Search

Background Trends of increasing area burned in many regions worldwide are leading to more locations experiencing short-interval reburns (i.e., fires occurring two or more times in the same place within 1–3 decades). Field and satellite indices of burn severity are well tested in forests experiencing a single recent fire, but the reliability of these indices in short-interval reburns is poorly understood. We tested how a commonly used field index (the Composite Burn Index, CBI) and satellite index (the Relative differenced Normalized Burn Ratio, RdNBR) compared to eight individual field measures of burn severity in short-interval reburns vs. areas burned in one recent fire, and whether results depended on whether the first fire was stand replacing (fire that is lethal to most dominant trees). Results Correspondence between both CBI and RdNBR with individual burn severity measures differed in short-interval reburns compared to single fires for some metrics of burn severity. Divergence in the relationship between both CBI and RdNBR vs. field measures was greatest when short-interval reburns followed a prior stand-replacing fire, and measures were more comparable to single fires when the first fire was non-stand replacing (i.e., lower severity). When short-interval reburns followed prior stand-replacing fires, CBI and RdNBR underestimated burn severity in the second fire for tree-canopy metrics (e.g., canopy cover loss, tree mortality), as young forests in early developmental stages are more sensitive to a second fire. Conversely, when short-interval reburns followed prior less-than-stand-replacing fires, both CBI and RdNBR overestimated burn severity for forest-floor metrics, as past low severity fires leave behind live fire-resistant trees and can stimulate resprouting understory vegetation. Finally, neither CBI nor RdNBR accurately detected deep wood charring—an important phenomenon that occurs in short-interval reburns. Conclusion Our findings inform interpretability of commonly used indices of burn severity in short-interval reburns by identifying how individual burn severity metrics can be under- or over-estimated, depending on the severity of the fire preceding a reburn. Adjustments to burn severity measurements made in short-interval reburns are particularly critical as reburned areas increase.

Burn severity in forests is commonly assessed in the field with visual ordinal estimates such as the Composite Burn Index (CBI). However, how CBI (a composite of several individual field measures) relates to independent quantitative measures of burn severity (e.g. fire-caused tree mortality, surface charring) has not been widely tested. Here, we use field data from 315 plots in 14 fires in the north-western USA to ask: (1) how CBI relates to eight independent field measures of burn severity; and (2) how these relationships vary across gradients of pre-fire forest structure. Overall, CBI corresponded well with most independent field measures, but some measures of extreme burn severity (e.g. deep charring on trees and snags) were not captured by CBI. Additionally, some measures of canopy burn severity corresponded to lower CBI values in forests with larger average tree size (diameter and height) – potentially from decoupling of surface and canopy fire effects in stands with larger, fire-resistant trees. Our findings suggest continued broad utility of CBI, while highlighting how the correspondence of aggregate plot-level CBI to different measures of burn severity can vary with forest conditions. We also suggest considerations for broadening CBI to account for more extreme levels of burn severity.