Daily Global Area Coverage (GAC) Images for Africa from the NOAA AVHRR Radiometers, 1982-1992, from SAI/JRC

General information about NOAA-AVHRR can be queried by interested
users in the category 'Sensor' and 'Source'. Some basic information is
given hereafter.

The Advanced Very High Resolution Radiometer (AVHRR) onboard NOAA 6,
8, 10 and TIROS-N measured in four spectral bands, while the NOAA 7, 9
and 11 are measured in 5 bands. The primary objective of the AVHRR
instrument is to provide cloud top and sea surface temperatures
through passively measured visible, near infra-red and infra-red
spectral radiation bands. Nevertheless these data are widely used for
terrestrial applications, such as land cover mapping and vegetation
monitoring.

The available data provide a long term AVHRR Global Area Coverage
(GAC) data set with particular emphasis placed on the continent of
Africa. Normalized Difference Vegetation Index (NDVI), channel 2
reflectance, channel 3 and 4 brightness temperatures, an approximate
surface temperature and a cloud probability image are available on a
daily basis from January 1982 to December 1992 for the whole African
Continent at a resolution of 5 km.

The remaining data from the entire GAC time series (July 1981 to the
present) will be processed by the end of 1995.

Channels 1 and 2 are converted to radiance using the August desert
calibration coefficients published by Holben et al. (1990).
Radiances in channels 1 and 2 are converted to 'top of atmosphere'
reflectances (ToA). Brightness temperatures are calculated for
channels 3, 4 and 5 using the inverse Planck function (Kidwell 1991).

Channels 4 and 5 are corrected for non-linearity of sensor response to
give true brightness temperatures using pre-launch correction
coefficients (Planet 1988). The brightness temperature and
reflectance images are scaled to 8 bit integers.

The ToA reflectances and brightness temperatures are used to identify
cloudy pixels. The channel 1 and 2 reflectances are then used to
compute the NDVI, and channels 4 and 5 true brightness temperatures
used to compute an approximate surface temperature using a 'split
window' technique (Price 1984). This does not take variations in
emissivity into account, and so is only of limited accuracy.

The NDVI, channel 2 reflectance, channels 3 and 4 brightness
temperature, surface temperature and the cloud probability channel are
then geometrically corrected. Geometric correction involves three
steps; navigation using the ELPs from the raw data, correlation with a
reference image data base to provide ground control points for fine
correction, and resampling into a daily continental scale mosaic.

For Africa a Mercator map projection is used. The Africa GAC mosaic
map center coordinates are 0#161#, 17.25#161# E, pixel size at the
equator is 5 by 5 km, giving an image of 1800 lines by 1600 columns,
with top left co-ordinates 37.59#161# N, 18.64#161# W, bottom right
37.59#161# S, 53.64#161# E. As a function of the Mercator projection
the resolution degrades by approximately 20% at the northern and
southern limits; for example the ground area represented by each pixel
is 30 km2 at 35#161# N or S, (4 km, x axis by 7.5 km, y axis) compared
with 25 km2 at the equator.

The processed data sets are stored as ERDAS 7.4 format files on a
gigatek optical disk juke box. Data are currently made available
through formal collaborative research agreements between outside
laboratories and the Joint Research Centre. In such instances data
costs are for marginal cost of reproduction only.

These data will become available to the international research
community through the EC and European Space Agency initiative, the
Centre for Earth Observation (CEO).

The data sets have already provided new information concerning inter
and intra annual variations in vegetation fire dynamics for Africa and
have been used to derive forest seasonality information through the
JRC`s thematic projects such as TRopical Ecosystem and Environment
observations by Satellite (TREES).

See separate entry for TREES.

Example data can be found on the CEO World Wide Web home page:
"http://www.ceo.org/".