You are here

Harvard Forest >

Harvard Forest Symposium Abstract 2014

  • Title: Deciduous tree phenology: from individuals to the landscape
  • Primary Author: Stephen Klosterman (Harvard University Herbaria)
  • Additional Authors: Arturo Martinez (Carnegie Mellon University); Andrew Richardson (Northern Arizona University)
  • Abstract:

    Vegetation phenology monitoring has yielded a decades-long archive, documenting the impacts of global change on the biosphere. However, the coarse spatial resolution of remote sensing obscures the organismic level processes driving landscape phenology, while point measurements on the ground have a limited extent. To characterize the phenology of individual trees at the landscape scale, we flew an unmanned aerial vehicle (UAV) every 4 to 7 days during spring greenup in 2013, and regularly throughout the rest of the growing season at Harvard Forest.



    We obtained RGB imagery of the forest canopy over an area encompassing a 250 meter MODIS pixel and georeferenced it with publicly available aerial photography (Massachusetts Office of Geographic Information). Tree crowns were identified, traced, and associated with species (Figure 1) using a geographic species map from an ongoing project at Harvard Forest (courtesy of David Orwig, Harvard Forest). Dates indicating the start, middle, and end of spring were calculated for individual trees, corresponding to the dates of 10%, 50%, and 90% amplitude in time series of green chromatic coordinate (Klosterman et al., 2014). For a physiological interpretation of digital image metrics of tree phenology, we obtained a record of direct visual assessment of leaf phenology for the dominant species in the study area during spring 2013 (courtesy of John O'Keefe, Harvard Forest), including the dates of bud break and 75% leaf size.



    For the three most well represented deciduous trees at Harvard Forest, red oak, red maple, and yellow birch, the observed date of bud break was four to six days later than the start of spring date derived from UAV images. Leaves also reached 75% of their mature size three to ten days later than the end of spring date from UAV data. While the dates from direct assessment and UAV image analysis are similar, these results indicate that image-based metrics may precede physiological assessments from an aerial view, as well as from the oblique view of tower-based cameras (Keenan et al., in press).



    Preliminary analysis of spring green up in 2013 indicates that trees with a later end of spring date than the scene average cluster spatially in several areas of the study region (Figure 2), indicating this phenomenon may be due to microenvironmental variation. In future work we will complete the partial analysis shown in Figure 2, explore what landscape features correlate with end of spring phenology, and compare individual tree phenologies derived from UAV images with stand scale metrics from the Harvard Forest PhenoCam and landscape metrics from remote sensing.



    References:



    Keenan, T. F., Darby, B., Felts, E., Sonnentag, O., Friedl, M., Hufkens, K., O’Keefe, J., Klosterman, S., Munger, J. W., Toomey, M. and Richardson, A. D.: Tracking forest phenology and seasonal physiology using digital repeat photography: a critical assessment, Ecol. Appl., in press.



    Klosterman, S. T., Hufkens, K., Gray, J. M., Melaas, E., Sonnentag, O., Lavine, I., Mitchell, L., Norman, R., Friedl, M. A. and Richardson, A. D.: Evaluating remote sensing of deciduous forest phenology at multiple spatial scales using PhenoCam imagery, Biogeosciences Discuss., 11(2), 2305–2342, doi:10.5194/bgd-11-2305-2014, 2014.



    Figure captions:



    Figure 1: Georeferenced UAV imagery with region of interest (white line) for a tree crown, and species map location (orange dot) of stem.



    Figure 2: Phenology map of study area, with individual trees outlined in the color of their end of spring date. Later ends of spring cluster in several areas, especially the northwest. The black outline indicates the ground area of a 250m MODIS pixel.

  • Research Category: Ecological Informatics and Modelling, Group Projects, Physiological Ecology, Population Dynamics, and Species Interactions

  • Figures:
  • Fig1_Low_Res.jpg
    Fig2.jpg