The Effects of Modern-Day Cropland and Pasture Management on Vegetation Fire: An Earth System Modeling Approach

Fire is a major component of the global carbon cycle, with some estimates of the associated emissions reaching 2.5 PgC/yr. This and the other impacts of biomass burning have driven efforts to improve its simulation in Earth system models. Recent global fire models usually include both bioclimatic and anthropogenic drivers of fire, with the latter (via population density and sometimes economic status) serving to increase or suppress burned area. Some models have added the representation of fire used in deforestation and cropland management, the extent and seasonal timing of which may not be accounted for by the usual approach to anthropogenic influence.

Human land use can also limit fire by fragmenting landscapes, but this process is not included in most global models. Moreover, although people often use fire to manage grazing lands for livestock, these practices have not been explicitly modeled (except as performed by pre-industrial societies). This could be important for regions such as sub-Saharan Africa, where the seasonality of pasture burning tends to differ from that of other lands, potentially influencing savanna-forest dynamics.

Recent efforts elucidating the effects of cropland and pasture on fire regimes at regional scales provide insight into these processes. Using this new understanding, we have developed a fire model with structurally distinct modules for burning of croplands, pasture, and primary and secondary lands, as well as fire use for deforestation. Parameters for each are rigorously constrained using remote-sensing observations of burned area. This structure allows us to disentangle agricultural practices and fragmentation effects from the endogenous processes driving fire on non-agricultural land, resulting in a better ability to simulate how fire works at large scales. This is critical for modeling the future of fire and all the parts of the Earth system that it affects, including vegetation distributions, nutrient cycling, and biosphere-atmosphere interactions.

In this work, we introduce the structure of the fire model and how it was parameterized. We then compare its performance with observations of burned area and previous global fire models, with a focus on regions where cropland or pasture predominates and especially where people use fire in their management.