Print PageThe seasonal cycle of plants (i.e., phenology, including budburst, flowering, senescence and dormancy) is affected by various environmental factors such as temperature and photoperiod. Species-level climatic-phenology models for budburst showed good agreement with observations from phenology network across different climatic types, while those for plant senescence were rare and only tested in limited geographical range. Remote sensing provides regional and global scale observations of vegetation dynamics, offering a way to assess ecosystem response to the changing climate. Previous work linking remote sensed phenology with local meteorological records was limited by either the temporal length or the spatial resolution of the remote sensing data. Since its launch in 2000, Moderate Resolution Imaging Spectroradiometer (MODIS) provided us ten years’ data of vegetation dynamics at relatively high spatial resolution (250m and 500m). The objective of this study is to test climatic-phenology models which incorporate both spring warming and winter chilling requirements in budburst models and temperature/photoperiod requirements in senescence models with remote sensing observations. Here we used MODIS 8-day 500m reflectance data of New England (tile ID: h12v04) to calculate the Enhanced Vegetation Index 2 (EVI2). EVI2 was then fit with a double-logistic curve from which we derived the budburst and senescence time. Daily temperature records from 150+ weather stations in the study area were deployed to construct the climatic-phenology models. The weighted average budburst and senescence time of pixels around each station was used in conjunction with meteorological records to parameterize both budburst and senescence models. The “goodness” of phenology models was analyzed based on cross-comparison and model selection criteria. Retrospective analysis was conducted based on the “best” model and the historical climate data in this region.
Results/Conclusions:
Remote sensing data showed the spatial and temporal variation in vegetation phenology of deciduous forests in New England. A three-parameter spring warming model was selected based on the model prediction capability. Parameters such as base temperature showed the spatial distribution closely related to regional temperature. In addition, species and soil types also contribute to the variation in the trend. Retrospective analysis suggested that the rate of change of budburst, senescence time and the length of growing season varies from station to station. We will further discuss the possibility to extend this analysis to other deciduous forests.
