Modelling the atmospheric response over southern Africa to SST forcing in the southeast tropical Atlantic and southwest subtropical Indian Oceans

Transcription

1 INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. 29: (2009) Published online 9 April 2009 in Wiley InterScience ( Modelling the atmospheric response over southern Africa to SST forcing in the southeast tropical Atlantic and southwest subtropical Indian Oceans K. Hansingo and C. J. C. Reason* Department of Oceanography, University of Cape Town, Private Bag X3, Rondebosch 7701, South Africa ABSTRACT: The sensitivity of the atmospheric circulation and rainfall over southern Africa to various regional sea surface temperature (SST) patterns observed during Benguela Niño events is investigated using an atmospheric general circulation model (AGCM). The model used is the United Kingdom Meteorological Office (UKMO) HadAM3 and experiments using idealizations of observed regional SST anomalies during various Benguela Niño events are performed. It is found that SST forcing in the tropical southeast Atlantic induces a regional baroclinic response and that a Benguela Niño is capable of forcing anomalous rainfall along the Angolan and northern Namibian coastal regions on its own, via changes in low level moisture convergence, uplift and evaporation over the SST forcing. An experiment with the tropical southeast Atlantic SST anomaly shifted slightly further north produced a larger circulation and rainfall response in the model. Additional experiments with various SST anomalies in the southwest Indian Ocean combined with those in the southeast Atlantic were performed. These experiments are motivated by the fact that South Indian Ocean SST anomalies of varying signs often occur at the same time as Benguela Niño events. The results of these experiments suggest that, depending on its sign, magnitude and location, SST forcing from the southwest Indian Ocean may augment or oppose the southern African rainfall anomalies occurring during a Benguela Niño event to varying degrees. Copyright 2009 Royal Meteorological Society KEY WORDS atmospheric modelling; South Atlantic; South Indian Ocean; SST forcing Received 5 September 2007; Revised 10 March 2009; Accepted 11 March Introduction Although the occurrence of anomalously warm sea surface temperature (SST) events in the tropical southeast Atlantic Ocean in late austral summer (so-called Benguela Niños) together with anomalously wet conditions in the Namib coastal desert has been known for about two decades (Shannon et al., 1986), the mechanisms behind the enhanced rainfall are not well understood. During some Benguela Niños, the rainfall anomalies are mainly confined to the coast whereas in others they extend well inland (Rouault et al., 2003). A strong warm event in the southeast Atlantic with a clear link through the thermocline to the equatorial Atlantic Ocean and with strong impacts on regional fisheries and significant increase in rainfall over coastal Angola/Namibia is generally classified as a Benguela Niño (Shannon et al., 1986; Reason et al., 2006a). Recent examples of Benguela Niños occurred in 1984, 1986 and 1995 and 2001 (Rouault et al., 2003; Florenchie et al., 2003, 2004). The composite SST anomaly plot for these events is shown in Figure 1(a). The centre and magnitude of the maximum SST anomaly varies among these events, * Correspondence to: C. J. C. Reason, Department of Oceanography, University of Cape Town, Private Bag X3, Rondebosch 7701, South Africa. cjr@egs.uct.ac.za from about 15 S to about 17 S and from about 2 C to about 4 C, respectively. Some of the Benguela Niño events occur at about the same time as an El Niño Southern Oscillation (ENSO) event or when there are also sizeable SST anomalies in the South Indian Ocean. For example, the 1984 and 1995 Benguela Niños were accompanied by cool SST anomalies in the southwest Indian Ocean while warm anomalies occurred there during the 1986 and 2001 events. Given the fact that the southern African rainfall anomalies seem to vary substantially in magnitude and spatial extent during Benguela Niños and that sometimes there are also SST anomalies at the same time in the South Indian Ocean, the question arises as to whether it is the SST forcing in the South Indian Ocean or that in the tropical southeast Atlantic SST anomalies that is mainly driving the observed rainfall anomalies. To try and address this question, this study investigates the sensitivity of the regional circulation and rainfall to various idealized SST anomalies in the southeast Atlantic and South Indian Oceans using the UKMO HadAM3 atmospheric general circulation model (AGCM). These SST anomalies are smoothed representations of the observed patterns and are about twice as large as the actual observed magnitudes. As in previous idealized AGCM experiments (e.g. Reason Copyright 2009 Royal Meteorological Society

