data acquired
1 the ground
required in
der to
Lon techni-
_ number of
. Monitor (OCM)
5 now planned
1 is known as
1985 ; McEvoy
je data,
tans gorm are
4 16. The
ically
commonly
npressLon
If the data.
n hatios o$
btain this
med data. This
ed and we
value of 255
^. system
nite .
25 fo T.
54545 Of
h a Linear
d to be
to calculate
Lc effects.
o any high
ning some
t is unlikely
wn to better
By comparison, the error due to compression and reconstruction is comapa-
tively small and can be regarded as an acceptable price to be paid for the
Amprovement in coverage.
4 DETERMINATION
We have used a simple power Law expression to determine chlorophyll
concentrations. This is an algorithm which is widely used amongst
experimenters. Explicitly, the chlorophyll concentration, C, is gtven by :
53 de «(Y
where L, , L ze atmospherically corrected radiances in CICS bands
1 and 3
a , b = adjustable constants.
?
The constants a and b are nonmally determined by Least-squares fitting
to sea-thuth measurements and their values can vary considerably.
Data compression causes errors in L, and Lg, which then cause errors
in the value of C. Theory shows that the percentage error in C is directly
proportional to ithe percentage evons in Ly and Lg and to the value of
b but {independent of a. In our experiments, we have used values of b
ranging from -2 up to -8.286, which represents the range used by experimen-
ters. Errors in the chlorophyll concentration were found to be almost
directly proportional to b as predicted by theory. Using b=-§.286 (the
worst case) errors were found to be 22%ams.
This figure should be compared with the errors in chlorophyll
concentration due to the other system errors (see section 3). In particular
a 2% error in the atmospheric correction will Lead to an error in C of
approximately 60%, for b=-8.286.
We should also consider the errors which are found in surface measu-
nements. Because of the variation in concentration from point to point, a
surface measurement may be quite unrepresentative of the average concentra-
tion within a CICS pixel (800m square). Errors from this source can be up
to 50% (Charlton, 1979).
5 COVERAGE IMPROVEMENT
We have considered a satellite carrying an instrument package similar
£o ERS-1, namely : ‘
OCM - 2.8 Mbit/s
SAR (wave mode) - 600 kbits/4
Mtimeter - 10 kbits/4
Wind 4cattermeter - 1 kbit/s