If you are analyzing biomass burning you might consider using burned area rather than active fire data as active fire data only gives you a partial picture.
[from the FIRMS FAQ https://earthdata.nasa.gov/faq/firms-faq#ed-active-fire-burnedarea: "It is not recommended to use active fire locations to estimate burned area due to spatial and temporal sampling issues. Determining this to an acceptable degree of accuracy is generally not possible due to nontrivial spatial and temporal sampling issues. For some applications, however, acceptable accuracy can be achieved, although the effective area burned per fire pixel is not simply a constant, but rather varies with respect to several different vegetation and fire-related variables. See Giglio et al. (2006) for more information"
There is a global MODIS burned area (MCD64A1) product which is available at: https://modis-fire.umd.edu/ba.html
Please note, this method may underestimate burned area if there is persistent cloud during the fire season.
If you don't need high spatial resolution then you could alternatively use the GFED4 global burned area data set (0.25-degree spatial resolution, daily + monthly temporal resolution), which is derived exclusively from MCD64A1 from ~August 2000 onward. To find out more about GFED4 go to: https://globalfiredata.org/pages/data/