Timeliness in Forest Change Monitoring: A New Assessment Framework Demonstrated using Sentinel-1 and a Continuous Change Detection Algorithm

The development of near-real time forest monitoring systems, which are used to create alerts for events such as logging or fire, have brought attention to an important consideration for remote sensing applications: not only is accuracy important but so is timeliness. In the context of forest monitoring, timeliness is measured by the lag between when a change occurs in the forest and the creation of an alert about the event. Timeliness is not captured in typical approaches to map accuracy assessment, and there is no community-accepted approach for evaluating accuracy of forest change as a function of time. Consequently, there is no straightforward method of comparing monitoring systems in the context of accuracy and timeliness, which is needed for determining their usefulness in achieving real-world objectives. Here, we propose an assessment framework that characterizes forest change detection accuracy as it relates to timeliness. Using a change monitoring algorithm that applies the Continuous Change Detection and Classification algorithm (CCDC) to Sentinel-1 radar data in Madagascar, we demonstrate how the assessment framework can be used to calculate a lag-dependent F1-Score and two new metrics of performance: Initial Delay, or the minimum time required to create an alert as observed in reference data, and the Level Off Point, or the lag at which accuracy stabilizes. Critical to the assessment process is the integration of daily Planet imagery, which provides precise information on the timing of changes. These metrics are reported for our novel radar-based alert system using a (lag, F1) syntax, for example (12, 0.08) for Initial Delay and (101, 0.57) for the Level Off Point. These metrics define two key moments of the accuracy/time lag curve and provide a holistic way to evaluate and compare the usefulness of forest monitoring systems in real-world applications.