An Investigation into Fuel Moisture Content: Dead or Alive
Kevin Varga, UCSB, kvarga@ucsb.edu (Presenter)
Charles Jones, UCSB, cjones@eri.ucsb.edu
Understanding wildfire dynamics requires comprehension of meteorology, climate, ecology, combustion, and complex topography. The interactions between these factors alter the amount of water within vegetation, also known as fuel moisture content (FMC), thus affecting the flammability. Better prediction of FMC can help communities increase their resilience and can help wildfire behavior analysts model fire spread. In this study, we create a machine learning model to predict live FMC. Our predictors include meteorological outputs from a 32-year Weather Research and Forecasting (WRF) Model climatology, Landsat observations, and static topography data. Our predictands consist of ten thousand in-situ FMC observations, spanning eight chaparral species, from the National Fuel Moisture Database. Lag correlation analysis is performed to determine the strongest relationship between predictors and predictands before running the random forest model. Dead FMC is being calculated using semi-empirical equations adapted from the Nelson dead fuel model. After successful live and dead FMC models are created, a historical, gridded dataset of FMC will be constructed. FMC variations will then be connected with different weather and climate events, as well as different wildfire behavior case studies. This high resolution modeling of FMC can also be used to better inform resilience efforts in the region of interest, such as Santa Barbara County's Regional Wildfire Mitigation Program (RWMP).
Poster: Poster_Varga_3-31_61_35.pdf
Associated Project(s):
Poster Location ID: 3-31
Presentation Type: Poster
Session: Poster Session 3
Session Date: Thu (May 11) 3:00-5:00 PM
CCE Program: TE