For monitoring bioproduction in the lakes, information
were first assessed separately for each scene using
the index ‘Mean’ (IRS1,IRS2). For this purpose
summary statistics of the index were calculated based
on all core lake areas for the individual dates (see
tab.3). In order to get a better understanding about the
differences between the lakes in one scene, the
frequency distribution of the means of all core lake
areas was analyzed for this index which led to a
classification of the lakes into equidistant classes. Only
exception were the minimum and maximum values
which form two broader classes representing the
extreme values (see fig.5) These results and the
comparison between the different dates give a first
impression of the seasonal behavior of the lakes.
In a second step the temporal changes between the
dates were further analyzed by calculation of the
differences between the values of the index for
consecutive datasets. Table 4 shows the summary
statistics of these differences which were also
classified based on histogram analysis (see fig.6)
following the same principles as for figure 5.
Information of all scenes available for the observed
time were combined to distinguish between lakes of
similar seasonal behavior. Spatial variations of
calculated indices within a dataset and variations
between scenes are caused by differences of lake
water properties. Thus, the principal component
analysis (PCA) could be used for assessing these
differences in a compressed way, because the main
variation within the data is represented by the first
principal component (PC). The first PC allows grouping
the lakes into classes of similar seasonal behavior of
bioproduction. The input data for the PCA were the
means (IRS1,IRS2) of all 6 acquired scenes.
4.4 Results for analysis of individual dates
The approach is based on the relation between field
measurements and the spectral properties of the lakes
and intends to show the relative spatial and temporal
changes of bioproduction in lakes. Figure 4 shows the
relationship between ground measurements of
chlorophyll-a content of the 3 lakes Gr. Wummsee, Gr.
Zechliner See, Braminsee and calculated indices for
similar dates (see tab.2).
85 T A Braminsee
e 307 a ô Ó
9 25 +
E 20
2 LS teur ses A Difference (IRS 1, IR S 3)
= 190 108 OMemn(RS1,IRS3)
a ; Gr. Wumms ee i
0 20 40 60 80 100
Chlorophyll-a concentration in pg/l
Fig. 4: Relation between calculated indices and
measured chlorophyll-a for three lakes for comparable
dates in 1997 (May, June, Sept.,1, Sept.,25)
It is assumed that the spectral properties of the other
lakes follow this relationship and can be used for
relative differentiation of chlorophyll-a content between
the lakes observed during the whole time.
Based on relation between the calculated means
(IRS1,IRS2) and chlorophyll-a content the
differentiation of high or low chlorophyll-a contents can
be assessed for all lakes.
MEAN Aug| Feb 2|May 4| Jun 2| Sep 1 Se
anstansm I i d B 97 Yor 97 "o7 25
96 97
Mean 10,9 6,5 9,9 8,51 11,4 8,7
Min 32 27 2,8 0,7 2,7 3,9
Max 24,7| 12,8| 29,55| 29,1] 32,7] 41,3
Stdev 4,6 2,1 5,3 4.8 6,3 6
Tab. 3: Summary statistics for Mean (IRS1,IRS2) of all
core lake areas for each dataset
A comparison of the summary statistics of the means
(IRS1,IRS2) for the different datasets (see tab. 3)
shows significant differences between the dates. The
minimum in February corresponds to a low level of
bioproduction during winter time. The two maxima in
August and early September are in accordance to
periods of high bioproduction in summer. The mean
value for May indicates a first algal bloom in spring.
The data also show a greater variation between the
lakes for the same date than for one lake through all
dates. The lowest variation can be observed in
February representing the time of minimal
bioproduction in all lakes. However, all of the other
dates show high and widely varying values which
reflect the big range of trophic state conditions in the
study area.
— $—— Augl8 96
— —- —Feb02 97
- - - À- - - Ma/04 97
—9—— Ju0297
——3—— S8p01 97
——68—— $8025 97
— N
C1 ©
cp po
Number of lakes
S
Fig. 5: Frequency distribution of Mean (IRS1,IRS2)
calculated from means of all core lake areas
4.5 Results of multitemporal analysis
For characterizing the behavior of lakes in the study
area the relative changes between different dates (see
fig.5) were analyzed as well as the absolute amount of
changes (see fig.6) between dates.
Fig. 5 shows a significant increase of the number of
lakes characterized by higher values between February
and May which continues partly until June, and further
until the beginning of September. Between May and
132 International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998
Jur
dec
are
tha
dec
will
dev
anc
dat
Anc
bio
cor
val
Ch
Me
IF
St
Tat
Fig.
Eac
vari
ran
dev
cha
19€
ma
win
enc
valt
97)
whi
Jun
blot
The
cha
the
am
ats
