Potentially limited detectability of short-term changes in boreal fire regimes: a simulation study

Abstract

Climate change is expected to increase area burned in the boreal plains ecozone of Canada in the early 21st century (2001–50). I examined the influence of inter-annual variability in area burned and short observed time series on the probability of detecting if an increase has occurred, using a null model of present and future fire regimes. A wide range of fire cycles are consistent with annual area burned in the late 20th century (1959–99). Fire cycles estimated from the reciprocal of the average annual burn fraction over a 50-year period are not very precise, and overestimate the fire cycle if years with large annual area burned have not recently occurred. Under the default assumptions, the probability of detecting a doubling of annual area burned during 2001–50 is 73% if it occurred instantaneously, but only 31% if it occurred gradually. Imprecise estimates and uncertainty in the ability to detect changes in fire cycles poses challenges for implementing aspects of sustainable forest management. Alternate empirical or model-based statistics, such as return periods for annual areas burned of a given magnitude, may be useful for inferring frequencies and magnitudes of large fire years that have not yet been observed.