Modulation of Daily Rainfall in East Africa during the Long and Short Rains By the Madden Julian Oscillation and Convectively Coupled Equatorial Waves

Rainfall variability and especially related extreme events have a large socio-economic relevance for East African countries. Recent years have seen both extreme droughts, but also floods and an as of yet not well-understood decline in the long rains during March-April-May (MAM). For the mostly rain-fed agriculture, the onset, withdrawal, and breaks in the two major rainy seasons, the long and short (October-November-December, OND) rains, are of pivotal importance. It has been shown in previous studies that the Madden-Julian Oscillation (MJO) modifies intraseasonal variability including onset dates of the rainy seasons in East Africa. However, a comprehensive study using (a) an extensive network of rain gauges for various decades, (b) investigating the role of a sole and combined influence of the MJO and Convectively Coupled Equatorial Waves (CCEWs), and (c) studying the wave-related mean changes in the regional moisture budget and convective instability is currently lacking for the region.

The present study presents results from such a comprehensive approach. It investigated the modulation of daily rainfall by the MJO and four types of CCEWs for the six East African countries (South Sudan, Ethiopia, Kenya, Uganda, Rwanda, and Tanzania) between 1983 and 2013 for the long and short rains, respectively. It used daily rainfall observations from 91 weather stations and CHIRPS (Climate Hazards InfraRed Precipitation with Stations) satellite-based, gauge-calibrated rainfall estimates. Eight wave phases were determined regionally from wave-filtered Outgoing Longwave Radiation (OLR) using the method of Yasunaga and Mapes (2012). The dynamic and thermodynamic analysis included wave phase composites of anomalous rainfall, vector wind, moisture budget terms, Convective Available Potential Energy (CAPE), and Moist Static Energy (MSE). The interference of simultaneous occurrences of the MJO and Kelvin/Equatorial Rossby Waves (KWs/ERWs) was also investigated using a novel type of a wave interaction diagram. The composite differences have been tested for significance using a bootstrap method.

The MJO and some CCEWs exert a statistically significant influence on daily rainfall in East Africa for both rainy seasons with a stronger signal during the long rains. It is found that the MJO reveals the strongest rainfall modulation exceeding 6 mm/day between dry and wet MJO phases for 10% of the rainfall stations during MAM. The corresponding values for the KWs and ERWs are 4.5 and 4 mm/day, even though the ERWs show a much weaker variance in wave-filtered OLR. The impact of the MJO and CCEWs for the short rains is was considerably weaker and mixed Rossby-gravity and eastward inertio-gravity waves generally show the weakest influence, if any. For the central Kenyan highlands, the MJO and KWs substantially enhance the likelihood for strong daily rainfall in excess of 25 mm/day for both seasons and the long rains, respectively. The wave interaction diagrams show that the MJO can override the dry signals of KWs and ERWs and significantly enhance their wet phases.

Consistent with the above findings, ERA-Interim composites independently confirm much stronger modulations of winds, moisture transports and flux convergences, as well as convective instability during the long rains for the MJO and CCEWs. Maximum (minimum) rainfall anomalies occur during the transition from anomalous easterlies (westerlies) to anomalous westerlies (easterlies). The MJO expectedly shows a much stronger modulation of dynamic and thermodynamic fields than CCEWs in both seasons. For the MJO, the increase in CAPE and MSE is clearly evident at the coast, whereas the signal in these quantities is rather weak over the highlands, perhaps partially due to problems with ERA-Interim profiles over East African orography. According to the zonal wave propagation, the MJO and KW (ER) signals approach East Africa from the west (east).

The East African orography modifies MJO- and CCEW-related rainfall anomalies both in amplitude and phase. An example for regional differences is that positive rainfall anomalies are already in place at the coastal lowlands and the Ethiopian highlands in phase 4 of the MJO, which is one phase before strong wet anomalies occur all over East Africa during phases 5 and 6. Consistent with their smaller spatial scales, KWs and ERWs cause more localized rainfall anomalies. These regional details are much better represented in the station data when compared to CHIRPS data.

This comprehensive study clearly demonstrates the potential for sub-seasonal rainfall forecasting given an improved representation of the MJO and CCEWs in weather and seasonal forecasting models. It also confirms the necessity of using data from a sufficiently dense surface station network due to limitations in satellite rainfall estimates – even though CHIRPS is gauge-calibrated, it appears that some of the gauges used in the present study were not available to this product.