Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986
Water quality monitoring of Lake Balaton using LANDSAT MSS
data
H.Shimoda, M.Etaya & T.Sakata
Tokai University Research & Information Center (TRIC), Tokyo, Japan
L.Goda & K.Stelczer
Research Center for Water Resources Development (VITUKI), Budapest, Hungary
ABSTRACT: Water quality monitoring of Lake Balaton in Hungary was studied using LANDSAT MSS
data. Ground truth measurements were done simultaneously with the data acquisition of MSS
data and fourteen items of water qualities were measured on the lake. After certain pre
processing of MSS data, linear multi regression analyses were made between MSS data and
ground truth data. Nine items among the water qualities showed correlations to the MSS
data, especially transparency, chlorophyll-a, UV extinction and oxygen saturation showed
sufficient and strong correlations. These four kinds of water quality patterns were clearly
extracted.
1. INTRODUCTION
A water management has become one of the
most important element for our life now.
In this field, it is necessary to know the
present status of water qualities
periodically in order to control the water
resources. A sattelite remote sensing is
a powerful tool in this field.
In Hungary, studies of water managements
are being done very actively. Lake
Balaton is the largest water source in
this country. The regulation of this lake
is a continuous task of undiminishing
importance. In this report, a joint
research for water quality monitoring of
Lake Balaton using LANDSAT MSS data by
TRIC in Japan and VITUKI in Hungary is
described.
2. STUDY AREA AND DATA ACQUISITION
In this study, the target area is the west
part of Lake Balaton including Keszthely
and Fonyod. System corrected LANDSAT MSS
image covering the study area, acquired on
the 2nd of July, 1981, was used for this
study. Lake Balaton and the target area
are shown in Figure 1 and Figure 2,
respectively.
Ground truth measurements on the lake
were done simultaneously with the data
acquisition of MSS data. Fourteen items
of water qualities were measured at thirty
points shown in Figure 3. These items are
as follows.
1) Transparency (TR)
2) Suspended solid concentrations (SS)
3) Chlorophyll-a (CH)
4) Water temperature (TM)
5) Oxygen saturation (OX)
6) Light energy on the water surface
(LE)
7) Light energy reflected from the
water surface (RF)
8) Light energy penetration (PE)
9) UV extinction (UV)
10) Acid soluble phosphorus
concentrations (PS)
11) Acid soluble calsium concentrations
(CA)
12) Acid soluble magnesium
concentrations (MA)
13) The ratios of acid soluble calsium
and phosphorus (CP)
14) The ratios of acid solbule magnesium
and phosphorus (MP)
3. PREPROCESSING
The first step of data analyses was a
preprocessing. In LANDSAT MSS data, there
exists striping noises which were caused
by the differences among responses of six
detectors. This striping noise has big
influences especially on a water quality
monotoring because of low reflectances of
waters. These noises were carefully
eliminated using following three
algorithms.
1) Mean and standard deviation matching
2) Histgram equalization
3) Random noises addition.
The method of 1) or 2) is generally used
to eliminate striping noises from LANDSAT
MSS image. However, these method can not
fully eliminate scan line noises, because
of quantization errors. In this study,
these residual scan line noises were
corrected by the method of 3) in order to
improve the image quality. Random noises
cancel the quantization errors and allow
to make histograms continuously for each
detectors. Figure 4 and Figure 5 show the
original MSS image and destriped image.
4. IMAGE ANALYSES AND RESULTS
Image analyses were made by TIAS(Tokai
Image Analysis System) 2000.
Multi regression analyses were made
between fourteen items of ground truth
data and four kinds of image signatures,
i.e. original MSS values, mean values of 3
x3 pixels window, normalized values within
4 bands and ratios of band 4 and band 5.
