rrors in
winter
period,
| under
2» radar
d Blyth
iver in
atellite
) flood
id the
s, with
od: 30
anuary
' aerial
> three
en and
nd the
f gray
; to be
to the
mages
aerial
of the
at the
stream
t used
f river
cost-
in. the
1al or
n situ
:nsing
areas
on, no
urrent
ically
detect
snsor.
single
on of
hat if
exist,
lation
board
band,
nages
ation
nhoff
| HH
8°. A
e the
poral
data
uring
eside
n the
ding
1 the
SAR
A COST-EFFICIENT APPROACH WITH
METEOROLOGICAL SATELLITES
In large river catchments or in the case of large-scale flooding
meteorological satellites alone can provide useful information,
provided the cloud cover is not significant or not stagnate.
NOAA AVHRR images are widely used for meteorological
purposes and for different analysis over sea and ocean, but its
use over land surface is strongly limited by its sparse spatial
resolution of 1.1 km at nadir. Hien (1988) describes an approach
for utilizing AVHRR data for flood monitoring. In this method
four channels of the AVHRR sensor are used to differentiate
between water bodies and land, to classify the cloud cover and
to eliminate the effects of some types of clouds. A method
applied widely uses the ratio of Channel 1 (visible) and Channel
2 (near-infrared) to distinguish between water and land. This
evaluation is based on the differences of the albedo of water and
land surface in the two channels. In the visible channel, water
has higher albedo than land, while in the near-infrared channel,
land has higher albedo than water. The ratio of CH2/CH1 thus
can be used to enhance the results of the water-land
classification from AVHRR images. The effect of cloud cover
can be reduced by using the split-window channels of the sensor.
When cloud contamination is not serious (thin clouds), some of
the cloud influence, including shadows of the clouds, can be
removed. A precise DEM and/or detailed topographic maps are
needed to make reliable estimates on water bodies from the
AVHRR image. High-resolution optical images (Landsat, SPOT,
etc.) can help in gathering other information like land use
classification. This method is presently being tested in a study
area of Thailand and the Mekong River delta. The validation is
planned on detecting different water bodies on the AVHRR
image, including lakes, reservoirs, the coastal line and flooded
rivers. Images are supplied through the receiving station at the
Asian Institute of Technology in Bangkok, Thailand, which
started its operation in October 1997. The first step of validation
process is to correlate the results of the classification process for
water body detection with known parameters, such as the area
of lakes and reservoirs, and the coastal line of Thailand. Further
steps in the application to be carried out are (a) to combine the
results with active microwave images and make a comparison
of classification during the yearly flooding and (b) to couple the
data derived from the analysis with hydrological models used
for flood simulation in Thailand.
CONCLUSIONS
Flood monitoring is an essential tool for planning of emergency
relief and repairs to communication, transportation and other
services and for the production of river flood risk maps for
future reference. Flood monitoring and the detection of water
bodies are important also for coupling the data to hydrological
models of flow routing and flood forecasting on a catchment
scale. By their nature, most floods occur in bad weather
conditions, which can severely restrict the use of aircraft, and
extensive cloud cover precludes the use of most earth observing
satellites, which rely on operating at optical wavelength. The
synthetic aperture radar (SAR) can penetrate clouds and
darkness, and therefore is an promising tool for flood mapping.
In this paper, the multisensor approach of remote sensing — the
analysis of images obtained by different sensors — in case of
flood assessment is discussed. In large areas of the Earth, aerial
photography is unavailable, and in situ measurements do not
supply enough data for mapping applications. In these cases,
satellite remote sensing can partly solve the problem, helping to
collect information on the inundated areas over large regions.
The initial conditions, mapped with high-resolution images of
optical sensors (Landsat, SPOT, etc) and the data from
meteorological, optical or radar images taken during the
flooding can be analyzed in a suitable computer software
system, often combined with a GIS. Four main characteristics of
satellite imagery that highly influence the application of
satellite-based remote sensing for flood monitoring are: (1)
spatial resolution (or pixel size) of the image, (2) return period
of the satellite, (3) availability of images and (4) costs of the
data. Further characteristics of importance are wavelength, view
angle and polarization (microwave systems) of sensors.
Different analysis show that the inundated areas are usually
underestimated, mainly because of insufficient image-resolution,
but proper comparisons are difficult to make, because ground
truth measurements are often missing. As a cost-efficient
approach further research on utilizing meteorological satellites
for inundation mapping, such as that discussed in the paper is
urgently needed.
Satellite remote sensing is widely used in some fields, like
meteorology, but the problem of operative application in flood
monitoring is yet to be solved. However, with the advent of the
planned satellites that can cover the Earth more frequently and
sensors designed directly to measure hydrological parameters,
space technology may be used for flood monitoring with a
higher accuracy around the globe in the future.
REFERENCES
Biggin, D.S.; Blyth, K. (1996). A comparison of ERS-1 satellite
radar and aerial photography for river flood mapping. Journal of
Water and Environmental Management, 1996, vol. 10, no. 1, p.
59
Boni, G.; Conti, M.; Dietrich, S.; Lanza, L.; Marzano, F.S.;
Mugnai, A.; Panegrossi, G.; Siccardi, F. (1996). Multisensor
observations during the flood event of 4-6 November, 1994 over
Northern Italy. Remote Sensing Reviews, 1996, vol. 14, pp. 91-
117
Engman, E.T. (1996). Remote sensing applications to
hydrology: future impact. Hydrological Sciences, 41 (4) August
1996, pp. 637-647
Giacomelli, A.; Mancini, M.; Rosso, R. (1995). Assessment of
flooded areas from ERS-1 PRI data: an application to the 1994
flood in Northern Italy. Phys. Chem. Earth, 1995, vol. 20, no. 5-
6, pp. 469-474
Hien, HM. (1998). Use of NOAA/AVHRR data for flood
disaster monitoring in the Mekong river delta, southern Vietnam.
Proc. International Symp. on Information Technology Tools for
Natural Disaster Risk Management, INCEDE, University of
Tokyo, March 1998, pp. 2-2-1 to 2-2-9
Imhoff, M.L.; McCandless, S.W. (1988). Flood boundary
delineation through clouds and vegetation using L-band space-
borne radar: a potential new tool for disease vector programs.
Acta Astronautica, 1988, vol. 17, no. 9, pp. 1003-1007
Planet, W.G. (editor) (1988). Data extraction and calibration of
TIROS-N/NOAA radiometers. NOAA Technical Memorandum
NESS 107 — Rev. 1, U.S. Department of Commerce, 1988
Intemational Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 783
EN
X a
