A Qualitative Characterization of Spring Vegetation Phenology Using MODIS Imagery for the Piedmont of North Carolina from 2000 to 2007

Recent studies have shown vegetation phenology around the world is being altered by increased variability in temperatures associated with a warming climate. The onset of spring and the duration of the growing season in many eastern states has been pushed forward an average of 2-5 days and lengthened by as much as 10-15 days respectively, as a response to climatic forcing. Analyzing phenological changes to forest dynamics is aided by the use of satellite imagery with high temporal and spatial resolution to accurately estimate the timing of recurrent events associated with the flush of green vegetation at the beginning of spring in deciduous forests. This study used daily MODIS images at 250m processed to Normalized Difference in Vegetation Index (NDVI) for the spring greenup from 2000-2003 and 2007. Of the 792 available images, 20 sites along the Piedmont, coastal plain, and mountains of North Carolina were filtered (a lowpass Savitsky-Golay convolution filter) to remove atmospheric noise, and used to estimate relevant phenological parameters. Onset of spring, length of growing season, rate of green-up, as well as, maximum green-up, were identified using a segmented regression technique. Over the study period, the Piedmont sites exhibited high variability in dates of onset among sites ( 5days) and negatively between years (6 days), with concurrent variability in growing season length. Furthermore, using the NDVI response in regressions of climate variables at the AmeriFLUX site in Duke Forest from 2001-2003, showed growing degree-days since last freeze and mean soil temperature as most significantly in agreement with phenological change. Future studies should focus on acquiring daily satellite imagery to monitor the changes and variability seen among sites and years with careful attention given to severe weather anomalies. Creating maps of relevant climatic variables may provide a more accurate means of predicting phenology and determining the influence of site-specific environmental variables.