Climate variability over the Greater Horn of Africa based on NCAR AGCM ensemble

Summary The variability and extreme wet anomalies in the Greater Horn of Africa (GHA) climate are investigated based on a multi-year National Center for Atmospheric Research (NCAR) AGCM ensemble data. While the GCM ensemble average reproduces realistic inter-annual variability of rainfall pattern over the GHA sub-region compared to observations, there is a distinct northward shift in the simulated regions of rainfall maxima throughout the season. However, in agreement with observations and many previous studies, the inter-annual variability derived from leading mode of EOF analysis is dominated by ENSO-related fluctuations. On the other hand, the spatial pattern corresponding to the second mode (EOF2) exhibits a unique dipole rainfall anomaly pattern (wet/dry conditions) over the northern/southern halves of our domain during all the three months of the short rains season. When the 3–10 year periodicity is filtered out from the 40-year EOF2 time series of the ensemble mean data, three distinct quasi-decadal regimes in the rainfall anomalies is exhibited for both monthly and seasonal mean data. It is also evident from our results that a combination of anomalous surface and mid-tropospheric flow from northwestern and eastern Atlantic Ocean and easterly flow from the Indian Ocean played a significant role in setting up the non-ENSO related 1961 floods. Coversely, during the ENSO-related 1997 floods, the mid-troposheric flow was characterized by anomalous westerly flow originating from the Congo rainforest that converged with the flow from Indian Ocean along the East Africa coast and over eastern/northeastern Kenya. The anomalous moisture flux convergence/divergence in both the ensemble and NCEP reanalysis is also consistent with the mid-trospheric flow anomalies that are associated with the two wet events.     

An investigation of the relationship between sub‐Saharan rainfall and global sea surface temperatures

Abstract

The relationship between sea surface temperature (SST) and rainfall index anomalies over sub‐Saharan Africa for the 15‐year period, 1970–84, has been examined. The objectively analysed monthly mean SST data were used for the global oceans between 40°S and 60°N. The rainfall data consist of annual mean rainfall indices for the Sahel and Soudan belts over north Africa.

An Empirical Orthogonal Function analysis of the SST data has been carried out for the Atlantic, Indian and global ocean regions. The results show that the most dominant eigenmode, EOF1, is characterized by warming over the central eastern Pacific, cooling over the eastern mid‐latitude Pacific and warming over the entire Atlantic and Indian ocean basins. The second EOF for the Atlantic Ocean SST analysis shows a dipole (north‐south see‐saw) pattern. The third EOF for the Atlantic SST analysis has the same sign over the entire Atlantic basin. Global SST EOF2 and EOF3 correspondió Atlantic SST EOF3 and EOF2, respectively.

The correlation between the sub‐Saharan annual rainfall index, which mainly represents the summer season rainfall from June to September, and SST EOFs shows that EOF1 has statistically significant monthly correlations for the Sahel and Soudan regions and that the warm El Niño‐like phases of SST EOF1 correspond to drought conditions. This result suggests that the large‐scale SST anomalies may be responsible for a significant component of the observed vacillation of sub‐Saharan rainfall. Some preliminary GLA GCM simulation results that support the above findings are also presented.     

Simulation of the sensitivity of Lake Victoria basin climate to lake surface temperatures

Summary The response of Lake Victoria basin climate to changes in the lake surface temperatures (LST) has been examined using NCAR-Regional climate model (RegCM2). In the control run uniform lake surface temperature of 24°C was prescribed and the model integrated for four months, starting at the beginning of September, 1988. In the anomaly experiments the LST was perturbed by ±1.5°C, and kept constant during the entire period of the integrations.Simulation results show significant relationship between basin-wide spatial distribution of rainfall and changes in LST. In general during the short rains at warmer/cooler LSTs, significant increase/decrease in the simulated rainfall occurs over the lake surface and surrounding areas. Rainfall exceeding the amount in the control run by more than 50%, particularly over the western, south/southwestern and central parts of the lake is simulated in the run in which the LST is 1.5°C warmer than the control. It is also evident from our results that different parts of the lake basin respond differently to LST changes which is in contrast to the common characterization of the lake basin as a single homogeneous climate regime in many previous studies.In general the results show that regions with largest response to LST anomalies during the short rains are collocated with the ITCZ. In October when the ITCZ is directly located over the lake, the largest response (maximum rainfall) is also located over the same region. As the season progresses and the ITCZ shifts out of the lake into northern Tanzania, the regions of rainfall maxima also shift with it. This appears to explain the unexpected reduction in over-lake rainfall in December in spite of the LST being warmer than control by 1.5°C. We believe this is a direct consequence of the enhanced convection to the south of the lake (over ITCZ) and the tendency of the system to conserve local moisture budget over the lake.     

