EVALUATION OF THE CLIMATE FORECAST SYSTEM (CFS) MODEL IN PREDICTING THE MADDEN JULIAN OSCILLATION (MJO) DURING BOREAL WINTER
Friska Chania(1), Muhammad Ridho Syahputra(1,2), Nurjanna Joko Trilaksono(1,2)

1. Undergraduate Program of Meteorology, Faculty of Earth Science and Technology, Institut Teknologi Bandung, Bandung, Indonesia&#8239-

2. Atmospheric Science Research Group, Faculty of Earth Science and Technology, Institut Teknologi Bandung, Bandung, Indonesia&#8239-&#8239-

Abstract

The Madden Julian Oscillation (MJO) has garnered the attention of scientific practitioners as it represents a source of predictability for subseasonal to seasonal (S2S) forecasts, bridging the gap between weather and seasonal predictions. One S2S model with potential for predicting sub-seasonal conditions in Indonesia is the Climate Forecast System (CFSv2). Currently, MJO index predictions are presented in the form of MJO phase diagrams- however, limitations in predicting the MJO index may constrain the analysis of existing models. Therefore, this study aims to evaluate the performance of the CFSv2 model in predicting the MJO index using the ^mjoindices^ tools developed by Hoffman.



The calculation of the MJO index is performed using the ^mjoindices^ tools from reanalysis data and operational CFSv2 model data for the boreal winter period (November, December, and January) from 2011 to 2021. Model performance evaluation is conducted deterministically with metrics such as RMSE and CC, and probabilistically with CRPS for evaluating the probabilistic distribution of weekly MJO index averages, and Brier score for evaluating weekly MJO active events. The criteria for an active MJO event are an MJO index greater than 1 for five consecutive days.



This study demonstrates that the ^mjoindices^ tools successfully calculated the OMI index and established an MJO index prediction system utilizing the ensemble prediction system of CFSv2 to forecast the phase and magnitude of the MJO index for the next 4 weeks. Although the performance evaluation of the CFSv2 model using RMSE and CC metrics did not show significant relationships between predictions and observations, probabilistic metrics such as CRPS and Brier score provide a more comprehensive picture. CRPS indicates that the model is more accurate in short-term predictions with the best value of 0.368 at lead 1, though accuracy decreases as lead time increases. The Brier score at lead 1 is 0.352, indicating lower accuracy for weekly predictions. However, the model^s predictability does not experience significant accuracy decline across subsequent leads.

Keywords: Madden Julian Oscillation (MJO), Model performance, CFSv2.

Topic: Atmospheric Sciences