Hydrological predictors of fire danger: using satellite observations for monthly to seasonal forecasting

Catastrophic wild fires are caused by many contributing factors. While some factors such as ignition or weather are challenging to predict with long-lead time, other factors such as fuel accumulation and fuel moisture tend to demonstrate sufficient persistence that they can be used as early fire danger predictors. We now have several satellite observations that contribute to estimating the state of the terrestrial hydrological cycle. In this talk, we present results from a body of research work on the relationship between hydrological proxies for fuel moisture and fuel accumulation, and fire burned area at several months lead time. First we demonstrate the statistical relationship between soil moisture and burned area across the continental United States. Then we describe the development of a multi-variate prediction system with burned area forecasting skill at the GACC level in the US. Finally, we present the development of a spatially distributed gridded fire danger forecast at 2-3 months lead time. The data sets used to determine these hydrology-fire relationships and their predictive capability include the FPA Fire Occurrence Data set and GFED burned area data products, as well as model-assimilated GRACE soil moisture, AIRS vapor pressure deficit and MODIS EVI data products.