Abstract ID: 132
Improving Sub-seasonal to Seasonal Model Performance in The Tropics Using a Machine Learning Approach
Lead Author: Nurdeka Hidayanto
Indonesian Agency for Meteorological, Climatological and Geophysics, Indonesia
Keywords: S2S, Precipitation, Machine Learning, Tropics, GBM
Abstract: Subseasonal-to-seasonal (S2S) prediction play an important role as a bridge between conventional short-range weather forecasts and long-range seasonal forecasts. The S2S prediction has a time range which corresponds to beyond 2 weeks but less than a season, and practically is needed by specific sectors such as agriculture, water resources, and weather early warning for tropical cyclone. However, the studies investigating the performance of S2S prediction in Indonesia are still rare in Indonesia. . This study aims to investigate the performance of S2S model in forecasting rainfall in Indonesia and improve its performance by post-processing using Machine Learning (ML) techniques. Ten ML methods were utilized to improve the performance S2S model in forecasting rainfall in Indonesia. The S2S output data from the European Centre for Medium Range Weather Forecasts (ECMWF) from 1998 to 2017 were used along with 280 points observation data from meteorological stations and rain gauge in Indonesia. The results indicate that in general the ML model has the ability to improve the performance of the S2S model by more than 70% compared to the only S2S models as they were. The Gradient Boosting Machine (GBM) model got the best performance compared to the others.
Co-authors:
Agita Vivi Wijayanti (Indonesian Agency for Meteorological, Climatological and Geophysics)
Donaldi Sukma Permana (Indonesian Agency for Meteorological, Climatological and Geophysics)
Muhammad Rifki Taufik (Indonesian Agency for Meteorological, Climatological and Geophysics)
Kharisma Aprilina (Indonesian Agency for Meteorological, Climatological and Geophysics)
Ummu Ma’rufah (Indonesian Agency for Meteorological, Climatological and Geophysics)
Jose Rizal (Indonesian Agency for Meteorological, Climatological and Geophysics)