2 1002 K. HANSINGO AND C. J. C. REASON Figure 1. NCEP National Center for Atmospheric Research (NCAR) reanalysis February April (FMA) southern African (a) optimally interpolated SST composite anomaly for FMA 1984/6/95 and 2001 derived from the 1982 to 2005 climatology. Positive anomalies greater then 0.5 C are shaded and contour interval is 0.5 C. (b) Climatological moisture flux vectors and moisture convergence (shaded) at 850 hpa and (c) climatological 500 hpa omega with negative values, indicating rising air, less than 0.01 Pa/s shaded. Contour interval is 0.01 Pa/s and vector size is shown. Base period is 1970 to (d) Southern Africa FMA CMAP rainfall climatology. Contour interval is 1 mm/day. and Jagadheesha, 2005a), the SST forcing is increased so as to reduce the noise and help isolate the response in the model to the forcing. The HadAM3 model was chosen due to its previous wide application to southern African climate variability (e.g. Tadross et al., 2005; Reason and Jagadheesha, 2005a, 2005b; Washington and Preston, 2006). An alternative approach is to use a regional climate model. However, given that the study wished to include both Atlantic and Indian Ocean SST anomalies and thus would require a large regional domain, the computational facilities available to the authors precluded the use of a regional model at a realistic horizontal resolution. In addition, the use of a global model avoids any issues with boundary conditions that can cause problems in a regional climate model application. Climatologically, the western Indian Ocean is the major source of moisture for late austral summer rainfall (February-March-April FMA) over southern Africa as shown in Figure 1(b). The Indian Ocean sourced low level warm moist air converges with drier cooler air from the South Atlantic over southwestern Africa south of 15 S. The latter is not immediately obvious from Figure 1(b), a mean plot. However, north of 15 S, periods of significant low level westerly flow from the tropical southeast Atlantic Ocean towards Angola are often associated with wet spells over subtropical southern Africa (Cook et al., 2004; Reason, 2007). A moisture convergence region extends inland from a surface heat low (Angola low) towards the meridional arm of the Inter-tropical Convergence Zone (ITCZ) that is located over the eastern Congo basin. Moisture convergence over eastern Angola, western Zambia, southern Congo and western Tanzania together with rising motion observed in the 500-hPa omega field in Figure 1(c) is favourable for rainfall over these regions near the convergence zone and the meridional arm of the ITCZ. The FMA season (Figure 1(d)) contributes more than 40% to the annual rainfall total over Africa south of 15 S (Washington and Preston, 2006) and substantial variability exists on interannual time scales (Richard et al., 2000). Since the circulation over southern Africa is more tropical in nature during this season (D Abreton and Lindesay, 1993), variability in features such as the ITCZ and the low level moisture confluence region extending towards the Angola low (Reason et al., 2006b), the Angola low itself, and the tropical extratropical cloud bands that stretch from this low to a midlatitude disturbance passing south of Africa may all contribute to rainfall over southern Africa. 2. Model description and experimental design The HadAM3 model has been used in various African studies, e.g. in investigating intraseasonal (Tennant, 2003) or interannual (Reason and Jagadheesha, 2005a, 2005b) climate variability and in sensitivity studies (Washington and Preston, 2006) as well as for operational seasonal

3 ATMOSPHERIC MODEL RESPONSE TO SOUTH ATLANTIC AND SOUTH INDIAN SST 1003 forecasting. A detailed evaluation of this model, its biases and the main parameterizations of the subgrid scale physics are provided in Pope et al. (2000). Previous studies with this model (e.g. Reason et al., 2003; Reason and Jagadheesha, 2005a, 2005b; Washington and Preston, 2006) indicate that HadAM3 correctly simulates a unimodal annual cycle with maximum rainfall in the austral summer (October March) and minimum rainfall in the austral winter (April September) over subtropical southern Africa. These authors indicate that the model does overestimate summer rainfall and this is most pronounced in the early summer season (October December) with the positive bias continuing into the late summer months (January March). Reason and Jagadheesha (2005a) also observed that the model is less successful with magnitudes of winter rainfall over the