Ensemble Model Output Statistics Untuk Prediksi Suhu dan Kelembaban Jangka Pendek Terkalibrasi

Sari, Desi Erliana (2018) Ensemble Model Output Statistics Untuk Prediksi Suhu dan Kelembaban Jangka Pendek Terkalibrasi. Undergraduate thesis, Institut Tekologi Sepuluh Nopember.

[thumbnail of 0621144000079-Undergraduate_Theses.pdf]
Preview
Text
0621144000079-Undergraduate_Theses.pdf - Accepted Version

Download (3MB) | Preview

Abstract

Informasi tentang prediksi cuaca yang cepat dan tepat menjadi suatu hal yang penting karena mempengaruhi berbagai bidang kehidupan. Upaya BMKG dalam memberikan informasi mengenai kondisi cuaca di Indonesia yaitu dengan mengembangkan Numerical Weather Prediction (NWP). Dengan memanfaatkan data luaran NWP dilakukan penelitian untuk prediksi cuaca jangka pendek dengan melakukan ensemble menggunakan luaran Model Output Statistics (MOS). MOS dapat didekati menggunakan metode PCR, PLSR, stepwise regression, dan ridge regression. Namun hasil ensemble sering kali bersifat underdispersi sehingga dibutuhkan kalibrasi ensemble dengan menggunakan metode Ensemble Model Output Statistics (EMOS). Kebaikan model dievaluasi dengan menggunakan Root Mean Square Error (RMSE), Continuous Ranked Probability Score (CRPS), dan rank histogram. Hasil penelitian pada kedua stasiun pengamatan menunjukkan bahwa model MOS terbaik berdasarkan nilai RMSE yaitu model regresi stepwise dan PLSR. Untuk prediksi terkalibrasi berdasarkan nilai CRPS yang diperoleh, model EMOS lebih terkalibrasi dibandingkan dengan raw ensemble dalam memprediksi temperatur dan kelembaban========================================================================================================Information of fast and precise weather prediction is important due to its affects on many areas of life. BMKG had has an efforts in providing information about weather conditions in Indonesia by developing Numerical Weather Prediction (NWP). By utilizing NWP's output, this research conducted to predict short-term weather using ensemble Model Output Statistic (MOS). MOS can be approximated by PCR, PLSR, stepwise regression, and ridge regression methods. However, the ensemble result is often underdispersion, so it needs an ensemble calibration using the Ensemble Model Output Statistics (EMOS) method. This model goodness is evaluated by using Root Mean Square Error (RMSE), Continuous Probability Score (CRPS), and verification rank histogram. The results of the research on both stations showed that the best MOS model based on the RMSE value is a stepwise regression and PLSR. For calibrated predictions based on CRPS values obtained, EMOS models are more calibrated than the raw ensembles in predicting temperature and humidity

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Ensemble Model Output Statistics, Kalibrasi, Model Output Statistics, Numerical Weather Prediction
Subjects: Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression
Q Science > QA Mathematics > QA278.5 Principal components analysis. Factor analysis. Correspondence analysis (Statistics)
Divisions: Faculty of Mathematics, Computation, and Data Science > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User: Desi Erliana Sari
Date Deposited: 30 Jun 2021 06:32
Last Modified: 30 Jun 2021 06:32
URI: http://repository.its.ac.id/id/eprint/56683

Actions (login required)

View Item View Item