EN
s
where a' and b' are not constants but depend on the type. of aerosol (6,21) and
cannot be determined without experimentation. Thus, in the absence of an exact
knowledge about a' and b', no matter how well refined the atmospheric correction
algorithm one uses, one is left with a sense of uncertainty. But, at present
we are interested in comparing the results obtained from the two algorithms, an
| exact knowledge of a' and b' should not matter as long as the same values of a'
and b' are used with both algorithms. In this work we have estimated the
| aerosol optical thicknesses from reference (6). A new expression for al], >)
| was obtained by ignoring the assumptions (a) and (b) and was used in the NEW
algorithm.
| The aximuthal angle of the satellite-sensed radiation depends on the
scanner tilt angle as well as on the position of the picture element in question
| along a given scan line. Assuming a flat ocean surface, an expression for the
| azimuth angle of the satellite-sensed radiation was obtained and was implemented
in the NEW algorithm.
In order to compare the water-leaving radiances obtained from the OLD. and
NEW algorithms, we have selected a typical CZCS scene from the Irish Sea which
was imaged on 15 April 1981. Figure 1 shows that the calibration lamp temp-
erature on this date was about 38K below the nominal value of 2000K. The
inflight calibration constants were computed using the preflight values of the
lamp radiances. The slopes for the CZCS channels 1 to 4 were found to be,
respectively, about 39%, 33%, 30% and 23% higher than the preflight values. For
analysing the selected scene, we have used the inflight calibration constants.
| The values of the solar irradiances on the top of the atmosphere were taken from
| N reference (11).
| | Let P()) denote the water-leaving radiance as a percentage of the satell-
Hi ite-sensed radiance. In Fig.2 we have plotted the values of P(A) obtained from
il the OLD algorithm as a function of P(A) obtained from the NEW algorithm. Fig-
| i ures 2(a) to 2(d) are the results for the CZCS channels 1 to 4 respectively.
| 1M These results show that there seems to be a simple relation between the old and
| | new values of P(\). A least squares analysis produces the following relations:
i
| y = -3.898 4 3.015x,^ 4 - 448nm, R' - 0.96
| v= 1.300 + 0.987x;: ‘à = 520m, R',- 1.0 "m
1 yr 1.688 0.984x, -A.- S50nm, R^ s 1.0
| MN y =,0.370 + L.052x, ° à =.670nm, RB. =.1.0
|
|
|
where x and y stand for the new and old value of P(A) respectively, and R is the
correlation coefficient. These results indicate that the values of x and y are
fairly close to each other. However, the chlorophyll algorithm involves the
il ratio of the water-leaving radiance in the CZCS channel 1 to that in the CZCS
channel 3. In order to compare this radiance ratio, in Fig.3 we have plotted
the quantity P(443nm)/P(550nm) obtained from the OLD algorithm as a function of
the corresponding quantity obtained from the NEW algorithm. A least squares
analysis using the data presented in this figure yields the following relation
Y = 0.052 + 0.834X, HR? = 0.96 (4)
where X and Y represent the new and the old value of the radiance ratio respect-
ively. This relation shows that the deviation of Y from X is about 11%. Thus,
the OLD algorithm underestimates the radiance ratio in question by about 11%.
Such an underestimate of the radiance ratio should have some effect on the para-
meters of the chlorophyll-like pigment algorithm. Further investigation on this
point is being pursued.
SOME OTHER SOURCES OF ERRORS AND UNCERTAINTIES
Many authors (for example, see refs. 4,9,22-24) have discussed various
types of errors involved in the remote measurement of the water colour. Some
of the requirements for a better accuracy of the atmospheric correction method
and in-situ measurements have been pointed out by Sérensen (14). In addition
to the sources of errors and uncertainties discussed in the above cited refer-
ences there are many other sources of errors and uncertainties, some of which
702
EE
TA TE rt ST a < he, SE E RR
EE cse S 755 N guet
