reactors encouraged the migration of these particles to the eastern
part of the reservoir.
The image (from pair 1986 - 1988) shows the changes in the
ecological situation as of the third year after the accident.
The pine forests have undergone clear changes everywhere,
taking on a blotched appearance. The greatest changes can be
seen on the northern edges of the forests. There are no substantial
changes in the condition of deciduous forests. There is a clear
change in the dead 'red' forest, especially in those areas where the
trees were uprooted and soil removed.
The condition of vegetation in the town of Pripyat has
deteriorated noticeably. The most contaminated part of the
cooling water reservoir is that into which used process water is
released.
The image (from pair 1986 - 1995) shows the ecological changes
which have occurred up to the present day. Radionuclides
produced by human activity have had an impact on pine forest
everywhere, and the ecological
situation has improved only in places. The general condition of
deciduous forests has not seen any substantial changes. Grass
vegetation has proved to be the most resistance to the impact of
radionuclides. The water in the Chernobyl NPP cooling reservoir
has become noticeably cleaner, and the condition of vegetation in
the town of Pripyat has improved.
2.2 Classification
First of all, before starting the studying of environment state
influence on vegetation, it is necessary to divide vegetation on
homogeneous parts as precise as possible, using ground truth
and remote sensing data. This problem is very difficult because
of all the forests are mixture within the area and influence of
toxicants on some species of vegetation
is different. The landscapes classifiction had been carried out
using modules of ERDAS IMAGINE software. It was indicated
all types of landscapes on classified multiband images for
example on SPOT 23.08.95. Besides this it was shown the
peculiarities of distribution of some type of vegetation which are
absent on initial vegetation map, used for supervised
classification. For example, it was shown the heterogeneous
character of some kinds of vegetation cover which cause the
specific spot view of classified images. It was observed the
replacing of the pine tree forests with deciduous forests on the
edge of pine tree forests, along the roads etc. It was specified
limits of some forests with complicate boundaries. The big
variety of vegetation species was observed within the long-fallow
lands and meadows. But, the influence of radionuclide
pollution on spectral characteristics of vegetation more week
than influence on ones of species peculiarities.
2.3 The vegetation spectral reflectance research
The studying of spectral reflectance of vegetation of multiband
multitemporal images was carried out using set of ground truth
observation . The map of soil radioactive pollution with Cs-137,
created using soil sample analisys, had been used. The analysis
carried out for multitemporal images have shown the
dependencies between spectral characteristics of major
boiocenouses in the test sites in red and near-infrared bands on
the one hand and the soil content of CS-137 on the other. All the
Intemational Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998
plots show a quite simple unimodal relation between the
analyzed values. According to these plots one can conclude that,
the higher the contamination of soils by radionuclides, the higher
the spectral brightness of plant in each band. For pine tree forest
this relation is manifested most clearly (Lyalko et. al. 1996 . The
radionuclides influence content of chlorophyll in plant as well as
structure of cells. The maximum distinctions for anomaly and
background values of radionuclide content in soil are observed
for 1986 year and the minimum one for 1995 year. The reduction
of distinctions connected with increasing of background values
because of radionuclides migration. Over 10 years, it is going the
relaxing of radionuclides influence on plant, but vegetation is
not fully recovered.
It was developed another approach using the analysis of
determination of shift of red edge position (REP) of the one
dimensional plots of spectral reflectance and wavelength. The
values of spectral reflectance had been taken from Landsat TM
multiband image (29.05.1988) for sites with different level of
contamination by Cs-137. It is known the REP of spectral
reflectance is the source of information about ecological state
(Collins, et. al, 1983; Horler et. al. 1983; Demetriades-Shah et.
al., 1992; Baret et al., 1992; Lyalko et. al., 1992, et. al.). REP is
the value of wavelength in interval 680 -760 nm, where first
derivative of spectral plot is maximum, that is mean that second
derivative at this point is equal to zero. The approach of REP
determination had been developed using Landsat TM data for
spectral reflectance of pine tree forest. To examine the method,
the sites with concentration of Cs-137 from 5 up to 25 Ci/sq.km
were measured. For this sites, the plots of spectral reflectance
had been done using Landsat TM 2,3,4,5 band data. For plot
construction, according to method, 1t was supposed the possibility
of using of supplementary data which produced from initial data.
As a result the 11 points are used for plot construction with
values of wavelength from 520 up to 830 nm: 520, 560, 600,
610, 630, 660, 690, 710, 760, 830 nm. The centered values of
spectral brightness were used in points with wavelength 610 and
710 nm . This values are approximated using sum of Chebyshev
polynomials from 1 to 4 degrees (Sportjuk et. al. 1993). The
initial values and approximated curve are shown ın figure 1. For
approximated curve the first derivative of reflectance spectra had
been determined in interval 660 - 760 nm and the point of first
derivative maximum had been indicated. This approach had been
used for REP determination of average spectral curve obtained
from summarizing of 65 initial spectral curves for sites with
level of Cs-137 contamination in soil about 5 Ci/ sq.km. The
same procedure had been carried out for 7 spectral curves
connected with level of Cs-137 contamination 25 Ci/sq.km
approximately. For two average curves the plots of first
derivatives and REP are demonstrated in figure 2. The REP is
equal to 725 nm (for lowS Ci/sq.km) and 721.5 nm (for 25
Ci/sq.km) according to calculations. Of course, this result will be
tested for another vegetation species but it demonstrate potential
possibility of indication of polluted sites of homogeneous forests
using REP determination on the basis of Landsat TM data.
The relationship between the spectral brightness values of
various forms of vegetation and the Cs-137 content in the soils
was used as a basis for a map showing the contamination of soil
with toxicants, principally Cs-137, in the zone affected by the
integra
vegeta!
The la
around
of the
radion
relief |
determ
south «
trace €
associe
edges
concern
southe
relief.
valley
in the
barrier
The sj
map
investi
repres
posed
horizo
where
deposi
partici
active
of un
arounc
0.8 —-
50«
Figur
where
