Noroozi, Aliakbar
as For all scene elements on land, a reference reflection value (a function of cover class, slope and sunangle) was derived,
as called “reference image”.
is The detection problem consists of detecting clouds or haze by the rule:
IF smoke-fraction «0 THEN cloud or haze i.e. smoke fraction —0.
The estimated pollution load is derived from a fit of the sum over (t) of smoke-fraction (t) spatial distribution to
observed point samples of soot and oily deposits.
In this phase of the study it was not possible to automatically detect and trace smoke clouds going through the
condensation level, forming clouds and eventually producing black rain.
As mentioned preprocessing and data selections were performed. Radiometric correction was based on parametric
lly | histogram matching (mean and standard deviation) over a cloud and smoke free area (West of Kuwait).
he
on Considerable effort was spent on the generation of reference reflectance data for the land area. Minimum and maximum
operators plus manmade context masks were used to remove cloud, snow, haze and smoke from a sequence of images.
so
d. Estimating the smoke-concentration over water required a manual fitting of the haze level such that the smoke pattern
over land matched the pattern over water.
th
3. TRACKING OF SMOKE PLUMES AND SOOT
of
The image dated March 11, 1991 shows a smoke plume over the Gulf and Saudi Arabia; however, differentiating
of between the fire sources and the smoke is difficult on this Band 2 image. Therefore, Band 3 of the NOAA-AVHRR
| sensor, which is very suitable for detection of fires, was used for detection of the burning oil wells. The Kuwait oil
he wells fires show a clearly visible colour on band 3. It should be indicated that there are two major fire sources, one to
the North and the other one to the South of Kuwait City (Plate 1).
in
he
Im
est
nd
ve
'er
ng |
|
|
in.
le |
le
ith
Ice |
r cc
by
ler
Plate 1: Oil field fires is appeared in white spots (SCWMRC, February 1998)
The smoke plumes over the Persian Gulf were steered by the wind direction, so that different altitudes, they were
moved into various directions. The direction of the smoke movement was eastward (towards the Iran) during March and
April 1991.
| International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 1021