Google+ Followers

28 January 2010

Penemuan Pengetahuan Memakai Model Neuro-Fuzzy

Kolokium di Lapan, Bandung
Kamis, 18 Februari 2010
Pukul : 10:00 – 11:00

Knowledge Discovery Using Neuro-Fuzzy Models
Case Study : Weather/Climate and Space Weather

Penemuan Pengetahuan Memakai Model Neuro-Fuzzy
Studi Kasus : Cuaca/Iklim dan Cuaca Antariksa

The Houw Liong
KK Fisika Sistem Kompleks, FMIPA, ITB

Abstrak

Hakekat sains ialah menemukan pengetahuan yang terandalkan (model konseptual,model empiris, model fisis, kaidah,hukum sebab akibat,teori) dari data pengamatan dan mengujinya dengan melihat kecocokannya antara prediksi pengetahuan tsb dengan data pengamatan . Metoda ini sekarang dikenal sebagai Data Mining atau Knowledge Discovery from Data (KDD) yang didukung oleh metoda statistik (clustering, regresi, korelasi, PCA, …) dan metoda inteligensi artificial (ANN, fuzzy logic ,neuro-fuzzy, SVM,….) .
Studi kasus diambil dari cuaca/iklim dan cuaca antariksa untuk melakukan prediksi jangka panjang, jangka menengah dan jangka pendek, khususnya untuk mengantisipasi curah hujan ekstrim yang mengakibatkan banjir besar di Indonesia dan super storm dari aktivitas matahari yang dapat mengakibatkan kelumpuhan telekomunikasi dan navigasi di Indonesia yang diperkirakan mungkin bisa terjadi pada akhir tahun 2012 atau awal tahun 2013 atau pada puncak aktivitas matahari di masa depan.

22 January 2010

Prediksi Curah Hujan daerah Jakarta dengan ANFIS

Prediksi Curah Hujan daerah Jakarta berdasarkan ANFIS

R.Gernowo, The Houw Liong, F. H. Widodo, S. Nuryanto

 

Prediksi Curah Hujan bulanan dan pentad daerah Ciliwung hilir dan hulu berdasarkan ANFIS

Puncak hujan terjadi pada akhir minggu Januari dan bulan Februari yang biasanya menimbulkan banjir tahunan di daerah Jakarta.

 

 

 

 



 

 

 

 

 

 

 

 

 

 

 

 

 



 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


14 January 2010

CLIMATE MODEL BASED ON SOLAR ACTIVITY

CLIMATE MODEL BASED ON SOLAR ACTIVITY
The Houw Liong
P.M. Siregar
R. Gernowo
F. H. Widodo

The galactic cosmic rays collide with air molecules in the upper atmosphere and produce secondary particles. Generally the charged particles so produced cannot penetrate to lower layers of the atmosphere, except gamma ray, neutrons and the muons. When gamma ray, neutrons and muons interact with the air molecules or water molecules, they become charged and together with aerosols particles act as condensation nuclei for the formation of clouds. The cosmic ray becomes the source of ions in the air besides radiation coming from earth originated by the radio isotope radon.
During the sunspot minimum, the intensity of the galactic cosmic ray that penetrates earth atmosphere becomes maximum which in turn increases the coverage of clouds. This implies that solar irradiation reaching the earth will be minimized. Conversely, during solar activity maximum or sunspot maximum, the intensity of galactic cosmic ray reaching lower levels of the atmosphere decreases, less cloud condensation nuclei are produced, hence the cloud cover decreases, furthermore extra energy received from flares during prominent eruptions, maximizes the amount of solar energy reaches the earth.
Although global cloud cover produces a warming effect or the greenhouse effect, but a cooling effect due to reflections against direct solar irradiation is more dominant factor [Svensmark,2007].
Furthermore during solar activity maximum, the intensity of ultraviolet that penetrates the earth increases. Solar activity maximum usually is followed by increasing coronal mass ejection. Both effects caused greater amount of energy penetrates the earth and this will influence the climate through the dynamics of the atmosphere and oceans.
Using rainfall data in Indonesia from NCEP Reanalysis at
http://www.cdc.noaa.gov/cdc/data.ncep.reanalysis
and sunspot numbers time series, we can get relations between sunspot numbers and rainfall in various Indonesian regions. The determination of sunspot numbers on yearly basis against the yearly rainfall for various regions in Indonesia based on time series data are shown in Figures 3, 4 and 5 .
From Figure 3 we can conclude that eastern Indonesia (Jayapura region) which represented Eastern Indonesian Maritime Continent is strongly influenced by ENSO.
After 1976 sunspot numbers maximum SMax and sunspot numbers minimum SMin correspond to precipitations above normal also to La Nina and maximum eruptions CME corresponding to precipitations below normal and also to El Nino.[4, 5] In Pontianak region which represent middle Indonesian Maritime Continent, the yearly precipitation is mainly determined by sunspot cycles (Figure 4). Precipitations above normal occur at sunspot maximum SMax, and precipitations below normal at sunspot minimum SMin. Precipitations in east Indonesia which represent North Australia Indonesian Monsoon are influenced by ENSO similar to those observed in Jayapura region. (Figure 3) Precipitations in Jakarta region or Jabodetabek are weakly influenced by ENSO. The peaks of yearly precipitations correspond to the peaks of sunspot numbers, but at the sunspot numbers minimum which correspond to galactic comic ray maximum, the yearly precipitations also maximum.(Figure 5).
The west Indonesian region is mainly influenced by IOD that also correlated to solar cycle. [5]
By analyzing monthly rainfall time series in Jakarta region, we can predict 6 months ahead of time (Figure 8.) and for short range predictions we can use pentad rainfall time series and WRF model to anticipate extreme rainfall in Jakarta region


