A methodology for Hazard Extraction and Analysis of Trends (HEAT) is proposed and conducted on a data set of wildfire incident response forms, known as ICS-209-PLUS. The HEAT processes: (1) extract a set of hazards from a data set, (2) calculate hazard-relevant metrics in a primary analysis, (3) analyze trends over time in metrics using time series, and (4) examine potential explanations for metric trends using a secondary analysis. Hazards are extracted from narrative data in the ICS-209-PLUS based on a framework previously developed by the authors, using natural language processing. Metrics examined for each hazard include operational time to occurrence, rate of occurrence, frequency, and severity. Primary results include a taxonomy of hazards present in the data set with relevant quantitative metrics. The most frequent hazards identified are environmental and include hazardous terrain. Most hazards occur on average between 35-55% containment. Incidents with hazards tend to have a higher average severity score when compared to the average score for all incidents. Time series of the metrics and relevant predictors, including fire characteristics, fire intensity, and operations, are created to facilitate further analysis. Secondary results used to determine which factors best predict hazard frequency include a correlation matrix and regression analysis. These findings are relevant to safety for current, as well as emerging wildfire operations, and are an exploratory first step in developing historical data-driven risk assessment models.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Wildfire Emergency Response Hazard Extraction and Analysis of Trends (HEAT) through Natural Language Processing and Time Series


    Beteiligte:


    Erscheinungsdatum :

    2021-10-03


    Format / Umfang :

    1822964 byte





    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch