Assessing disturbance severity with lags in post-fire mortality in subalpine forests of the Sierra Nevada

Recent changes in high elevation forests worldwide indicate that climate and land use changes threaten forest structure and long-term ecosystem stability, with implications for carbon sequestration and ecosystem refugia. California's high elevations are often assumed to be stable carbon sinks due to relatively low frequency and severity of disturbance, however the frequency and area burned have increased significantly over the past century and the severity of disturbances may be underestimated using current methods. Anecdotally, subalpine tree mortality after fire is generally low in the first year, with substantial increases in mortality in the five years following. Typically, fire severity is mapped within a year of a fire, which may not capture mortality of trees severely weakened but that retained needles. Delayed mortality and compounded disturbances may lead to an underestimation of fire effects using standard fire severity maps, which are typically generated within a year after the fire. For areas burned and surrounding unburned areas, we first classified yearly tree mortality from high resolution NAIP imagery, spectrally unmixed imagery in the years between NAIP, and lidar. We then analyzed the temporal patterns of tree mortality using Gaussian process, a Bayesian machine learning method, and compared the patterns of timing between areas with and without drought in the years following the fire. While mortality cannot be attributed to specific causes or agents without extensive field data, we found patterns of lagged mortality years after fire that are currently unaccounted for when fire severity is mapped. We found the timing of additional post-fire mortality varied between drought and non-drought conditions, indicating that fire and drought stress may lead to an increase in fire severity due to increased stress. This enhanced metric of disturbance severity is expected to improve our assessment of changing fire regimes, and thus understanding of current and projected trends in carbon sequestration in high elevation subalpine forests.