Setiawan, Nugroho Syarif (2019) Filtering Data Time Series Magnetotelurik Menggunakan Empirical Mode Decomposition (EMD). Other thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
03411540000013_UNDERGRADUATE_THESES.pdf Download (4MB) | Preview |
Abstract
Salah satu masalah penting dalam pengolahan data Magnetotelurik (MT) untuk diselesaikan adalah cara untuk melemahkan atau menghilangkan noise pada data. Noise dapat mengganggu informasi primer pada data Magnetotelurik sehingga menyebabkan kesalahan dalam memberikan gambaran persebaran resistivitas bawah permukaan. Penelitian ini menerapkan filtering Empirical Mode Decomposition (EMD) pada data time-series Magnetotelurik hingga dihasilkan komponen-komponen Intinsic Mode Functions (IMFs). IMF ini memiliki sifat yang adaptif dan dapat menekan sifat non-stasioneritas yang disebabkan oleh keberadaan noise. Data kemudian diubah ke domain frekuensi dengan transformasi Fast Fourier. Spektrum dalam domain frekuensi dibandingkan antara data yang dilakukan proses filtering EMD dengan data yang langsung dilakukan proses transformasi Fast Fourier. Proses pengolahan data dilanjutkan sampai ke inversi 1D untuk mengetahui pengaruh prose filtering EMD pada hasil persebaran resistivitas bawah permukaan. Pengolahan data time-series MT menggunakan filtering EMD secara keseluruhan menghasilkan kurva time-series yang lebih halus dan mampu menekan sifat non stasioneritas yang disebabkan oleh adanya gangguan pada perambatan gelombang EM atau noise yang terekam saat pengukuran. Hasil model cross section 1D yang didapatkan juga memiliki kecenderungan pola persebaran resistivitas yang mirip dengan model validasi, meskipun dari segi nilai resistivitas dan kedalaman masih agak berbeda.
=================================================================================================================================
One of problems in Magnetotelluric data processing is how to reduce the effect of noise which can disturb the primary information in the data and may lead into incorrect resistivity distribution imagary. In this paper, we propose a processing method using empirical mode decomposition (EMD) in original of Huang. This process used to filter out Magnetotelluric time series data from high frequency noise and resulting smooth frequency spectrum. The signal itteratively decomposed into several elementary oscillations called intrinsic mode functions (IMFs). These IMFs are local averages which are extracted through effective and simple collection of algorithms, also have adaptive characteristic and could pressed the non-stationarity caused by the presence of the noise. The filtered data transformed into frequency domain using fast Fourier transform, then we calculate the values of impedance, phase, and apparent resistivity. Obtained result data then compared with the data that processed without EMD filtering. These processes done until 1D inversion to see the differences at resistivity distribution model. MT time-series data processing using EMD gives smoother smoother time series curve in totality, also could suppress the non-stationarity character caused by disturbance at EM waves propagation or by noise that recorded at data acquisition process. The resulting 1D cross section model gives resistivity distribution pattern that mostly the same with validation model, although the resistivity values and the depth are slightly different.
| Item Type: | Thesis (Other) |
|---|---|
| Additional Information: | RSGf 538.3 Set f-1 2019 |
| Uncontrolled Keywords: | Empirical Mode Decomposition (EMD), Intrinsic Mode Function (IMF), Magnetotelurik |
| Subjects: | Q Science > QC Physics > QC446.3.H37 Harmonics (Electric waves) Q Science > QC Physics > QC610.3 Electric conductivity Q Science > QE Geology > QE601 Geology, Structural |
| Divisions: | Faculty of Civil Engineering and Planning > Geophysics Engineering > 33201-(S1) Undergraduate Thesis |
| Depositing User: | Nugroho Syarif Setiawan |
| Date Deposited: | 30 Apr 2024 02:19 |
| Last Modified: | 30 Apr 2024 02:19 |
| URI: | http://repository.its.ac.id/id/eprint/64417 |
Actions (login required)
 |
View Item |