Abstract Patterns of energy and available moisture can vary over small (<1 km) distances in mountainous terrain. Information on fuel and soil moisture conditions that resolves this variation could help to inform fire and drought management decisions. Here, we describe the development of TOPOFIRE, a web-based mapping system designed to provide finely resolved information on soil water balance, drought, and wildfire danger information for the contiguous United States. We developed 8-arc-second-resolution (~250 meter) daily historical, near real-time, and 4-day forecast radiation, temperature, humidity, and snow water equivalent data and used these grids to calculate a suite of drought and wildfire danger indices. Large differences in shortwave radiation and surface air temperature with aspect contribute to greater snow accumulation and delays in melt timing on north-facing slopes, delaying fuel conditioning on shaded slopes. These datasets will help advance our understanding of the role of topography in wildland fire spread and ecological effects. Integration with national programs like the Wildland Fire Assessment System, the Wildland Fire Decision Support System, and drought early warning systems could support more proactive management of wildland fires and refine the characterization of drought in mountainous regions of the United States.
more »
« less
Predicting Toxic Gas Levels in Structure Fires Using Machine Learning
This study presents a machine learning framework for real-time hazard assessment in fire environments using gas sensor data. A two-stage predictive model was developed with simulated data from PyroSim, a fire dynamics tool. The regression model predicts gas concentrations throughout a burning structure, while the classification model determines whether the levels are safe or pose an immediate danger to life and health (IDLH). To enhance performance, lag features captured short-term temporal dynamics, and the Synthetic Minority Over-sampling Technique (SMOTE) addressed class imbalance, improving the detection of rare but critical hazardous conditions. This approach significantly increased the model's accuracy in identifying immediate danger events. Designed for real-time deployment, the system processes sensor input to provide instant hazard classification. Preliminary results indicate that the models accurately predict and classify hazardous conditions with high confidence, enabling real-time use in autonomous systems. Future efforts will focus on sensor calibration, system integration, and field validation to advance automated risk assessment and improve responder safety in fire emergencies.
more »
« less
- Award ID(s):
- 2447856
- PAR ID:
- 10672985
- Publisher / Repository:
- IEEE
- Date Published:
- ISBN:
- 979-8-3315-6501-5
- Page Range / eLocation ID:
- 0683 to 0690
- Subject(s) / Keyword(s):
- Machine Learning Regression Classifier Simulation Firefighter Gases Sensors
- Location:
- Yorktown Heights, NY, USA
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Benim, Ali Cemal (Ed.)Accurate quantification of wildfire risk is essential for balancing wildfire mitigation and prescribed fire management in grassland ecosystems, yet existing fire danger indices do not explicitly distinguish seasonal fuel dynamics from daily weather variability. This study presents a hierarchical framework for quantifying wildland fire risk by explicitly separating seasonal wildfire potential from daily weather-driven fire activity. The framework introduces the Daily Burned Area Ratio (DBAR) as a quantitative measure of realized wildfire risk and decomposes it into the Seasonal Burned Area Ratio (SBAR) and the Daily Burn Activity Index (DBAI). Wildfire records from the U.S. Forest Service Fire Program Analysis Fire-Occurrence Database, Oklahoma Mesonet weather observations, and remotely sensed vegetation data collected between 1995 and 2020 were used to develop and evaluate the framework in the Flint Hills of Kansas and Oklahoma. SBAR was modeled using grass curing and air temperature to characterize the seasonal baseline of wildfire activity, whereas DBAI was modeled using dead fuel moisture content (DFMC) and wind speed to quantify day-to-day departures from that seasonal baseline. The SBAR model accurately reproduced the characteristic bimodal wildfire regime of the Great Plains, whereas the DBAI model identified DFMC as the dominant control on daily wildfire activity, with wind speed providing an important secondary influence. Compared with the Burning Index (BI) and the Grassland Fire Danger Index (GFDI), the hierarchical framework achieved superior performance in discriminating fire days from non-fire days. Global sensitivity analysis further demonstrated that the framework provides a more balanced representation of the influences of grass curing, relative humidity, air temperature, and wind speed than the conventional indices. By explicitly separating seasonal fuel dynamics from short-term weather variability, the proposed framework provides an ecologically interpretable, locally calibratable, and operationally practical approach to wildfire risk assessment. Because the seasonal and daily components can be calibrated independently, the framework is readily transferable to other grassland ecosystems and provides a flexible foundation for adaptive wildfire and prescribed fire management under changing climatic conditions.more » « less
