B41K-0196:
Standardizing PhenoCam Image Processing and Data Products

Thursday, 18 December 2014
Thomas E Milliman1, Andrew D Richardson2, Stephen Klosterman2, Josh M Gray3, Koen Hufkens2, Donald Aubrecht2, Min Chen2 and Mark A Friedl4, (1)Univ. of New Hampshire, Institute for the Study of Earth, Oceans and Space (EOS), Durham, NH, United States, (2)Harvard University, Cambridge, MA, United States, (3)Boston University, Earth and Environment, Boston, MA, United States, (4)Boston University, Boston, MA, United States
Abstract:
The PhenoCam Network (http://phenocam.unh.edu) contains an archive of imagery from digital webcams to be used for scientific studies of phenological processes of vegetation. The image archive continues to grow and currently has over 4.8 million images representing 850 site-years of data. Time series of broadband reflectance (e.g., red, green, blue, infrared bands) and derivative vegetation indices (e.g. green chromatic coordinate or GCC) are calculated for regions of interest (ROI) within each image series. These time series form the basis for subsequent analysis, such as spring and autumn transition date extraction (using curvature analysis techniques) and modeling the climate-phenology relationship. Processing is relatively straightforward but time consuming, with some sites having more than 100,000 images available. While the PhenoCam Network distributes the original image data, it is our goal to provide higher-level vegetation phenology products, generated in a standardized way, to encourage use of the data without the need to download and analyze individual images.

We describe here the details of the standard image processing procedures, and also provide a description of the products that will be available for download. Products currently in development include an “all-image” file, which contains a statistical summary of the red, green and blue bands over the pixels in predefined ROI’s for each image from a site. This product is used to generate 1-day and 3-day temporal aggregates with 90th percentile values of GCC for the specified time-period
with standard image selection/filtering criteria applied. Sample software (in python, R, MATLAB) that can be used to read in and plot these products will also be described.