Extreme precipitation in South Sulawesi triggered by equatorial waves and its representation in MetUM forecasts

The Indonesian island of Sulawesi lies at the heart of the Maritime Continent, a region prone to extreme rainfall. On seasonal timescales, rainfall frequency and intensity increases during the monsoon season (Nov-March). On subseasonal scales, rainfall is modulated by the Madden-Julian Oscillation (MJO) which increases moisture and moisture convergence in its active phase. Higher frequency modes include convectively coupled equatorial waves which influence rainfall variability on daily timescales. On 22nd January 2019, these large-scale meteorological drivers coincided resulting in the South Sulawesi region experiencing its largest flood on record. Specifically, the extreme rainfall event was linked to a convectively coupled Kelvin wave (CCKW) and a convectively coupled equatorial Rossby wave (CCERW) embedded within an active MJO envelope, as well as a cross equatorial cold surge (Latos et al, 2021). Interactions between these modes resulted in increased moisture transport and convergence that lead to the development of a mesoscale convective system (MCS) over Java and the Java Sea on 21st January 2019. This MCS traversed towards Sulawesi bringing extreme rainfall to the region in the evening and overnight on the 22nd reaching its peak mid-afternoon. Cases like this present a unique challenge for forecasters, to not only to accurately represent the individual equatorial modes but their interactions. In the present work we study the MCS in reanalysis and satellite data and discuss how the various equatorial modes contributed to its development. Then we examine its representation in different convection permitting Met Office Unified Model (MetUM) configurations. We find that the MetUM performed well in capturing the trajectories of the equatorial modes, however the representation of the MCS itself varies between ensemble members and model configurations. We further examine how well the RAL1T+ configuration represents the equatorial modes through comparing filtered fields of daily model data at fixed lead times to those in observations.