Relationship of In-situ Live Fuel Moisture Measurement and MODIS data in Los Angeles County, California USA

Southern California possesses a Mediterranean climate with semi-arid to arid characteristics and contains shrublands that are naturally at a high risk to wildfire. Since beginning of the 21^st century, intensity of wildfires in California has greatly increased with respect to recorded history. Live Fuel Moisture (LFM) measures the percentage of moisture content within live shrubs and assesses the availability of fuel at risk to wildfire. Vegetation Indices (VIs) derived from remotely sensed satellite data have been applied to estimate LFM based on spectral response to changes in vegetation conditions. The in-situ LFM field data collected from chamise chaparral covers 3-5 acres with a temporal resolution of 2-weeks. In this study, LFM field data and remotely sensed satellite data within Los Angeles County, California from 2001 to 2017 were analyzed. Five VIs were calculated using the bands within the visible and infrared spectrums from MODIS collections with a 500-meter spatial resolution and an 8-day temporal resolution. These VIs, Enhanced Vegetation Index (EVI), Normalized Difference Infrared Index (NDII), Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI) and Visible Atmospherically Resistant Index (VARI), monitor the phenology of ecosystems. Linear regression was applied to calculate the coefficient of determination (R²) to assess the strength of correlation between linearly interpolated LFM data and time smoothed VI data. The VI time series were compared with the in-situ LFM time series for each site as well as with each other. Furthermore, datasets acquired on Terra and Aqua were compared as they collect data at different times of day. In summary, preliminary results showed that the strength of the performance of each VI varied site to site as well as year to year, and a single VI did not consistently dominate. In addition, inter-annual variability in weather conditions impacted the dynamics of LFM and remotely sensed VIs, as well as the strength of correlation, indicating that VIs may become less sensitive to LFM dynamics under extremely dry conditions. The goal of utilizing easily accessible remote sensing data to produce LFM estimation will benefit land management and planning by reducing time and money spent on ground data collection to monitor wildfire risk.