Harvard Forest >

Harvard Forest Symposium Abstract 2012

• Title: A Segmentation Approach for Estimating Forest Structural Characteristics from Lidar and Radar: Analysis and Error Assessment
• Primary Author: Paul Siqueira (University of Massachusetts - Amherst )
• Additional Authors: Razi Ahmed (UMASS Amherst); Caitlin Dickinson (University of Massachusetts - Amherst )
• Abstract:

  One of the nine NASA funded Bigfoot sites for connecting remote sensing measurements to ground process observations of carbon flux and net primary production, the Harvard Forest has been a resource for a wide variety of ecological studies on spatial scales extending from the microscopic to macroscopic. Typical characteristics of the region that are relevant to this study are a mean carbon content of 120 MgC/ha, an average height of 24m, a mean basal area of 40 m2/ha, and on the order of 1000 trees/ha. In the summer of 2009, in addition to a 15 hectare ground validation effort, the Laser Vegetation Imaging Sensor (LVIS) and UAVSAR overflew the Harvard region, collecting full waveform lidar and L-band SAR data over a 30 km area (10 km x 30 km), for developing biomass and vegetation structure estimation algorithms relevant to NASA’s DESDynI mission. This rich and consistent data set provides an opportunity to explore relationships between a wide variety of ground validation data and remote sensing sources (lidar, SAR and InSAR),and to better understand methods of combining these fundamental data sources for studying ecosystems, carbon balance and vegetation three-dimensional structure. In this poster we explore a variety of remote sensing techniques for estimating biomass and structure. Important themes in this work are: i.) use of a segmentation algorithm to analyze regions of “self-consistent” response to lidar and radar ii.) inclusion and description of a rigorous error analysis and iii.) use of polimetric decomposition algorithms and multibaseline InSAR for characterizing forest structure.

• Research Category: