Validation of fuel moisture content estimates from the Australian Flammability Monitoring Systems for coastal shrublands in the Perth region

The Yanchep Bushfire, which burnt around 12 300 ha in December 2019, was triggered by an extreme heatwave event. The burnt area included structures, suburbs, and reserves, with an extensive native vegetation component. The site has very heterogeneous vegetation and incorporates threatened species. Since vegetation moisture is a crucial factor in determining fire behavior, several remote sensing methods have been developed to calculate live fuel moisture content (LFMC) to replace costly and time consuming field based methods. In previous studies, we developed a method to estimate LFMC from coarse resolution (500m) Moderate Resolution Imaging Spectroradiometer (MODIS) data. However, higher spatial resolution satellite sensors can detect LFMC changes in smaller scales which are essential for fire management applications. Therefore, we adapted previous methodology for estimating LFMC using Sentinel 2 imagery to produce LFMC maps at 20m spatial resolution.

In this research, we used field measured LFMC of dominant plant species on three sites in Yanchep to validate Sentinel 2 based LFMC. Spatial and temporal collocations were performed to identify closest matches of observed and estimated LFMC values. Collocated pairs of measurements were analyzed statistically to determine the relationship between the observed and estimated LFMC and with that the performance of the Sentinel-2 LFMC estimates.

Dominant plant species individually showed weaker response to Sentinel-2 based LFMC estimates comparing to the averages of all vegetation species. Moreover, the weighted average of dominant species (by land cover percentage) indicated the highest correlation with estimated LFMC. This demonstrated good performance of the LFMC Sentinel-2 algorithm given the limitations of the field data used as ground truth.  

This research will serve as a basis for validating Sentinel 2 based LFMC with ground measurements to better understand uncertainties in both LFMC measured on the ground and retrieve by the satellite. Remote sensing based LFMC maps are an invaluable resource for bushfire risk assessment and prescribed burn activities.