The rate of drought-induced tree mortality has increased across the world, and is expected to continue to increase further as temperatures (and associated with them, evaporative demand) continue to rise. However, in situ monitoring of this heterogeneous process is cumbersome and expensive. Predicting where droughts will lead to tree mortality is critical for forest management and understanding forest risk, but plant physiological models differ widely in their mortality predictions. A number of simple climatic indices have also been used as indicators of mortality risk, but these do not take into account differences in plant drought response strategies that also affect mortality risk. Greenhouse studies have suggested that depletion of plant water content may be a better predictor of mortality than the xylem water potential typically used in plant physiological predictions of tree mortality. Observation of vegetation optical depth (VOD) derived from passive microwave radiometry are proportional to canopy water content, and are available globally and with high temporal frequency. They could thus be used to predict tree mortality at large scales. Here, we compare the fractional area of mortality (FAM) predicted based on VOD to mortality rates observed using aerial detection in the Sierra Nevada during the 2012-2016 California drought.
Results/Conclusions
Relative water content determined from VOD outperformed the commonly used climatic water deficit index in estimating mortality rates. Furthermore, a random forest statistical model for FAM with several climatic, topographic, and vegetation indices indicated relative water content derived from VOD is the most important input in the random forest model, and is more than twice as important as other variables. Although the results of this study are limited by the kilometer-scale resolution of the VOD dataset used and the imperfections of the aerial detection dataset, our results nevertheless demonstrate that microwave-derived estimates of vegetation water content can be used to study drought-driven tree mortality, and may be a valuable tool for mortality predictions if they can be combined with variables measured at higher spatial resolution