southwestern Cape region of South Africa for certain years between 1985 and 2000, although it was able to represent the sign of the anomalies each year. Despite the model s shortcomings in simulating rainfall magnitudes correctly, it has some skill in capturing the observed interannual tendency in rainfall over the winter rainfall dominated southwestern region of South Africa (Reason and Jagadheesha, 2005a). The model also has some skill in capturing the observed interannual variability in rainfall over the summer rainfall region over southern Africa (Washington and Preston, 2006). Given these strengths of the model, it seems appropriate to apply it to investigate the sensitivity of the regional atmosphere to idealized SST patterns. In many sensitivity experiments with AGCMs, a control experiment is required with which to compare the model response to perturbations to SST forcing. In this study, the control experiment is one in which the model is forced with a 24-year ( ) global climatology of optimally interpolated SST (Reynolds and Smith, 1994) and integrated for 10 months starting from January. This experiment consists of five ensemble members, each initialized with different initial atmospheric conditions arbitrarily picked. Idealized experiments involved adding an idealized SST anomaly pattern(s) to the 24-year climatology over a particular region(s). An ensemble of five integrations each starting from different initial atmospheric conditions has been performed with various idealized SST anomalies imposed over the southeast Atlantic region and the southwest Indian Oceans (Figures 2 and 3). Each ensemble member begins from a different 1st January state and these are then integrated with a constant SST anomaly pattern added to the climatology for the January May period. The month of January is regarded as the spin-up period. To try and isolate the role that Benguela Niños and SST anomalies in the southwest Indian Ocean may play in influencing the regional circulation, the following idealized experiments are performed with the HadAM3. The first experiment, experiment 1, uses an idealization of the Benguela Niño SST anomaly pattern occurring during FMA 1995 (Figure 2). Idealized SST anomaly patterns for additional sensitivity experiments are shown in Figure 3. Experiment 2 is the same as experiment 1, except that cool SST anomalies are added south of Madagascar (Figure 3(a)). Note that the 1995 Benguela Niño was associated with cooling in the southwest Indian Ocean south of Madagascar. In Experiment 3, the SST anomaly in the southeast Atlantic is shifted equatorward (Figure 3(b)), similar to the location in the composite plot (Figure 1(c)). The purpose of this experiment is to investigate the sensitivity of the regional atmosphere to a Benguela Niño with SST anomalies located deeper in the tropics. Experiments 4 and 5 are similar to the previous one except that a positive SST anomaly is added in the southwest Indian Ocean. Note that the 1986 and 2001 Benguela Niños were characterized by warm anomalies in the southwest Indian Ocean. In experiment 5, the southwest Indian Ocean anomaly is closer to southern Africa (Figure 3(d)) than in experiment 4 (Figure 3(c)). Previous studies (e.g. Reason and Mulenga, 1999; Behera and Yamagata, 2001; Reason, 2001, 2002) have shown that warming in the southwest Indian Ocean may result in anomalous rainfall over southern Africa. The results below present Figure 2. Idealized SST anomaly representing a Benguela Niño forcing occurring during 1995 in the southeast Atlantic Ocean. Contour interval is 0.5 C.

4 1004 K. HANSINGO AND C. J. C. REASON Figure 3. Sensitivity of Benguela Niño forcing: (a) same as Figure 2 but with a cool southwest Indian Ocean SST anomaly added, (b) same as Figure 2 but shifted northward, (c) same as (b) but with a warm southwest Indian Ocean SST anomaly added, (d) same as (c) but with the Indian Ocean SST anomaly closer to the subcontinent. FMA mean differences between each experiment and the control experiment. A student t test is used to determine the statistical significance of the model response in each case, considering each member of the ensemble as independent. 