How does ERA5 reanalysis data deal with wildfires?

Question

I am trying to investigate the likelihood of a fire occuring in the Amazon rainforest based on environmental variables, including ERA5Land data such as temperature, wind speed and relative humidity (derived from temperature and dew point temperature). While plotting the data, I noticed that when a fire occurs, it usually coincides with a significant drop in humidity within a short time frame. Notably the humidity starts decreasing rapidly before the fire is even detected. I was therefore wondering how the data assimilation in ERA5 deals with wildfire events. More precisely, I am worried that the satelite data shows a sharp decline in relative humidity, which is considered unphysical by the ERA5 model, resulting in a smoothing of the humidity decline such that it is observable already before the fire has started.

I am sorry if this is a bit of a random question, but any help is greatly appreciated.

Answer 1

To my knowledge, only specialized studies with small domains are used to model fire behavior in meteorological numeric predictions. Climatological models would be too coarse. Aerosol effects are included via satellite data assimilation, so smoke from large wildfires influence the model runs by reducing incoming solar radiation. However, any heat or humidity changes from fire would not be included except indirectly by assimilating temperature and humidity from observations. Modeling the meteorological influences of wildfires requires very fine scales in order to properly capture the interactions with the atmosphere, for example as done in meteorological / fire behavior coupled models such as WRF-SFIRE, CAWFE, and HIGRAD-FIRETEC.