Remote Detection and Modeling of Abrupt and Gradual Tree Mortality in the Southwestern USA
Abstract:Current climate models predict a warming and drying trend that has a high probability of increasing the frequency and spatial extent of tree mortality events. Field surveys can be used to identify, date, and attribute a cause of mortality to specific trees, but monetary and time constraints prevent broad-scale surveys, which are necessary to establish regional or global trends in tree mortality. This is significant because widespread forest mortality will likely lead to radical changes in evapotranspiration and surface albedo, which could compound climate change. While understanding the causes and mechanisms of tree mortality events is crucial, it is equally important to be able to detect and monitor mortality and subsequent changes to the ecosystem at broad spatial- and temporal-scales.
Over the past five years our ability to remotely detect abrupt forest mortality events has improved greatly, but gradual events—such as those caused by drought or certain types of insects—are still difficult to identify. Moreover, it is virtually impossible to quantify the amount of mortality that has occurred within a mixed pixel. We have developed a system that fuses climate and satellite-derived spectral data to identify both the date and the agent of forest mortality events. This system has been used with Landsat time series data to detect both abrupt and general trends in tree loss that have occurred during the past quarter-century in northern New Mexico. It has also been used with MODIS data to identify pixels with a high likelihood of drought-caused tree mortality in the Southwestern US. These candidate pixels were then fed to ED-FRT, a coupled forest dynamics-radiative transfer model, to generate estimates of drought-induced. We demonstrate a multi-scale approach that can produce results that will be instrumental in advancing our understanding of tree mortality-climate feedbacks, and improve our ability to predict what forests could look like in the future.