Prediksi Iklim dan Prediksi Iklim dan Banjir DKI dengan AnfisBanjir DKI dengan Anfis

The Houw LiongThe Houw Liong

P.M.SiregarP.M.Siregar

R.GernowoR.Gernowo

H.WidodoH.Widodo

S. NuryantoS. Nuryanto

Garis BesarGaris Besar

• Model Model Iklim dan CuacaIklim dan Cuaca

• Hard Computing & Soft ComputingHard Computing & Soft Computing

• Prediktibilitas , Chaos/Weak CausalityPrediktibilitas , Chaos/Weak Causality

• Kaidah Fuzzy & ANFISKaidah Fuzzy & ANFIS

• Fuzzy ClusteringFuzzy Clustering

• Deret Waktu Bilangan Bintik Matahari, Deret Waktu Bilangan Bintik Matahari, Curah Hujan & Tinggi Muka Air Curah Hujan & Tinggi Muka Air

• KesimpulanKesimpulan

Model AtmosferModel Atmosfer

Model AtmosferModel Atmosfer

Physical ModelPhysical Model

• Cloud Formation : Arakawa , KuoCloud Formation : Arakawa , Kuo• Interactions : lands, oceans, cryosphere, Interactions : lands, oceans, cryosphere,

biosphere biosphere • Forcing : Solar activity, volcanic eruptions, Forcing : Solar activity, volcanic eruptions,

cosmic rayscosmic rays

•PredictabilityPredictability•Chaos, Attractor, weak causalityChaos, Attractor, weak causality

Predictability of a Climate Predictability of a Climate ModelModel

Forecasting Based on Soft Forecasting Based on Soft Computing & Solar cycleComputing & Solar cycle

• Quasi periodic solar cycleQuasi periodic solar cycle

• Sunspot , flare , cme, galactic cosmic Sunspot , flare , cme, galactic cosmic rays, interplanetary magnetic field rays, interplanetary magnetic field and weatherand weather

• ENSO & IODENSO & IOD

• MJOMJO

Sunspot & cosmic raySunspot & cosmic ray

Galactic Cosmic RaysGalactic Cosmic Rays

Cosmic RaysCosmic Rays

Kaidah Samar Sugeno Kaidah Samar Sugeno (Sugeno Fuzzy Rules) (Sugeno Fuzzy Rules)

• Untuk x adalah AUntuk x adalah Ai i dan y adalah Bdan y adalah Bjj maka z adalah pmaka z adalah piix + qx + qjjy + ry + rijij

• Kaidah Belajar /Kaidah Belajar /Learning Rules :Learning Rules :

• δδv_k = - v_k = - η∂η∂e_tot/∂v_ke_tot/∂v_k

Adaptive Neuro Fuzzy Adaptive Neuro Fuzzy Inference SystemInference System

A1

A2

B2

B1 N

N

layer 1layer 2

layer 3

layer 4

layer 5

x y

1w

2w

1w

2w

x y

ANFISANFIS

• Layer 1 Layer 1 ::

•

• x x andand y y areare input input of of ode -i anode -i andd O1,i O1,i isis membership function ofmembership function of fuzzyfuzzy set set A=(A1,A2A=(A1,A2) and B=() and B=(B1 B1 ,,B2B2 )) with with membership function membership function A A isis : :

• •

• ai,bi, ai,bi, andand ci ci areare parameter parameterss• Layer 2 : Layer 2 : output as the product of input output as the product of input

membership functionsmembership functions : :•

2

1,

1,

( ), 1, 2,

( ), 3, 4,

i

i

i A

i B

O x for i or

O y for i

b2

i

i

A

a

cx1

1)x(

2,1i)y()x(wOii BA1i,2

• Layer 3 Layer 3 inin node -i : node -i :

•

• Layer 4 : Layer 4 : Node -i Node -i is is adaptiadaptive ve node node with with funfunctionction node : node :

2,1i,ww

wwO

21

iii,3

)ryqxp(wfwO iiiiiii,4

ANFISANFIS

• Layer 5 : Layer 5 : final final output :output :

