Are there papers, ideas or thoughts regarding the under/over estimation of PM2.5 when using NEI 2016 emissions modeling platform fire inventories (e.g. ptfire, ptagfire) instead of other inventories (e.g. FINN, GFED)?
The 2016 inventory used in EPA Modeling platforms involved a 18-month+ long collaboration with States, Tribes and MJOs. All agencies involved provided feedback and activity data (acres burned) on the ground that helped to generate the fire emissions inventory for year 2016. The other fire inventories you mention, FINN and GFED, consist of just the satellite detects and numerous assumptions (acres burned, fuel type, duration of burn). FINN and GFED do not know which fire type is occurring (wildfire or prescribed or ag fire). So as far as activity data used in emissions inventory generation the 2016 Collaboration has an incredible advantage over the FINN and GFED type guesses. There are still uncertainties with the 2016 Collaboration inventory process (e.g. emissions factor assumptions). Hope these thoughts help.
In addition, there are 3 versions of the 2016 modeling platform, each of which contains an estimate for the fire emissions. In these datasets, an attempt was made to classify fires as prescribed, wildfire, or agricultural. For each fire type, area burned, emission factors by pollutant was estimated and so each fire type was modelled differently. The FINN and GFED approaches are global in scope and do not distinguish fire types and so all fire types are modelled with similar parameters. So FINN and GFED are available globally but lack the detail that the 2016 inventory which was developed specifically for the US.
Here is a paper about the crop residue burning inventory methods that was developed for the 2014 NEI and extended to later years.
Pouliot G, Rao V, McCarty JL, Soja A. Development of the crop residue and rangeland burning in the 2014 National Emissions Inventory using information from multiple sources. Journal of the Air & Waste Management Association. 2017 Apr 27;67(5):613-22.