Evaluasi Kemampuan Luaran Model Subseasonal to Seasonal (S2S) dalam Menentukan Awal Musim Hujan Di Pulau Jawa

Abstract

Penentuan awal musim hujan sangat penting untuk pengambilan keputusan terutama pada sektor pertanian. Penelitian ini bertujuan untuk menguji kemampuan model ensambel subseasonal to seasonal (S2S) dalam menentukan awal musim hujan di Pulau Jawa periode tahun 2017-2021. Data luaran model S2S yang digunakan yaitu data lead time 44 hari dari Korea Meteorological Administration (KMA), National Centers for Environmental Prediction (NCEP), dan United Kingdom Meteorological Office (UKMO). Kriteria awal musim hujan ditentukan berdasarkan definisi agronomi. Metode quantile mapping digunakan untuk mengoreksi bias model S2S. Hasil menunjukkan model S2S terkoreksi mampu mereproduksi pola sebaran tanggal awal musim hujan data observasi, dengan nilai korelasi Pearson rata-rata 0,805. Kemampuan model S2S dalam menentukan awal musim hujan bervariasi untuk tiap titik kajian, hal ini dipengaruhi oleh pemilihan ensambel data, lead time, dan wilayah. Evaluasi kemampuan model S2S lead time 44 hari terkoreksi menunjukkan skill score rata-rata sebesar -0,015. Model S2S NCEP controlled forecast memiliki kemampuan yang lebih baik dari model UKMO dan KMA dengan nilai skill score tertinggi sebesar 0,207. Studi ini menemukan bahwa koreksi bias yang diterapkan masih belum dapat meningkatkan kemampuan model S2S dalam memprediksi awal musim hujan pada Pulau Jawa.

Determining onset of rainy season is very important for decision making especially the agricultural sector. This study aims to test the ability of the sub seasonal to seasonal (S2S) ensemble model in determining onset of rainy season in Java for the 2017-2021 period. The output data of the S2S model used is a 44-day lead time from the Korea Meteorological Administration (KMA), the National Centers for Environmental Prediction (NCEP), and the United Kingdom Meteorological Office (UKMO). The criteria for rainy season onset are determined based on the agronomic definition. The quantile mapping method is used to correct the bias of the S2S model. The results show that the corrected S2S model can reproduce the distribution pattern of the rainy season onset date of observation data, with an average Pearson correlation value of 0,805. The ability of the S2S model to determine onset of rainy season varies for each study point, this is influenced by the choice of data ensemble, lead time, and region. Evaluation of the ability of the S2S model with a corrected 44-day lead time shows an average skill score of -0,015. The S2S NCEP controlled forecast model has better abilities than the UKMO and KMA models, with the highest skill score value of 0,207. This study found that the bias correction applied still could not improve the ability of the S2S model to predict the onset of rainy season on Java Island.