CONCLUSION

The accuracies of very long range predictions (more than ten years ahead) of climate models are very poor due to the chaotic nature of the atmosphere-ocean system.
A climate model using ANFIS and fuzzy clustering can be used to predict extreme rainfall in Indonesian regions.
The fuzzy c-means clustering shows that the western Indonesian region is influenced mainly by IOD, the eastern Indonesian region is influenced mainly by ENSO and the middle region is mainly influenced by solar activity.
So, by knowing sunspot number time series as predicted by ANFIS and fuzzy clustering of climate regions we can predict the coming extreme weather for each regions in Indonesia.
By analyzing monthly rainfall time series in Jakarta region, we can predict 6 months ahead of time and for short range predictions we can use pentad rainfall time series and WRF model to anticipate extreme rainfall in Jakarta region.

06 January 2010

NUMERICAL CLIMATE MODEL in Indonesian Region

NUMERICAL CLIMATE MODEL in Indonesian Region
The Houw Liong
R. Gernowo

Climate model can be constructed by using the law of physics for the atmosphere i.e.: The Navier-Stokes equation, the conservation of mass, the conservation of energy, the equations of states, including schemes for cloud formations, carbon and sulfur cycle, interactions between atmosphere and land surface, oceans, cryosphere, and biosphere, furthermore we have to include forcing by volcanic eruptions, the solar activity and galactic cosmic rays.
In Indonesia scientists from LAPAN [Ratag, 2002] use Global circulation Model (GCM) with 9 levels vertical resolution and 3.20x5.60 horizontal resolution based on sigma coordinates. Limited local area model (DARLAM) is used to obtain higher resolution 25 km x 25 km. The model also has to include schemes for cloud formations, carbon and sulfur cycle, and interactions between atmosphere and land surface, oceans, cryosphere, and biosphere.
With a scenario that the concentration of CO2 will be doubled in 100 years, the temperature anomaly can be calculated. In 50 years from 1990 most Indonesian regions the temperature very likely will raise by 0.5 to 1.0 degrees Celsius. Some regions the temperature very likely will raise by 1.0 to 1.5 degrees Celsius. This climate model cannot predict the precipitations in Indonesian regions well. However the verification of this forecasting is not convincing because the in Indonesia many good weather stations are located in cities. From these stations the average increase of temperature from 1860 to 2000 is about 10 C, but actually the increase is due to urban warming, not global warming.
The forecasting of rainfall is still poor, for instant using rainfall data from Bandung the best forecasting is obtained using cloud formation according to Arakawa scheme the correlation with the data is 0.46.
It needs some modification before it can be use to predict rainfall well.
Furthermore a realistic model has to include forcing by volcanic eruptions, the solar activity and galactic cosmic rays which is poorly represented in climate model.
Due to the nature of chaotic system [Lorenz, 1960] and the difficulties to formulate complete model, and to get accurate data of boundary conditions and initial conditions are difficult, the accuracy of long term prediction is poor.
The satellite data is regarded even by NOAA administration to be the most reliable but they don’t use it in releases as it is only available for 30 years. It has shown a cooling since 2002.(Figure 6.)
Short range prediction according to WRF (weather research and forecasting), a regional numerical weather model developed by Pennsylvania State University/ National Center for Atmospheric Research (PSU/NCAR), with a horizontal grid resolution of 5 km. (Figure 7.)
By inspecting the model output in the area if interest, it is found that the largest concentration of convective rainfall over Jakarta and other area in the northern coast of West Java occurred particularly during 31 January - 2 February 2007 UTC (Universal Time Coordinated), as shown in Figure 7 & 9. It is important to note that the daily-accumulated rainfall on 1 February 2007 was more concentrated in land and sea area, while almost all of the rainfall on 31 January 2007 was distributed along the coastal area. More interestingly, large convective rainfall occurred in both inland and Jakarta coastal area on 1 February 2007.