Prototype for monitoring and forecasting fall foliage coloration in real time from satellite data

While determining vegetation phenology from the time series of historical satellite data has been widely investigated throughout the last decade, little effort has been devoted to real-time monitoring and short-term forecasting. The latter is more important for numerical weather modeling, ecosystem forecasting, forest and crop management, and health risk warning. In this study we developed a prototype approach for the real-time monitoring and short-term forecasting of fall foliage status (including low coloration, moderate coloration, near-peak coloration, peak coloration, and post-peak coloration) using temporal satellite observations. The algorithm combined the climatology of vegetation phenology and temporally available satellite observations to establish a set of potential temporal trajectories of foliage development at a given time. These trajectories were used to identify foliage coloration phases in real time, to predict the occurrence of future phenological events, and, furthermore, to analyze the uncertainty of monitoring and forecasting. With an increase in satellite observations, monitoring and forecasting were continuously updated. The approach developed was tested using MODIS (Moderate Resolution Imaging Spectroradiometer) data at a spatial resolution of 500 m across northeastern North America and evaluated using field measurements at the Harvard Forests of the northeastern United States and standard MODIS foliage coloration phases. The results indicate that short-term forecasting can be well implemented in more than half a month before the occurrence of a foliage phase, and that the accuracy of the real-time monitoring of both near-peak-coloration and peak-coloration occurrence is less than 5 days in most mixed forests and deciduous forests. (C) 2012 Elsevier B.V. All rights reserved.