Close Window

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 — Carnegie Institution (US-PI)
Jose Natalino Macedo Silva — Brazilian Forest Service (SA-PI)

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).



Objectives





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



1995-1999



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

Close Window