84
F. Recknagel
Tab. 5.3. Multivariate time series database of 9 lakes
Lakes
Years
Sampling
No ofWater
Noof
No of
Frequency
Quality
Phytoplankton
Zooplankton
Parameters SoeciesJGrouos Soecies/Grouos
Biwa
1984-91
weekly to
10
21
-
(Japan)
monthly
Burrinjuck
1976 - 97
weekly to
8
8
6
(Austra1ia)
monthlv
Kasumigaura 1984 - 93 fortnightl y
10
10
3
(Jaoan)
to monthlv
Myponga
1970 - 97
weekly to
10
25
-
(Australia)
monthlv
Saidenbach
1979-84 fortnightly
9
5
1
(Germanv)
Soyang
1984 - 99
monthly
7
12
12
(Korea)
Tuusulanjaervi 1972 - 87 fortnightly
7
10
-
(Finland)
to monthlv
Veluverneer
1976-99
weekly to
11
14
6
(Holland)
monthlv
Wolderwijd
1976-99
weekly to
11
14
6
(Holland)
monthlv
INPUTSX;
INITIAL POPULATION GENERATION I
GENERATION j
Orthophos phatcmcJI 0
Nitrate mcII
0
WaterTe mperalure oe 0
SeuhiDe pthm
0
Solar Ra diatioR Jlcmlld 0
pH
0
Rotlrera Indlvldualsll 0
Cladocer • individualsll 0
indhidualsfl 0
I
Y"", (X .... X~
I
y .tl (Xl""X~
IChromosomc 11
IChromosonM: 11
I
y .f,(X""x~
IChtomosome 21
k:hromosomc 1J
I
Y"'C3 (Xt,,·XJ
IChromosome JI
IChrofDOllOOle 31
I
IChromosome41
IChromosomc41
W
y =f" (Xl"'X~
~
Crossoycr
N
I y. fN(X''''X~ I ~hromosomc~ Ehromosome ~
LEARNING ALGORITHM
ERROR (Y OBS • Y CALe)'
I: a) ~ € { N evolvOO artilicial neural networks )
b) ~ € { N evolvOO dllTe ... ntiaJ equations )
c) f, € {N evolvOO IF ... THEN .•. ELSE· rule sets I
IChromosome 11
lChromosome 21
IChrollJ05ODlCJI
IChromosome 41
Ehromosomc ~
r
CALCULATED
OUTPur~ALe
Aleat
Abundance
OBSERVED
OUTPUT",.,
Alpl
Abundante
Fig. 5.5. Application of evolutionary computation to evolve ANN (a). EDE (b)
and ER (c) from lake databases to be embodied in algal specific agents