3. Idealization of a Benguela Niño: experiment 1 Figure 4(a) (c) shows rainfall, latent heat flux and moisture flux anomalies generated by the model when forced with the SST anomaly pattern in Figure 2. The model s response to the SST forcing in experiment 1 includes enhanced rainfall (Figure 4(a)) and increased evaporation (Figure 4(b)) near the SST anomaly as well as a low level cyclonic anomaly (Figure 4(c)) that extends westwards together with a high pressure anomaly aloft (not shown). Note that the bold vectors in this figure denote differences that are statistically significant at the 95% level. This response is somewhat similar to the baroclinic response obtained by Haarsma et al. (2003) and Robertson et al. (2003) to a regional SST anomaly in the South Atlantic Ocean during summer. The model produces positive rainfall anomalies over western Angola and northern Namibia that stretch southeastward over South Africa, with other areas of increased rainfall over parts of southeastern Africa/southwest Indian Ocean (Figure 4(a)). Positive anomalies are also evident over the Gulf of Guinea and neighbouring coastal West Africa consistent with the observations in Reason and Rouault (2006) who found a strong positive correlation between rainfall there and SST anomalies in the tropical southeast Atlantic. Rainfall over the region extending southeastward from Angola is usually associated with tropical extratropical cloud bands or tropical temperate troughs (TTTs), which extend from the Angola low to a westerly disturbance passing south of Africa. Figure 4(c) shows a cyclonic anomaly off Angola with a NW SE orientated area of moisture convergence stretching to the southwest coast of South Africa that is located near and just to the west of the rainfall band in Figure 4(a). This pattern is suggestive of increased TTT activity. A broad band of enhanced latent heat flux is evident across Namibia and over South Africa (Figure 4(b)) and is associated with the positive rainfall anomaly band. This band extends from the large positive latent heat flux anomalies generated over the SST forcing. Areas of increased latent heat flux, such as those in Figure 4(b), may serve as sources of moisture and energy for storms as a result of enhanced surface evaporation over the warm SST anomaly. Although not the subject of this study, it is important to mention that the relationship between SST and latent heat flux is complex and cannot be explained by thermodynamic considerations alone (Zhang and McPhaden, 1995). In investigating relationships between SST and latent heat flux in the equatorial Pacific, Zhang and McPhaden (1995) found that at lower SST (near and below 27.5 C) the latent heat flux increases with SST, whereas at higher SST (above 27.5 C) the latent heat flux decreases with SST. Other parameters such as wind speed and humidity need to be considered as well in order to understand how the latent heat flux may change. The latent heat results in Figure 4(b) are in agreement with this SST latent heat flux relationship because the ocean region offshore of Angola/northern Namibia has relatively lower SST than that in the equatorial western

5 ATMOSPHERIC MODEL RESPONSE TO SOUTH ATLANTIC AND SOUTH INDIAN SST 1005 Figure 4. Experiment 1 anomalies for (a) rainfall and (b) surface latent heat flux anomalies. Contour levels are as shown. Panel (c) shows moisture flux (vectors) and moisture convergence (contours) anomalies at 850 hpa. Negative (positive) values indicate increased (reduced) moisture convergence and contour interval is 0.5 g/kg m/s. Vector size is shown. The shading and dark vectors in this and succeeding figures denote areas with the difference significant at 95% level as estimated by a t test. and central Pacific. Another thing to note is that since this is an atmosphere only experiment, feedbacks between the ocean and the atmosphere, which take place in the real world through surface flux interactions, are not taken into account. Therefore, the results need to be interpreted in that light.

6 1006 K. HANSINGO AND C. J. C. REASON As mentioned above, increases in surface latent heat flux increase the supply of moisture to the lower atmosphere. Although the availability of moisture is necessary for precipitation to occur, other factors such as moisture convergence, uplift and atmospheric instability also need to be considered. The westerly moisture flux anomalies over the South Atlantic north of 10 S and the cyclonic anomaly circulation off the coast of Angola lead to relative low level moisture convergence over central Angola and much of the Congo (Figure 4(c)), with weak positive rainfall anomalies occurring over the former. The large cyclonic moisture flux anomaly (Figure 4(c)) centred over the Mozambique Channel leads to relative convergence over Tanzania and other areas of tropical eastern Africa and the tropical southwest Indian Ocean on its northern arm with associated positive rainfall anomalies there. This Mozambique Channel feature is produced in all the experiments and global plots of the anomalous winds (not shown) for each experiment reveal that it seems to be a result of the SST forcing modulating the standing waves in Southern Hemisphere. These standing waves are known to be important to southern African