Remote characterization of fire behavior during the FireFlux II experiment

The FireFlux II field experiment was conducted on January 30th, 2013 in south-east Texas, USA, under high fire danger conditions. The experiment was designed to study the behavior of a head fire progressing through a flat, tall grass prairie, and it was informed by the use of a coupled fire-atmosphere model. Vegetation properties and fuel moisture were measured shortly before the experiment. Near-surface atmospheric conditions were monitored during the experiment using an elaborate meteorological instrumentation array. Fire behavior was observed through a combination of remote and in-situ sensors. Clements et al. (2019) presented the analysis of the experiment micrometeorology and in-situ fire behavior observations acquired using a thermocouple array. In this paper, we extend the study of fire behavior during the FireFlux II experiment with the analysis of remote sensing observations. Two thermal infrared and two visible cameras were deployed during the experiment. One thermal and one visible camera were mounted on a helicopter, whereas the other two cameras were installed on a 40-m-height tower next to the burn unit. The tower infrared camera covered a reduced area of interest coincident with the thermocouple array and it allowed monitoring the fire spread as well as measuring the spatially-resolved evolution of brightness temperature. Imagery collected from the helicopter allowed extending fire behavior measurements to the complete burn unit. While airborne IR footage was saturated and did not allow estimation of emitted radiant heat, its analysis allowed tracking fire progression through the plot and therefore estimating rate of spread and fire time of arrival. The existence of in-situ temperature observations provides an outstanding opportunity to validate remote sensing methodologies. In addition, the combination of remote observations with in-situ temperature measurements allows a comprehensive characterization of fire behavior, including spatially-resolved fire rate of spread and fire time of arrival, fire radiative power, Byram’s fire line intensity, and air temperature during fire front passage. The current version of this paper presents preliminary results from this analysis. Such results demonstrate the usefulness of the selected datasets and the potential of the proposed methodology, encouraging further work. Possible applications of the resulting dataset include (i) the validation of existing fire behavior models that are able to predict any of the measured variables, (ii) the development of data-driven fire behavior models, and (iii) the investigation of the relative contribution of radiative and convective heat transfer mechanisms to fire spread.