Machine Learning Historical and Forecasted Gridded Live Fuel Moisture Content

Live fuel moisture content plays a significant and complex role in wildfire propagation. However, in situ historical and near real-time live fuel moisture measurements are temporally and spatially sparse within wildfire-prone regions. Routine bi-weekly sampling intervals are sometimes skipped if the weather is unfavourable and/or field personnel are unavailable. To fill these spatial and temporal gaps, we have developed daily gridded live fuel moisture products for various vegetation species like Chamise, Ceanothus, Manzanita and Sage that can be used, in conjunction with other predictors, to assess current and historical wildfire potential. LFM observations from the National Fuel Moisture Database were statistically related to dynamically downscaled high-resolution weather predictors using a random forest machine learning model. This model captures reasonably well the temporal and spatial variability of live fuel moisture content within California for each species. Over the past several years, we have built a multidecadal high-resolution LFM data set and an operational LFM forecast framework. The historical continuous gridded data set can be used to investigate past wildfire behavior as well place LFM forecasts into historical perspective. We will discuss LFM modeling techniques, validation efforts, and operational analytics used by agencies to help prevent catastrophic wildfire ignitions.