Four-dimensional variational (4DVAR) performance test with assimilation satellite and radar data (case study of a heavy rainfall Bengkulu March 4, 2019)

Abstract

Abstract Four-dimensional variational (4DVAR) is one of the assimilation techniques considering time integration to distribute observational data at time window intervals. In this study, we aim to evaluate the 4DVAR assimilation technique using satellite and radar data to simulate a heavy rainfall case in Bengkulu on March 4, 2019. The result shows that radar data assimilation (DA-RAD) can improve rainfall pattern over Bengkulu mainland areas, while the satellite data assimilation (DA-SAT) enhances rainfall over the ocean. In addition, for temporal analysis, the DA-RAD successfully correct the initial time of the event, producing the smallest error and the best correlation in statistical verification, also a small bias and higher accuracy for discrete verification. However, DA-SAT is more capable to improve rainfall accumulation with the lowest FAR value. In conclusion, compared to others, both satellite and radar can be used as assimilation data for 4DVAR methods as they have different roles in increasing the quality of model performance.