Mapping Surface Vapor Pressure Deficits From Geostationary Satellites for Fire Weather Monitoring

The increase in the wildfires were observed globally in accordance with global warming, and to real-
time monitoring of wildfire risk in broad scale is demanded for wildfire management to prevent the
spread of wildfires. Scientists invented a lot of indices to assess the wildfire risk. Vapor Pressure
Deficit (VPD) is one of the most important meteorological components for those indices. Compared
to other components of fire weather indices, VPD can change quickly from lower risk to higher risk
even in sub-hourly. Therefore, real-time fire risk monitoring requires high-resolution and high-
temporal VPD spatial map. Here, we developed VPD estimation method using the GOES Advanced
Baseline Imager (ABI) data. Unlike the polar-orbital satellite data, the ABI can observe target region
every 10 minutes, so that we can estimate VPD for fire weather in real-time. The method used to
estimate VPD is same with the algorithm of NASA Earth Exchange Gridded Daily Meteorology (NEX-
GDM), which estimate meteorological variables from ground weather observation and spatial
variables based on random forest (RF). We calculated RF importance to select bands of ABI as input
of the model. To validate our results, we compared the spatial pattern of our VPD data with the Real-
Time Mesoscale Analysis (RTMA) data over the conterminous USA. We sought possibility of
applying our method to the region where no real-time high-resolution weather data is available, such
as South America. The developed method can produce real-time high-resolution high-frequent VPD
data in the continental scale. The derived data from GOES ABI could contribute to improve the fire
weather monitoring and lead to prevent wildfires.