Hakim, Arief Rachman (2016) Pemodelan Spatial Extreme Value Dengan Pendekatan Max-Stable Proses (Studi Kasus: Pemodelan Curah Hujan Ekstrem di Kabupaten Ngawi). Masters thesis, Institut Technology Sepuluh Nopember.
Preview |
Text
1314201212-Master_Thesis.pdf - Accepted Version Download (3MB) | Preview |
Abstract
Kejadian ekstrem adalah suatu fenomena berskala pendek yang jarang terjadi dan
sulit dihindari, namun memberikan dampak yang cukup besar. Indonesia sebagai
daerah tropis ekuatorial mempunyai variasi curah hujan yang cukup besar.
Extreme Value Theory (EVT) digunakan untuk mengidentifikasi kejadian ekstrem
yang bersifat univariat. Max-Stable Proses (MSP) dan Spatial Extreme value
(SEV) adalah metode untuk mengidentifikasi kejadian ekstrem pada kasus
multivariate dan melibatkan unsur spasial. MSP menggunakan pendekatan
distribusi Generalized Extreme Value (GEV). Estimasi parameter distribusi GEV
menggunakan metode Maximum Pairwise Likelihood Estimation (MPLE).
Penelitian ini bertujuan untuk memodelkan SEV dengan pendekatan MSP pada
studi kasus data curah hujan ekstrem di Kabupaten Ngawi. Data yang digunakan
untuk menyusun model dan estimasi parameter adalah data curah hujan tahun
1991-2011, sedangkan untuk validasi model menggunakan data tahun 2012-2015.
Dependensi antar lokasi pengamatan ditunjukkan melalui plot koefisien ekstremal.
Nilai koefisien ekstermal berkisar antara 1,3 sampai 1,5, hal ini berarti terdapat
dependensi spasial. Validasi model dilakukan dengan mengestimasi return level
tahun 2012-2015 melalui pendekatan model Smith, Schlather, Brown-Resnick.
Hasil validasi, diketahui bahwa model Smith lebih baik daripada model Sclather
dan Brown-resnick. Hal ini ditunjukkan oleh nilai RMSE dari return level model
Smith, Schlather dan Brown-Resnick berturut-turut sebesar 25,317, 29,376,
33,477.
====================================================================================================
Extreme events is a short scale phenomena that are rare and hard to avoid, but it
gives a considerable impact. Indonesia as the equatorial tropical areas have
rainfall variability is large enough. One attempt to minimize the impact of extreme
rainfall loss was to determine the patterns and characteristics of extreme rainfall
events, so early anticipation can be done. Extreme Value Theory (EVT) used to
identify extreme events that are univariate. Max-Stable Process and spatial
extreme value are statistical methods for analyzing extreme events on multivariate
case and involve spatial element. The approach used in Max-Stable Process is
Generalized Extreme Value (GEV) distribution. Estimate parameters of GEV
distribution using Maximum Pairwise Likelihood Estimation (MPLE) method.
This study aims to modeling spatial extreme using Max-Stable Process on case
studies of extreme rainfall in Ngawi Regency. The data used to construct the
model and parameter estimation is the rainfall data of 1991-2011, and for the
validation of the model using the data of 2012-2015. The dependencies of rainfall
intensities across location were indicated by extremal coefficient plot. The
resulting extremal coefficient value is in the range of 1,3 to 1,5, it means there is a
spatial dependencies. Model validation is done by estimating the return level in
2012-2015 by Smith model, Schlather model, Brown-Resnick model approach.
The Results validation Smith is known that the model is better than the other two
models for prediction of return level using RMSE. RMSE value Smith model is
25,317, Schlather model is 29,3761 and Brown-Resnick model is 33,477.
| Item Type: | Thesis (Masters) |
|---|---|
| Additional Information: | RTSt 519.24 Hak p |
| Uncontrolled Keywords: | Spatial extreme value, Max-Stable process, Smith, Schlather, Brown-Resnick, return level |
| Subjects: | H Social Sciences > HA Statistics |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis |
| Depositing User: | Mr. Tondo Indra Nyata |
| Date Deposited: | 04 Mar 2020 06:53 |
| Last Modified: | 04 Mar 2020 06:53 |
| URI: | http://repository.its.ac.id/id/eprint/75290 |
Actions (login required)
 |
View Item |