33 
h, 86, 3039-3054, 
ESTIMATION OF SURFACE REFLECTANCES AND VEGETATION INDEX 
data’, Int. Journal 
USING NOAA / AVHRR : 
METHODS AND RESULTS AT GLOBAL SCALE. 
it calibration’, Int. 
an, D. Tanré, 'Sun 
of the 6th ISPRS 
k 
global vegetation 
B. BERTHELOT, G. DEDIEU, F. CABOT, and S. ADAM. 
LERTS, Unite mixte CNES-CNRS. 18 avenue E Belin, 31055 Toulouse Cedex, France. 
Tel: 61.28.14.15; Fax: 61.28.14.10 
pplied Optics, 22, 
ABSTRACT 
□) using pointable 
Radiometric processing of global scale AVHRR data is described. It includes calibration of raw signal and 
correction of atmospheric effects on satellite measurements. These corrections are based on climatologies. A 
Applied Optics, 
sensitivity study is made to test the retrieval of surface reflectances and NDVI under various assumptions on 
atmospheric characteristics. Variations due to change of aerosol optical thickness and water vapor content are 
î NOAA/AVHRR 
investigated as well as the modifications involved by accounting for ground elevation. The noise on temporal 
NDVI profiles which remains after atmospheric corrections is also discussed. 
ation indices and 
5, 1990. 
AA A VH RR and 
KEYWORDS: Surface reflectance, atmospheric corrections, NOAA/AVHRR, NDVI. 
ihelian grasslands 
1. INTRODUCTION 
Lite measurements 
AVHRR time series data sets are now available over ten years and constitutes an unique data set offered to the 
:) model. Part 2: a 
S, p.20791-20801, 
scientific community Long term studies such as vegetation evolution or net primary productivity estimates 
are made possible at global scale. However, it supposes that problems of calibration, correction of 
atmospheric and directionnal effects, cloud filtering, be resolved with enough accuracy if one wants to 
ton, 1981. 
face sur une série 
Measurements and 
interprete the interannual fluctuations as real variations of surface characteristics. 
Calibration of AVHRR data (Visible and Near Infrared channels) has focused the attention of 
AVHRR data users since several years in order to assess, improve, and preserve the quality of the data with 
time. The correction of atmospheric effects on a global data set requires first a correction model adapted to 
al effects from a 
data processing, in term of accuracy and volume of data, and second, to be able to specify characteristics of 
atmospheric parameters which are variables in time and space, mainly aerosol characteristics, water vapor and 
t’s surface for the 
ozone contents. 
Variations of top of the atmosphere (TOA) reflectances are quite different from surface reflectance 
on wheat canopy 
variations. They depend on ground reflectance level, width of the sensor spectral band, illumination and 
observation geometries, and atmosphere characteristics at the time of acquisition. According to Goward et al. 
‘Description of a 
' Remote Sensing, 
(1991), a change of 50 % can be observed between surface and TOA reflectances. From that evaluation, it 
seems that the correction of the atmospheric effects can lead to a better interpretation of the results, provided 
that atmosphere characteristics be known. 
^VHRR products: 
Atmospheric corrections based cm climatologies are assessed in this paper. We also studied the 
sensitivity of surface reflectances and NDVI retrieval to various assumptions on atmospheric characteristics. 
-7 meteorological 
Temporal evolution of reflectances and NDVI is also studied and residual noise is discussed. 
iristics for global 
2. DATA 
th American land 
Input data used in this study are shortly described in the following paragraphs. Although processing has been 
lite data’. Science, 
applied to global data set, in order to improve the clarity of the presentation the characteristics of the input 
data and different results will be presented on a latitudinal transect chosen arbitrarily at 20° East, on july 
ance of vegetation 
17th, 1989. The transect starts from the north of Europe (Finland) to mountains of East europe, then from 
Lybian desert to high plateau of South Africa. Elevations range from sea level to about 1500 m in the Balkans 
: or reducing noise 
(Edwards, 1989, Fig. 1). North latitude are noted with negative sign. Impact of altitude on atmospheric 
corrections is discussed in section 4.