LC-13 Abstract

Low-Cost Evaluation of EO-1 Hyperion and ALI for Detection and Biophysical Characterization of Forest Logging in Amazonia

Gregory Paul Asner --  University of Colorado
Natalino Silva -- EMBRAPA
Rodrigo Antonio Pereira -- Fundacao Floresta Tropical

Major uncertainties exist regarding the rate and intensity of logging in tropical forests worldwide: these uncertainties severely limit economic, ecological, and biogeochemical analyses of these regions.  Recent sawmill surveys in the Amazon region of Brazil show that the area logged is nearly equal to total area deforested annually, but conversion of survey data to forest area, forest structural damage, and biomass estimates requires multiple assumptions about logging practices.  Remote sensing could provide an independent means to monitor logging activity and to estimate the biophysical consequences of this land use. Previous studies have demonstrated that the detection of logging in Amazon forests is difficult and no studies have developed either the quantitative physical basis or remote sensing approaches needed to estimate the effects of various logging regimes on forest structure.  A major reason for these limitations has been a lack of sufficient, well-calibrated optical satellite data, which in turn, has impeded the development and use of physically-based, quantitative approaches for detection and structural characterization of forest logging regimes.

We propose to use data from the EO-1 Hyperion imaging spectrometer to greatly increase our ability to estimate the presence and structural attributes of selective logging in the Amazon Basin.  Our approach is based on four "biogeophysical indicators" not yet derived simultaneously from any satellite sensor: 1) green canopy leaf area index; 2) degree of shadowing; 3) presence of exposed soil and; 4) non-photosynthetic vegetation material.  Airborne, field and modeling studies have shown that the optical reflectance continuum (400-2500 nm) contains sufficient information to derive estimates of each of these indicators.  Our ongoing studies in the eastern Amazon basin also suggest that these four indicators are sensitive to logging intensity.  Satellite-based estimates of these indicators should provide a means to quantify both the presence and degree of structural disturbance caused by various logging regimes.

Our quantitative assessment of Hyperion hyperspectral and ALI multi-spectral data for the detection and structural characterization of selective logging in Amazonia will benefit from data collected through an ongoing project run by the Tropical Forest Foundation, within which we have developed a study of the canopy and landscape biophysics of conventional and reduced-impact logging.  We will add to our base of forest structural information in concert with an EO-1 overpass.  Using a photon transport model inversion technique that accounts for non-linear mixing of the four biogeophysical indicators, we will estimate these parameters across a gradient of selective logging intensity provided by conventional and reduced impact logging sites.  We will also compare our physically-based approach to both conventional (e.g., NDVI) and novel (e.g., SWIR-channel) vegetation indices as well as to linear mixture modeling methods.  We will cross-compare these approaches using Hyperion and ALI imagers to determine the strengths and limitations of these two sensors for applications of forest biophysics.  This effort will yield the first physically-based, quantitative analysis of the detection and intensity of selective logging in Amazonia, comparing hyperspectral and improved multi-spectral approaches as well as inverse modeling, linear mixture modeling, and vegetation index techniques. The study sites of this investigation include the Fazenda Cauaxi in the municipality of Uliolandia and the Tapajos National Forest in Santarem, Para (Tab.1).


There are two primary objectives of this project:  

  1. Test the efficacy of EO-1 Hyperion imaging spectrometer and ALI multi-spectral data for detection and quantification of forest structural damage resulting from selective logging in the eastern Amazon Basin.  

  2. Test the strength of traditional and novel spectral indices, linear mixture modeling, and photon transport inverse modeling in delivering reliable estimates of biogeophysical variation across a site matrix of logging intensity and time since harvesting.

Table 1. Physical and logistical characteristics of logging research sites.  Completed field and remote sensing data collections.


Fazenda Cauaxi

FLONA-Tapajos / Fazenda Fortaleza

Central Lat./Long.

343'S, 4817'W

33'S, 5458'W

Dry season

July - December

Logging treatments

Conventional and Reduced-impact Logging

Treatment Years


1996-1999 / 1991-1999

Preliminary Field Measurements

Stand density; tree heights; crown dimensions; LAI; fPAR; location and extent of roads, skids, logdecks; GPS; field spectrometry (400-2500nm): tissue optical properties and soil reflectance; vegetation cover

Species identification and mapping; trunk diameters (dbh); GPS; field spectrometry (400-2500nm): tissue optical properties and soil reflectance; LAI

Recent Remote Sensing

Landsat 5 TM; SPOT tasked; Airborne LIDAR; Airborne digital videography

Landsat 5 TM; SPOT; JERS-1 radar; Airborne LIDAR; Airborne digital videography