Search

Real-time forecasting of travel demand during wildfire evacuations is crucial for emergency managers and transportation planners to make timely and better-informed decisions. However, few studies focus on accurate travel demand forecasting in large-scale emergency evacuations. Therefore, this study develops and tests a new methodological framework for modeling trip generation in wildfire evacuations by using (a) large-scale GPS data generated by mobile devices and (b) state-of-the-art AI technologies. The proposed methodology aims at forecasting evacuation trips and other types of trips. Based on the travel demand inferred from the GPS data, we develop a new deep learning model, i.e., Situational-Aware Multi-Graph Convolutional Recurrent Network (SA-MGCRN), along with a model updating scheme to achieve real-time forecasting of travel demand during wildfire evacuations. The proposed methodological framework is tested in this study for a real-world case study: the 2019 Kincade Fire in Sonoma County, CA. The results show that SA-MGCRN significantly outperforms all the selected state-of-the-art benchmarks in terms of prediction performance. Our finding suggests that the most important model components of SA-MGCRN are evacuation order/warning information, proximity to fire, and population change, which are consistent with behavioral theories and empirical findings.

This report presents a list of case studies concerning large outdoor fires involving evacuation. The report has been developed within the activities of the Emergency Management & Evacuation (EME) Subgroup of the Large Outdoor Fires & the Built Environment (LOF&BE) group of the International Association for Fire Safety Science (IAFSS). This work is deemed to be a useful starting point to build a large database of case studies and identify common trends and differences across such type of incidents. The case studies presented are useful for the identification of critical issues related to evacuation scenarios due to large outdoor fires and could be used to identify lessons learnt based on the events. Spatial and temporal scales of past events can be reviewed and an analysis of their consequences could inform policy makers in further developing safety guidance and recommendations for emergency management and evacuation planning., This report is prepared as a part of activities by the Emergency Management and Evacuation (EME) subgroup, Large Outdoor Fires & the Built Environment (LOF&BE) working group, (the International Association for Fire Safety Science (IAFSS) permanent working group). This is the Part 1 of "Case studies of large outdoor fires involving evacuations". Part 2 is here 10.5281/zenodo.6544760

Reliable empirical data are required to understand and model realistic pedestrian navigation behaviours at public crowd-gathering places such as public transportation hubs and shopping malls. Naturalistic walking big data could provide reliable information for investigating realistic pedestrian behaviours and to overcome critical shortcomings of the data collected through controlled laboratory experiments. In this work, we investigate pedestrian navigation in indoor open spaces using naturalistic walking big data collected through video recordings. The extracted data include 299,082 trajectories of individual pedestrians who navigated through the atrium of the Informatics Forum building of the University of Edinburgh. We compare several pedestrian vector fields by calibrating several cellular automaton (CA) models and we finally identify a generalized vector field for pedestrians who are walking in indoor open space environments under normal walking conditions. The output of this study could be useful in enhancing CA-based pedestrian simulation models by representing pedestrian navigation as well as route-choice behaviours more realistically in those models. Simulation tools based on such enhanced models can facilitate practitioners, such as public building designers, to optimize designs considering naturalistic pedestrian behaviours in open spaces.

Perivascular Spaces (PVS) are a recently recognised feature of Small Vessel Disease (SVD), also indicating neuroinflammation, and are an important part of the brain's circulation and glymphatic drainage system. Quantitative analysis of PVS on Magnetic Resonance Images (MRI) is important for understanding their relationship with neurological diseases. In this work, we propose a segmentation technique based on the 3D Frangi filtering for extraction of PVS from MRI. Based on prior knowledge from neuroradiological ratings of PVS, we used ordered logit models to optimise Frangi filter parameters in response to the variability in the scanner's parameters and study protocols. We optimized and validated our proposed models on two independent cohorts, a dementia sample (N=20) and patients who previously had mild to moderate stroke (N=48). Results demonstrate the robustness and generalisability of our segmentation method. Segmentation-based PVS burden estimates correlated with neuroradiological assessments (Spearman's $\rho$ = 0.74, p $

Perivascular Spaces (PVS) are a feature of Small Vessel Disease (SVD), and are an important part of the brain’s circulation and glymphatic drainage system. Quantitative analysis of PVS on Magnetic Resonance Images (MRI) is important for understanding their relationship with neurological diseases. In this work, we propose a segmentation technique based on the 3D Frangi fltering for extraction of PVS from MRI. We used ordered logit models and visual rating scales as alternative ground truth for Frangi flter parameter optimization and evaluation. We optimized and validated our proposed models on two independent cohorts, a dementia sample (N=20) and patients who previously had mild to moderate stroke (N=48). Results demonstrate the robustness and generalisability of our segmentation method. Segmentation-based PVS burden estimates correlated well with neuroradiological assessments (Spearman’s ρ=0.74, p