•

5

i ii

i ii i

i

w fO w f

w

sspot

0

50

100

150

200

1948

1952

1956

1960

1964

1968

1972

1976

1980

1984

1988

1992

1996

2000

2004

2008

2012

sspot

NASA : Prediction of Solar NASA : Prediction of Solar Cycle 24Cycle 24

Pontianak RegionCorrelation Sunspot vs Precip =0.88

0.00

50.00

100.00

150.00

200.00

1948

1951

1954

1957

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

Years

Sun

spot

/Pre

cip

-50.000.0050.00100.00150.00200.00

ave-sunspot ave-precip

Jaya Pura

0.0050.00

100.00150.00200.00250.00300.00350.00

1948

1951

19541957

19601963

19661969

1972

19751978

19811984

19871990

1993

19961999

2002

Years

mm

/mon

th

0.00

50.00

100.00

150.00

200.00

Avg precip sspot

Jakarta

0.00

50.00

100.00

150.00

200.00

250.0019

4819

5119

5419

5719

60

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

Years

mm

/mon

th

0.00

50.00

100.00

150.00

200.00

Avg Precip Avg-sspot

Fuzzy ClusteringFuzzy Clustering• Fuzzy c-means AlgorithmFuzzy c-means Algorithm

• Fix c (2Fix c (2cc n) and select a value for n) and select a value for parameter m’, initialize the partition matrix parameter m’, initialize the partition matrix UU(0)(0), membership functions and the centers . , membership functions and the centers . Each step in this algorithm will labeled r, Each step in this algorithm will labeled r, where r=0,1,2,..where r=0,1,2,..

• Repeat updating the partition matrix for Repeat updating the partition matrix for rrth th step,U step,U (r)(r) until until

• )()1( rr UU

Fuzzy ClusteringFuzzy Clustering

• set r=r+1 set r=r+1

• Calculate the new c centers :Calculate the new c centers :

n

kmik

n

k kjxmik

ijv

1'

1.'

• Calculate the new membership Calculate the new membership functionsfunctions

1)1'/(2

1 )(

)()1(

m

c

j drjk

drikr

ik

• set r=r+1 set r=r+1

• Calculate the new c centers :Calculate the new c centers :

n

kmik

n

k kjxmik

ijv

1'

1.'

Pengelompokan Samar (Fuzzy Pengelompokan Samar (Fuzzy Clustering)Clustering)

IOD from PAOMA ForecastsIOD from PAOMA Forecasts

Multivariate Enso IndexMultivariate Enso Index

Prediksi Curah Hujan Ciliwung Prediksi Curah Hujan Ciliwung Hilir (DKI Jakarta)Hilir (DKI Jakarta)

Ciliw ung Hilir Prediksi September 2009 - Maret 2010

0

2

4

6

8

Jan-04 Oct-04 Jul-05 Apr-06 Jan-07 Oct-07 Jul-08 Apr-09 Jan-10Jumlah Bulanan Rata-Rata

Cur

Huj

an (

x100

mm

)

Data Prediksi ANFIS

Prediksi Pentad Ciliwung Prediksi Pentad Ciliwung HilirHilir

Ciliw ung Hilir Prediksi Oktober 2009-November 2009

0

0.5

1

1.5

2

2.5

3-Jun-08 3-Oct-08 3-Feb-09 3-Jun-09 3-Oct-09PENTAD

Cur

Huj

(x1

00m

m)

Data Prediksi ANFIS

Prediksi Curah Hujan Ciliwung Prediksi Curah Hujan Ciliwung HuluHulu

Ciliw ung Hulu Prediksi September 2009 - Maret 2010

0

2

4

6

8

Jan-04 Sep-04 May-05 Jan-06 Sep-06 May-07 Jan-08 Sep-08 May-09 Jan-10Jumlah Bulanan Rata-Rata

Cur

Huj

an (

x100

mm

)

Data Prediksi ANFIS

Curah Hujan Pentad Ciliwung Curah Hujan Pentad Ciliwung HuluHulu

Ciliw ung Hulu Prediksi Oktober 2009 - November 2009

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

3-Jun-08 1-Oct-08 29-Jan-09 29-May-09 26-Sep-09PENTAD

Cur

Huj

(x1

00m

m).

