Evaluation the Best Random Component in Modified Thomas-Fiering Model in Generating Rainfall Data for Akre station

Abstract

In this research, the effect of random component in the modified Thomas-Fiering model to generate daily rainfall data was studied, and Akre station considered a case study. A random component with special distributions: Normal random numbers, Wilson-Hilferty (W-H) transformation, truncated W-H, and Kirby modification to W-H transformation were used. The model applied to the daily rainfall data for Akre station for available years 2000–2006 and the model used to generate the rainfall data for the years 2006 and 2007. The results showed that the correlation coefficients between the observed and generated data were 0.82 for normal random numbers, 0.77 for W-H transformation, 0.89 for truncated –W –H, and 0.87 for KM to W-H transformation. The tests of Chi-square test, Kolmogorov–Smirnov test, root mean squared error (RMSE) test, and mean absolute error (MAE) test were used to compare between observed and generated data. All the results have passed the Chi-square test and Kolmogorov–Smirnov, where the calculated values were less than the tabulated value at 5% significance. For the test RMSE and MAE, the truncated W-H transform was the values of at least two. Therefore, W-H transform is the best for generating the rainfall data at Akre station

Prediksi curah hujan menggunakan metode Thomas-Fiering dengan modifikasi Kirby untuk rancangan sistem penyaliran tambang

“…Meskipun transformasi yang dimodifikasi oleh Kirby agak lebih rumit daripada transformasi asli atau truncated, namun dapat dengan mudah diorientasikan menggunakan software pengolah data. Persamaan modifikasi Kirby untuk transformasi Wilson-Hilferty (Aswad, Yousif dan Ibrahim, 2019) sebagai berikut: (Cahyadi dkk., 2020). Model ini biasanya mengkaji ulang data atau informasi terdahulu untuk menduga peluang kejadian tersebut pada keadaan sekarang atau yang akan datang dengan asumsi terdapat relevansi pada jalur waktu.…”