Trend analysis of time-series phenology of North America derived from satellite data

Remote sensing information has been used in studies of the seasonal dynamics (phenology) of the land surface since the 1980s. While our understanding of remote sensing phenology is still in development, it is regarded as a key to understanding land-surface processes over large areas. Phenologic metrics, including start of season, end of season, duration of season, and seasonally integrated greenness, were derived from 8 km advanced very high resolution radiometer (AVHRR) data over North America spanning the years 1982-2003. Trend analysis was performed on annual summaries of the metrics to determine areas with increasing or decreasing growing season trends for the time period under study. Results show a trend toward earlier starts of season in limited areas of the mixed boreal forest, and a trend toward later end of season in well-defined areas of New England and southeastern Canada. Results in Saskatchewan, Canada, include a trend toward longer duration of season over a well-defined area, principally as a result of regional changes in land use practices. Changing seasonality appears to be an integrated response to a complex of factors, including climate change, but also, in many places, changes in land use practices. Copyright ?? 2006 by V. H. Winston & Son, Inc. All rights reserved.