Can roads cause wildfires?

Posted by Design Studio

26 June 2024

Challenge 1: Tree Biomass

Context: Land fragmentation and its relation to fire and nature 


The breakup of the world’s land into smaller and smaller parcels of contiguous vegetation surface provides one of the clearest visual manifestations of human society’s impacts on the natural world.  Land fragmentation, as this is known, has been shown to have profound impacts on floral / faunal ecology and diversity, epidemiology, terrestrial microclimate, and human cultural diversity.  It has also been associated with a wide range of impacts on fire activity. Regional studies tend to suggest fires increase with fragmentation, while globally-aggregated analyses suggest fragmentation lowers fire activity.  Resolving the root of this apparent paradox is of practical importance because fragmentation is forecast to increase substantially by the end of the century, as are climatic conditions favouring increased fire activity -higher temperatures, aridity, wind gusting and gross primary productivity. Understanding and predicting where, when, why and by how much increases in fragmentation may alter fire propensity and behaviour with changing climatic conditions puts the assessment and planning of present and future fragmentation, respectively, within range of active ‘risk management’.  


Method: Representing fragmentation in a global intermediate-resolution model 

This study represents a first step in this direction.  Here, using road density as a proxy for fragmentation, we hypothesise that road-driven increases in fire activity can be captured by its measured microclimatic effects on vegetation patch edges -increased temperature, fuel desiccation, exposure to human activity and wind infiltration.  Fire limiting impacts of fragmentation are hypothesised as being the edge/road-induced barrier to fire spread, limiting individual fire sizes to vegetation patch size unless conditions for extreme fires are met.   These effects are represented at model sub-grid scales by transforming gridded road length estimates to circles of equal area, whose circumferences sum to that grid cells road length, and whose area is equal to that of the grid cell.


Result & Conclusion: Fragmentation has a globally small but regionally significant impact on fire

Implementing and running these at 0.5 degree grid resolution in the ORCHIDEE land surface model, we find that roads have a small negative impact on aggregate fire activity, but increase fires in tropical and other forest ecosystems. This is particularly true of areas with low to moderate population density, high fragmentation and pronounced dry season aridity. Fragmentation has a significant impact on proportional changes to fire activity in many of these forest ecosystems (>25% change in 7% of burned grid cells), indicating biome-specific fragilities.  Our results are spatially and proportionally consistent with both regional and global scale fire-fragmentation statistical analyses, paving the way for predictive modelling of its future impacts. Land fragmentation imposes potentially large and deleterious consequences for fire risk and fire regime stability, impacting ecosystem and carbon dynamics from regional to global scales.  This work provides a first globally-consistent framework for evaluating that risk in a quantifiable manner.  Extensions of this may look at the long term changes to fire return interval due to fragmentation, future fragmentation-fire impacts, and the net biome productivity consequences of prescribed burning in fragment edge ecosystems. 

Simulated fractional changes in burned area due to fragmentation in northern S. America (f), colour-bar), overlaid with BA and fragmentation-proxy trend data from Rosan et al. (2022)118, which were aggregated for Brazil’s Amazonian (circular points) and Cerrado regions (triangles), as comparison. Where both BA and fragmentation increased (+BA/+frag) over 2003-2018, points are coloured red and [(+BA/-frag)=orange; (-BA/+frag)=light blue; (-BA/=frag)=dark blue].

Read the publication in full: