International Archives of Photogrammetiy and Remote Sensing. Vol. XXXII Part 7C2, UNISPACE III, Vienna. 1999 
90 
I5PR5 
UNISPACE III - ISPRS/EARSeL Workshop on 
“Remote Sensing for the Detection, Monitoring 
and Mitigation of Natural Disasters” 
2:30-5:30 pm, 22 July 1999, VIC Room B 
Vienna, Austria 
normally made in the open sea - and often during night hours to 
avoid visual observation. From a practical point of view many 
spills of light oil often evaporate within hours but in other cases 
heavier oil may migrate under the influence of wind and current. 
There are cases of cmde oil that is submerged controlled by the 
buoyancy and is therefore not detectable. 
In the first case the observation relies on the backscatter contrast 
between the area of oil spill and that of the surrounding coast and 
sea, with the oil spill often having a veiy small backscatter due to 
the dampening of the short gravity waves (capillary waves). A 
low wind speed of 2 to 5 m/s, gives the best contrast. At higher 
wind speeds - above 7 m/s the contrast gradually disappears and 
the spill is eventually not detectable. This seems a serious limita 
tion but in fact the occurrence of wind still and large wind force 
is less frequent than we believe so that the techniques is useful in 
a great percentage of time. 
Another serious limitation is the fact that it is difficult to distin 
guish from the backscatter contrast between a natural slick or an 
intentional oil spill. Natural slicks are related to ocean dynamics 
that collect oil from for example algae into narrow patches on the 
surface. They may also be due to natural oil seepage giving slicks 
with a pattern formed by wind and currents. Finally, there may be 
sewage’s from towns or waste discliarges from for example fac 
tories of fish product, which also give signatures that may be 
mistaken for oil spill. These are disturbing ambiguities and is the 
reason why several attempts to devise a software package for 
automatic detection of oil spills have had limited success. How 
ever, it has been found that an experienced image interpreter is 
able in most cases to distinguish between natural slicks and ille 
gal spills based on pattern recognition and on his/her knowledge 
about coastal activities. The Norwegian system referred to previ 
ously relies upon a supervised automatic detection with alarms 
being evaluated interactively. 
A serious limitation stems from the lee effect that is often found 
close to the coast whereby sea roughening first takes place some 
kilometre from the shore. An oil spill in a near-shore area may 
therefore not be detected. Monitoring will therefore rely upon 
other means of observation such as airborne infrared or passive 
microwave radiometers. 
PASSIVE MICROWAVE SYSTEMS 
Microwave radiometers have been flown on Earth observation 
satellites since 1978 collecting almost global data sets that are 
being used extensively for climate studies including sea ice ob 
servations in both hemispheres. The data collected are so-called 
brightness temperatures tliat is a measure of the thermal pow'er 
that leaves the surface observed (Rayleigh-Jeans law). Being a 
passive system the spatial resolution depends upon tire size of the 
antenna measured in wavelengths and the distance. Thus, a 37 
GHz (8 mm wavelength) system with an antenna of about one 
metre gives a ground resolution of about 25 km from satellite 
heights. Constructed as a scanning system it may have a swath of 
about 800 km to give a complete global coverage within two 
days. 
This order of resolution lias been found very suitable for moni 
toring a climate feature such as desertification. Based on a physi 
cally-based model it has be shown that the briglitness tempera 
ture is directly related the to biomass and therefore the level of 
vegetation and used for monitoring seasonal and annuiti varia 
tions of vegetation in the Sahel region in Africa, for instance. It 
should be noted that passive microwave data (SMMR and 
SSM/I) are available for more than 20 years with day and night 
time observations and are likely to be acquired for another 20 
years. 
Likewise, it has been demonstrated that an airborne microwave 
radiometer can measure the tliickness of an oil slick on the ocean 
surface. Modelling shows that the brightness temperature vary 
with the oil tliickness since the film acts as a matching slab be 
tween the two media, sea water and air, resulting in an increase in 
brightness temperature relative to the surrounding surface. This 
is largely independent of waves. To overcome a thickness ambi 
guity the measurements are made simultaneously at minimum 
two frequencies. With a scanning system covering an oil slick it 
is in principle possible to determine the volume of tire slick. The 
techniques has been demonstrated experimentally, but an ambi 
guity arises due to the relatively coarse spatial resolution of the 
system (70 m at 17 GHz in a 2 000-m swath at 2 000-m altitude) 
and because an oil spill does not have a uniform tliickness within 
the footprint of the antenna beam. However, a volume underes 
timate of 10 to 20% that was found is not very important since 
the order of magnitude suffices in most cases (Skou et al., 1983). 
GENERAL CONSIDERATIONS OF DISASTER 
MONITORING 
The remote sensing techniques developed during the last many 
years have proven their value in a great number of applications 
and have reached a level of maturity so that it may be used opera 
tionally, the only major obstacle being that the repetition rate of 
observation may be insufficient for the application in mind. This 
appears to be the case with monitoring natural disasters. The 
techniques have limitations and give rise to ambiguities as 
pointed out above, but knowing them it is in my opinion still very 
reasonable to include it in a system designed for assisting the end 
users in managing natural disasters. 
“The technique is there - let us use it”. This, I expressed on a 
previous occasion after having witnessed still another series of 
presentations of examples of the use of remote sensing data in