Data Prediksii ANFIS

Analisis Hujan Ekstrim DKI Analisis Hujan Ekstrim DKI dengan WRFdengan WRF

Data Curah Hujan 2007 BMKG serupa Data Curah Hujan 2007 BMKG serupa

dengan Hasil WRFdengan Hasil WRF

KesimpulanKesimpulan

• ANFIS & ANFIS & Pengelompokan SamarPengelompokan Samar dapat digabungkan untuk prediksi/ dapat digabungkan untuk prediksi/ prakiraan prakiraan Iklim jangka panjang dan Iklim jangka panjang dan menegah.menegah.

• Prediksi Prediksi cuaca cuaca jangka pendek yang jangka pendek yang akurat memerlukan model atmosfer akurat memerlukan model atmosfer dan data radar & satellit cuaca.dan data radar & satellit cuaca.

Banjir 2002Banjir 2002

Ciliw ung Hilir Prediksi Agustus 2008 - April 2009

0

2

4

6

8

Jan-04 Jul-04 Jan-05 Jul-05 Jan-06 Jul-06 Jan-07 Jul-07 Jan-08 Jul-08 Jan-09Jumlah Bulanan Rata-Rata

Cur

Huj

an (

x100

mm

)

Data Prediksi ANFIS

Ciliw ung Hilir Prediksi September 2008-December 2008

0

0.5

1

1.5

2

2.5

3

3-Aug-07 3-Nov-07 3-Feb-08 3-May-08 3-Aug-08 3-Nov-08PENTAD

Cur

Huj

(x1

00m

m)

Data Prediksi ANFIS

Ciliw ung Hulu Prediksi May 2008 - April 2009

0

2

4

6

8

Jan-04 Jul-04 Jan-05 Jul-05 Jan-06 Jul-06 Jan-07 Jul-07 Jan-08 Jul-08 Jan-09Jumlah Bulanan Rata-Rata

Cur

Huj

an (

x100

mm

)

Data Prediksi ANFIS

Ciliw ung Hulu Prediksi May 2008 - Desember 2008

0

0.5

1

1.5

2

2.5

2-Sep-07 1-Dec-07 29-Feb-08 29-May-08 27-Aug-08 25-Nov-08PENTAD

Cur

Huj

(x1

00m

m)

Data Prediksii ANFIS

Banjir 2005Banjir 2005

PROBABILITASPROBABILITASINTENSITAS Curah HujanINTENSITAS Curah Hujan

INTENSITAS C.H. ( 1957 - 1988 ) JAKARTA (745 )

0

10

20

30

40

50

60

0-<=

5

>5-<

=10

>10-

<=15

>15-

<=20

>20-

<=25

>25-

<=30

>30-

<=35

>35-

<=40

>40-

<=45

>45-

<=50

>50-

<=55 >5

5

KELAS

PR

OB

(%)

INTENSITAS C.H. ( 1985 - 2003 ) BOGOR

0

10

20

30

40

50

60

0-<

=5

>5

-<=

10

>1

0-<

=1

5

>1

5-<

=2

0

>2

0-<

=2

5

>2

5-<

=3

0

>3

0-<

=3

5

>3

5-<

=4

0

>4

0-<

=4

5

>4

5-<

=5

0

>5

0-<

=5

5

>5

5

KELAS

PR

OB

(%

)

IODIODIOD

-0.60

-0.40

-0.20

0.00

0.20

0.40

0.60

194819511954

19571960

19631966

196919721975

19781981

198419871990

19931996

19992002

200520082011

20142017

Years

IOD

Solar activities &Solar activities & ClimateClimate

Microphysics Cumulus

Solar and its activities PBL

Earth Surface

surface T,Qv,Wind

Surface FluxSH,LH

IncomingSW,LWSurface

Emisi/albedo

Cloud FractionCloud Effects

Cloud detrainment

Elnino-Lanina Years

0,00

50,00

100,00

150,00

200,00

1948

1950

1952

1954

1956

1958

1960

1962

1964

1966

1968

1970

1972

1974

1976

1978

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

Years

Su

nsp

ot

Nu

mb

er

sspotLanina Elnino

Positive-Negative of Indian Dipole Mode Years

0,00

50,00

100,00

150,00

200,00

1948

1950

1952

1954

1956

1958

1960

1962

1964

1966

1968

1970

1972

1974

1976

1978

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

Years

Su

ns

po

t N

um

be

r

sspotNegative Positive