SENEGALESE LAND SURFACE CHANGE ANALYSIS AND BIOPHYSICAL PARAMETER ESTIMATION
USING NOAA AVHRR SPECTRAL DATA
David L. Toll
NASA/Goddard Space Flight- Center
Greenbelt, MD 20771
Fred M. Yukovich
Research Triangle Institute
Research Triangle Park, NC
ABSTRACT
Surface biophysical estimates were derived
from analysis of NOAA Advanced Very High Spectral
Resolution (AVHRR) spectral data of the
Senegalese area of west Africa. The derived
surface parameters were of solar albedo, spectral
visible and near-infrared band reflectance,
spectral vegetative index, and ground
temperature. Wet and dry linked AVHRR scenes
from 1981 through 1985 in Senegal were analyzed
for a semi-wet southerly site near Tambacounda
and a predominantly dry northerly site near
Podor. Related problems were studied to convert
satellite derived radiance to biophysical
estimates of the land surface. Problems studied
were associated with sensor miscalibration,
atmospheric correction, surface anisotropy of
reflected radiation, narrow satellite band
reflectance to broad solar band conversion, and
ground emissivity correction.
Solar albedo derived estimates for both sites
were indicated not to change markedly with
significant antecedent precipitation events or
correspondingly from increases in green leaf
vegetation density. The bright soil/substrate
contributed to a high albedo for the dry related
scenes, whereas the high internal leaf
reflectance in green vegetation canopies in the
near-IR contributed to high solar albedo for the
wet related scenes. Furthermore, the
relationship between solar albedo and ground
temperature was poor, indicating the solar albedo
had little control of the ground temperature.
NDVI and the derived visible reflectance were
more sensitive to antecedent rainfall amounts and
green vegetation changes than were near-IR
changes. The information in the NDVI related to
green leaf density changes primarily was from the
visible reflectance. The contribution of the
near-IR reflectance to explaining green
vegetation is largely reduced When there is a
bright substrate or the canopy has significant
woody vegetation amounts.
KEYWORDS: Albedo, NDVI, AVHRR, Senegal,
Land Change 1
1- TMRCJXX7TEDN
The western sub-Saharan region has over the
last decade undergone periods of severe drought
followed by periods of near—average rainfall
(Motha et al. 1980 and Nicholson 1985) . Analysis
of the polar orbiting NOAA Advanced Very High
Resolution Radiometer (AVHRR) spectral data may
provide a means to study regional land surface
changes associated with drought and anthropogenic
activities and provide further insight to land
surface and atmosphere interactions.
However, the extraction of biophysical
parameters of the land surface from the AVHRR
using directional and spectral radiance is
associated with numerous difficulties that were
examined for this study: 1) radiometric
calibration of the optical bands (i.e., the AVHRR.
bands without an on-board calibrator) (Hoiben et
al. 1989a); 2) land surface anisotropy sensed by
directional derived AVHRR sensors (Kimes et al.
1985); 3) solar albedo estimation from AVHRR
narrow band passes (Toll 1989); 4) heterogeneous
ground emissivity complicating the land surface
temperature estimate (Becker 1987); and 5)
atmospheric aerosol and water vapor temporal and
spatial variability affecting ground surface
derivations (D’Almeida 1986).
The primary objective of the study is to use
AVHRR derived spectral data to evaluate land
surface changes and to provide biophysical
information derived from the satellite: radiance
to examine land surface-climatology interactions.
To characterize the landscape, surface derived
estimates of solar albedo, spectral vegetation
index and ground temperature are extracted from
the AVHRR spectral data. Surf ace meteorological
data of the western sub-Saharan for Tambacounda
and Podor (available from the NOAA National
Climate Data Center, .NCDC) were used to report
climatic conditions and to describe the state of
the atmosphere (e.g., temperature, rain, aerosol
loading, wind, and clouds) for the time of the
satellite overpass.
2. STUDY SUB
The area selected for study is the Senegalese
area in the sub-Saharan encompassing two
ecological areas from north to south, the Sahel
(averaging 100-400 mm of precipitation) and the
Soudan (400-800 mm) (Nicholson 1983) . The study
region area is diverse because of a steep south
to north precipitation gradient, with mean annual
isohyets occurring parallel to equal lines of
latitude. The vegetation is largely steppe, with
changes to desert in the north and changes to
savannah in the the south (Monad 1986). The area
has extensive pastoral lands, largely from
insufficient rainfall to support agriculture.
Man induced effects from overgrazing of livestock
and gathering of firewood has compounded the
environmental effects from drought. Justice and
Hiemauz (1986) report these changes in addition
to drought have led to an overall lower net
primary productivity as derived through analysis
of NOAA AVHRR spectral data. Tucker et al.