-
Modern safety-critical systems increasingly rely on human-robot interaction to reduce disaster risk and support decision-making during emergencies. Vision-Language Models (VLMs) are promising for these settings because they can interpret complex scenes and communicate safety-relevant information, but they still require careful evaluation to ensure reliable safety reasoning. In particular, current evaluations often frame danger recognition as a binary decision (Safe/Unsafe), making it unclear whether a model is identifying true physical hazards or merely reacting to unusual scene elements. We address this limitation by introducing an explicit distinction between hazard and anomaly, and by separately recognizing hazardous and anomalous states. We evaluate several state-of-the-art VLMs across two datasets and multiple prompting strategies to test whether this distinction changes model behavior. Our results show that VLMs frequently misinterpret anomalousness as hazardousness, revealing an over-reliance on contextual irregularity as a proxy for danger. We further show that explicitly separating anomaly from hazard provides a more informative evaluation of VLM safety reasoning and exposes failure modes that binary safety judgments can obscure. Our public dataset is available on Roboflow: https://app.roboflow.com/vlm-in-context-anomaly-and-hazard-detection/camera-ready-roman-ds.more » « less
-
Modern safety-critical systems increasingly rely on human–robot interaction to reduce disaster risk and support decision-making during emergencies. Vision–Language Models (VLMs) are promising for these settings because they can interpret complex scenes and communicate safety-relevant information, but they still require careful evaluation to ensure reliable safety reasoning. In particular, current evaluations often frame danger recognition as a binary decision (Safe/Unsafe), making it unclear whether a model is identifying true physical hazards or merely reacting to unusual scene elements. We address this limitation by introducing an explicit distinction between hazard and anomaly, and by separately recognizing hazardous and anomalous states. We evaluate several state-of-the-art VLMs across two datasets and multiple prompting strategies to test whether this distinction changes model behavior. Our results show that VLMs frequently misinterpret anomalousness as hazardousness, revealing an over-reliance on contextual irregularity as a proxy for danger. We further show that explicitly separating anomaly from hazard provides a more informative evaluation of VLM safety reasoning and exposes failure modes that binary safety judgments can obscure.more » « less
-
Skateboarding as a method of transportation has become prevalent, which has increased the occurrence and likelihood of pedestrian–skateboarder collisions and near-collision scenarios in shared-use roadway areas. Collisions between pedestrians and skateboarders can result in significant injury. New approaches are needed to evaluate shared-use areas prone to hazardous pedestrian–skateboarder interactions, and perform real-time, in situ (e.g., on-device) predictions of pedestrian–skateboarder collisions as road conditions vary due to changes in land usage and construction. A mechanism called the Surrogate Safety Measures for skateboarder–pedestrian interaction can be computed to evaluate high-risk conditions on roads and sidewalks using deep learning object detection models. In this paper, we present the first ever skateboarder–pedestrian safety study leveraging deep learning architectures. We view and analyze state of the art deep learning architectures, namely the Faster R-CNN and two variants of the Single Shot Multi-box Detector (SSD) model to select the correct model that best suits two different tasks: automated calculation of Post Encroachment Time (PET) and finding hazardous conflict zones in real-time. We also contribute a new annotated data set that contains skateboarder–pedestrian interactions that has been collected for this study. Both our selected models can detect and classify pedestrians and skateboarders correctly and efficiently. However, due to differences in their architectures and based on the advantages and disadvantages of each model, both models were individually used to perform two different set of tasks. Due to improved accuracy, the Faster R-CNN model was used to automate the calculation of post encroachment time, whereas to determine hazardous regions in real-time, due to its extremely fast inference rate, the Single Shot Multibox MobileNet V1 model was used. An outcome of this work is a model that can be deployed on low-cost, small-footprint mobile and IoT devices at traffic intersections with existing cameras to perform on-device inferencing for in situ Surrogate Safety Measurement (SSM), such as Time-To-Collision (TTC) and Post Encroachment Time (PET). SSM values that exceed a hazard threshold can be published to an Message Queuing Telemetry Transport (MQTT) broker, where messages are received by an intersection traffic signal controller for real-time signal adjustment, thus contributing to state-of-the-art vehicle and pedestrian safety at hazard-prone intersections.more » « less
An official website of the United States government