Relationships Between QBO in the Lower Equatorial Stratospheric Zonal Winds and East African Seasonal Rainfall

Summary Teleconnections between the seasonal rainfall anomalies of March through May (“long-rains”) over eastern Africa (Uganda, Kenya and Tanzania) and the lower equatorial stratospheric (30-mb) zonal winds for the 32-year period 1964–1995 are examined using statistical methods. The analysis is based on the application of the simple correlation method and QBO/rainfall composite analysis. A statistical study of spatial correlation patterns is made in an effort to understand the climatic associations between the equatorial stratospheric zonal wind and regional rainfall at the interannual scale. The aim of this analysis is to establish whether this global signal can be employed as predictor variable in the long-range forecasts. The study is part of an ongoing investigation, which aims at designing a comprehensive and objective, multi-variate-forecast system of seasonal rainfall over eastern Africa. The correlation parameters include simultaneous (zero lag), and the non-zero lag correlations. The statistical significance of the correlation coefficient [r] is tested based on the Monte Carlo t-statistical method, and the standard correlation tables. Our results indicate significant positive simultaneous and non-zero lag correlations between rainfall over parts of East Africa and lower equatorial stratospheric zonal wind during the months of March–May and June–August. Significantly high correlations are concentrated over the western regions of eastern Africa with peak values of (+ 0.8) observed over these areas. These associations have been observed to be more prominent during lag than in the simultaneous correlations. Strong month to month lag coherence is observed after June prior to the onset of the March to May seasonal rainfall and persists for more than 4 months. Correlation indices for the eight homogeneous rainfall regions over eastern Africa which are derived from our Empirical Orthogonal Function/Cluster analysis shows a clear annual cycle with significant relationships between QBO and seasonal rainfall occurring during boreal summer (June–August). The season with the weakest relationship is December–February. It is however, noted that although the coherence between QBO-Index and rainfall during the long-rains is significantly high, there are some wet/dry years for which the relationship between the long rains and the lower equatorial zonal wind are not significant (for example in 1966, 1973 and 1983). These years have been associated with strong and prolonged ENSO events. Preliminary comparison of the QBO-Index and the newly found Indian Ocean dipole mode index (DMI) indicates that the two climate variables may be significantly related. Of the six high dipole mode events in the Indian Ocean that were observed in 1961, 1967, 1972, 1982, 1994 and 1997, all except 1967 coincided with the easterly phase of the QBO-Index and below normal rainfall over western highlands of eastern Africa. Contingency analyses indicate 60 percent likelihood for the occurrence of above normal rainfall during the westerly phase of the QBO and 63 percent likelihood of below normal rainfall during the east phase of the QBO. Our correlation analysis results indicate that about 36 percent of the variability of the long-rains season over eastern Africa are associated with the QBO-Index. Our results further show that the tendency of the lower equatorial stratosphe ric zonal wind prior to the season is a good indicator of the performance of the long rains of eastern Africa. A positive OND minus JJA QBO trend is a good indicator for the non-occurrence of drought over eastern Africa. Similarly, a negative trend is a good indicator for the non-occurrence of high rainfall over the region. The identified characteristics and domain of influence of the QBO signal in different regions of East Africa suggests that this global oscillator may offer useful input to objective multi-variate rainfall prediction models for eastern Africa. Received June 4, 1999 Revised November 25, 1999