You are here

Harvard Forest >

Harvard Forest REU Symposium Abstract 2012

  • Title: Alternatives to Remote Sensing of Phenology using Public Outdoor Webcams
  • Author: Dmitri Ilushin (Harvard College (Harvard University))
  • Abstract:

    Currently, there are hopes of expanding tree phenology tracking to a global scale, but established methods of phenology tracking are limited now by their low temporal granularity or their prohibitive costs in terms of manpower. With the growth of the internet, publicly available webcams have become more and more available, providing a set of data that covers all areas of the world with high temporal and spatial granularity. This summer our goal was to analyze the utility of public webcams in studying local phenological trends on a global scale. We used a repository of publicly available webcam images from AMOS (the Archive of Many Outdoor Scenes, administered by Washington University in St. Louis). Through visual inspection, we classified a subset of 1879 geolocated sites based on camera stability and overall image quality for the series of photos into one of three categories: stable locations with few FOV (Field Of View) shifts, consistently changing FOV sites, and poor locations. Of these, sets of images from cameras both marked as stable and containing vegetation were put into a program to create ROI (Region Of Interest) masks around vegetation and then processed to extract a time series of GCC (Greenness Chromatic Coordinate), the percentage of greenness to overall brightness, within an image's vegetated area. The resulting time series can then be compared with established phenological observation methods and help corroborate the utility of webcam imagery as an alternative method for studying tree phenology. By better understanding how trees respond to climactic events, we can further understand yearly NPP (Net Primary Productivity, a measure of overall carbon intake of trees) responses to increased carbon concentrations in the atmosphere. AMOS provides a new frontier to explore tree phenology and will help allow us to understand better how forests respond to climate change.

  • Research Category: Ecological Informatics and Modelling; Group Projects