Convectively coupled Kelvin waves contribution to hazardous weather in Sumatra.

The island of Sumatra is characterized by extremely high average precipitation accumulation dominated by a strong diurnal cycle of convection, which develops over steep its topography. As a result, the island often suffers from precipitation driven natural hazards, which account for more than 50% of all natural disasters. Such hazardous events inflict great socio-economic loss and damage.

This study addresses this topic via a synergistic methodology employing analysis of meteorological data (precipitation, surface winds) and three independent datasets of floods, in order to consistently analyze spatio-temporal variability in flooding events and the environmental conditions leading to them. Two flood databases were derived from crowd-sourcing: Twitter and local papers. Additionally, the database from Indonesian agency BNPB was used. All three datasets were analyzed independently for the 2014-2018 period. While not all floods are identified in every data base, the results obtained from our analyses agree for all key elements of this study, providing cross calibration and increasing confidence in our findings.

On a subseasonal time scale, the amount of rainfall over the island is variable as well and strongly modulated by eastward propagating modes of organized convection: the Madden-Julian Oscillations (MJO) and convectively coupled Kelvin waves (CCKW), both of which affect the local diurnal cycle through multi-scale processes. This study investigates the relationship between those two modes of organized convection and flooding in Sumatra using several data forms of flood validation. It is shown that CCKWs constitute a critical dynamical predictor for flood onset.

Although our results agree with importance of MJO, indicated by previous studies, we find that only about 27% of floods in Sumatra were immediately preceded by favorable MJO conditions and all of them were in fact also associated with a CCKW embedded within an envelope of enhanced MJO convection. This was the case during the flood in Padang on 31 May, 2017, when the MJO was active over the Indian Ocean but its enhanced precipitation had not yet reached Sumatra. Instead, a strong CCKW, which initiated over eastern Africa, brought anomalous precipitation exceeding 10 mm/day to Padang and triggered a flood.

Comprehensive analysis shows that nearly half of robust CCKW events, which propagated between 80E and 110E, were associated with floods in Sumatra. From a different perspective, nearly all of floods in Sumatra were preceded by anomalous precipitation associated with a CCKW and about 60% of floods in Sumatra were immediately preceded by a strong CCKW event. This percentage is substantially higher than for favorable MJO conditions, indicating stronger interaction of local convection with CCKWs. Even though CCKW activity is modulated by the MJO itself and more CCKWs are found when the MJO is active in the Indian Ocean, during analyzed 2014-2018 period nearly 30% were associated with CCKW alone. Therefore, this study shows that CCKWs are important contributor to extreme weather in Sumatra and constitute a potential source of predictability of such hazardous events.