climate (Mason and Jury, 1997). Similar modulations to these waves were obtained by Williams et al. (2008) in AGCM experiments with SST forcing in the central and eastern South Atlantic. National Center for Environmental Prediction (NCEP) re-analyses for austral summer 1995 indicate strongly negative anomalies in the 500-hPa omega field over the tropical southeast Atlantic and Angola/Namibia. Thus, rising air was also observed near the warm SST during the 1995 Benguela Niño consistent with the increased precipitation there. With a baroclinic type of response in the model and rising air at 500 hpa (not shown) over the tropical southeast Atlantic, this suggests enhanced convection near the region of forcing. 4. Sensitivity to location of the Benguela SST anomaly and to South Indian Ocean SST forcing Given the above response of the atmosphere to the idealized Benguela Niño forcing, we now examine the sensitivity of the model to firstly, an idealization of the 1995 regional SST anomalies (experiment 2) (i.e. as in experiment 1 but with cool SST anomalies also added in the southwest Indian Ocean), and then to idealizations of the SST patterns observed during other Benguela Niño events (experiments 3 5). The results are presented as differences between each of these experiments and the control, for rainfall and moisture flux Idealization of the 1995 Benguela Niño: experiment 2 As mentioned earlier, during the 1995 Benguela Niño, a cool SST anomaly occurred in the southwest Indian Ocean south of Madagascar together with strong warming in the tropical southeast Atlantic. During this event, extensive floods were reported over western Angola and northern Namibia, but further east, many parts of southern Africa received below average rainfall. Experiment 2 is an idealization of this event in which the AGCM is forced with SST anomalies as in Figure 3(a). When forced with these anomalies, the model produces rainfall anomalies similar to those in the previous experiment. However, a small reduction in the positive rainfall anomalies seen in experiment 1 (Figure 4(a)) is observed near the forcing in the southeast Atlantic and the southern Mozambique Channel in experiment 2 (Figure 5(a)). Despite enhanced low level moisture convergence anomalies (Figure 5(b)) near the SST forcing in the southeast Atlantic Ocean in experiment 2, positive rainfall anomalies of only about 2 mm/day are produced there (Figure 5(a)). On the other hand, rainfall anomalies of greater than 0.5 mm/day over western Angola/northern Namibia extend eastward over Zambia/Mozambique and Tanzania in experiment 2 as a result of enhanced low level moisture convergence (Figure 5(b)) over this area. Compared to experiment 1 (Figure 4(c)), the cyclonic anomaly centred over Madagascar appears to have weakened and shifted slightly northwestward in experiment 2, thereby causing enhanced anomalies in low level moisture convergence (Figure 5(b)) over the region extending northeast from southwestern Zambia and the positive rainfall anomalies noted above. Therefore, these results suggest that SST in the subtropical southwest Indian Ocean may lead to its own response over the region that can augment or oppose the rainfall impacts of a Benguela Niño over southern Africa. Note that previous AGCM experiments with SST anomalies imposed in this part of the South Indian Ocean (e.g. Reason and Mulenga, 1999; Reason, 2001, 2002) also show a significant circulation and rainfall response over southern Africa. Note that during the 1995 Benguela Niño, below average rainfall was observed over Zimbabwe/Mozambique but wetter than average conditions occurred over eastern South Africa and Tanzania (Rouault et al., 2003). Figure 5(a) also shows above average rainfall over eastern South Africa and Tanzania, but does not show dry conditions over Zimbabwe and Mozambique. This result suggests that some other forcing besides that idealized in Figure 3(a) may have contributed to the observed rainfall anomalies. Note that the 1995 Benguela event also coincided with a protracted El Niño event in the Pacific but that the corresponding Pacific or tropical Indian SST anomalies are not included in the model experiment. Thus, these model results suggest that independent of ENSO, a Benguela Niño is capable of producing anomalous rainfall over southern Africa despite cooling in the southwest Indian Ocean which is normally associated with below average rainfall over the subcontinent. However, the results also suggest that the 1995 observed rainfall anomalies were not solely due to regional Atlantic or South Indian Ocean SST forcing but were also contributed to by other SST forcing (e.g. associated with the ENSO event) or by other factors not included here.

