Monitoring forest disturbance events and recovery status with eDaRT 3.0

The worldwide demand for timely and accurate information about ecosystem dynamics at Landsat spatial scale is growing and as of today still exceeds the availability of information. We discuss recent enhancements and performance assessments of the UC Davis/USFS Landsat-based automated forest monitoring system called Ecosystem Disturbance and Recovery Tracker (eDaRT). The system has been operated since 2012 by the USDA Forest Service on California forests to generate most current disturbance maps: disturbance events on unprecedented, sub-annual scales and up to 16-day snapshots of their cumulative effects, in terms of variety of remote-sensing based metrics of change.

All usable Landsat images can be automatically processed, making the system suitable for providing input into long-term retrospective studies and near-real-time year-around continuous monitoring of change. eDaRT's highly scalable, flexible, and modular algorithms and their prototype parallel implementation allow one to rapidly process large areas (e.g. the State of California) without relying on supercomputing power, thus providing flexibility of customization, enhancements, and agility needed to meet evolving and diversifying user demand for forest health information.

Recent performance tests in California showed high sensitivity to detection of subtle subpixel level forest stress, mortality, harvest, and fires, down to 1% change in canopy cover; with false positive rates as low as 10%. The eDaRT development continues toward new functionality and further improvements in accuracy. It has been increasingly used as an experimental platform for collaboration between users (forest managers, researchers) and providers of customized remote-sensing based analytics.