennas operational globally, affecting spacecraft communications.

A comprehensive guide from WIRED on assembling an emergency kit for cars or go-bags, covering essential gear like water filters, flashlights, power banks, and air quality monitors to prepare for wildfires, hurricanes, earthquakes, and other disasters.

This paper systematically evaluates time series foundation models (TSFMs) for forecasting extreme PM2.5 concentrations from wildfire smoke using a 12-year dataset from California. Results show that fully-trained recurrent baselines like BiLSTM outperform TSFMs, challenging the assumption that larger pretrained models dominate in environmental forecasting.

Prosecutors used ChatGPT logs as evidence in a trial for an arson suspect, but the jury was unconvinced, leading to a mistrial.