Fire Dynamics Simulator & QUIC-fire simulations of the 2014 Camp Swift experiments

Background: Simpler models like BehavePlus are used in prescribed burn planning but rely on strong assumptions. There's a need to evaluate more complex, three-dimensional fire behavior models using relevant observational data from field experiments to potentially improve predictions and extend simpler models. Aims: This study conducted a basic evaluation of several multi-dimensional, time-dependent fire behavior models (FDS-PB, FDS-LS1, FDS-LS4, and QUIC-Fire) using an integrated, in space and time, field-scale observational dataset from the 2014 Camp Swift experiments. Methods: Model predictions were compared against observed head-fire arrival times, fire footprints, head-fire depths, the influence of vegetation on fire spread, fuel break breaching, and final burn area for three burn blocks. Key results: FDS-PB and FDS-LS4 generally provided the most accurate predictions of head-fire arrival times. All models overpredicted head-fire depths and failed to accurately capture the observed extinction of flank fires in camphorweed. FDS-PB was better at reproducing the breaching of the small fuel break. FDS-LS1 and QUIC-Fire significantly overpredicted head-fire arrival times. Conclusions: While some models showed promise in predicting head-fire spread, all models demonstrated limitations in capturing the complexities of wildland fire behavior, particularly the influence of vegetation on flank fire spread. Implications: This study underscores the critical and currently unmet need for comprehensive observational datasets for rigorous evaluation and improvement of wildland fire behavior models. The findings suggest that further model development is needed to accurately capture vegetation effects and other fire behavior characteristics for effective support of prescribed fire planning and other applications.