7 ATMOSPHERIC MODEL RESPONSE TO SOUTH ATLANTIC AND SOUTH INDIAN SST 1007 Figure 5. Experiment 2 anomalies for (a) rainfall and contour levels are as shown. (b) 850 hpa anomalies for moisture flux (vectors) and convergence (contours). Negative values indicate enhanced convergence and positive values indicate relative divergence. Contour interval is 0.5 g/kg m/s. The vector size is shown but the zero contour is not shown Benguela Niño shifted northward: experiment 3 In experiment 3 (Figure 3(b)), the forcing in the southeast Atlantic is shifted equatorward. This experiment is conducted on the premise that tropical forcing may produce a stronger atmospheric response locally, similar to over Peru during an El Niño in the Pacific. Because of higher SST in the tropics, a warm SST anomaly located further into the tropics raises the actual surface temperature to a correspondingly higher level than the one located more polewards, thereby increasing the chances of deep atmospheric convection developing in its vicinity. Therefore, it is expected that experiment 3 will generate more rainfall along the Angolan coast than experiments 1 and 2. Figure 6 shows the model s response to this northward shifted SST forcing. Consistent with the more tropical SST anomaly, positive rainfall anomalies of more than 5 mm/day (or about twice the magnitude of that in experiment 1, Figure 4(a)) are generated near the SST forcing and along the Angolan coast (Figure 6(a)). Rainfall anomalies of greater than mm/day extend from this region southeast and northeast to South Africa/southwest Indian Ocean and tropical eastern Africa, respectively. In general, more rainfall is generated over the subcontinent in this experiment than in the previous two. These results suggest that a Benguela Niño with maximum SST anomaly slightly farther north may produce more anomalous rainfall over western Angola/northern Namibia and elsewhere than if the SST forcing is situated further poleward. The cyclonic anomalies centred off Angola and over Madagascar lead to low level moisture convergence anomalies over northern Angola, eastern tropical Africa and southern Zambia (Figure 6(b)). Although these circulation anomalies are also produced

8 1008 K. HANSINGO AND C. J. C. REASON Figure 6. Same as Figure 5 but for experiment 3. in the above experiments, they appear to be stronger in experiment 3. The cyclonic anomaly off Angola tends to reduce the low level moisture flux off the adjacent land onto the South Atlantic Ocean (cf Figure 1(b)), whereas that over Madagascar tends to enhance moisture flux from the subtropical southwest Indian Ocean, thereby leading to more rainfall over South Africa. Stronger moisture convergence anomalies (Figure 6(b)) lead to positive rainfall anomalies stretching southward over Namibia and western South Africa (Figure 6(a)) Warm Indian Ocean SST anomalies south of Madagascar: experiment 4 In experiment 4, the model is forced with the same SST anomalies as in experiment 3 but with a warm anomaly added in the southwest Indian Ocean (Figure 3(c)), similar to that observed during the 2001 Benguela Niño. This experiment is conducted on the premise that warm SST anomalies in the subtropical southwest Indian Ocean may generate increased rainfall over southern Africa and hence could augment those due to SST anomalies in the southeast Atlantic. Both observations and GCM results in Reason and Mulenga (1999), Reason (2001) and Hansingo and Reason (2006) have shown that warming in the subtropical southwest Indian Ocean during austral summer leads to enhanced rainfall over large areas of southeastern Africa. Furthermore, Behera and Yamagata (2001) and Reason (2001, 2002) have shown that anomalous rainfall is produced over southern Africa during the positive phase of a South Indian Ocean subtropical dipole event in summer. This phase is characterized by warm

9 ATMOSPHERIC MODEL RESPONSE TO SOUTH ATLANTIC AND SOUTH INDIAN SST 1009 SST anomalies south of Madagascar and cool anomalies off the western coast of Australia. Therefore, given this previous research and the results in experiments 2 4, with the SST forcing in the tropical southeast Atlantic shifted northward plus the warm SST forcing imposed south of Madagascar (Figure 3(c)), one expects this experiment to generate more rainfall over southern Africa than experiments 1 3. However, Figure 7(a) shows that compared to experiment 3, there is only about the same amount of rainfall produced over southern Africa as well as over the SST forcing in the tropical southeast Atlantic, although the latter is much greater than in experiments 1 and 2. Somewhat larger positive rainfall anomalies are generated over northern Madagascar and the Mozambique Channel in this experiment compared to experiment 3. Experiment 4 effectively contains the positive (warm) pole of a South Indian Ocean subtropical SST dipole event. Reason and Mulenga (1999), Behera and Yamagata (2001) and Reason (2001, 2002) attributed anomalous rainfall over southeastern Africa during warm events in the southwest subtropical Indian Ocean to enhanced lower tropospheric easterlies transporting surplus moisture to this region, and hence enhanced moisture convergence and convection there. The warm SST anomaly in the southwest Indian Ocean acts as the source of surplus moisture. This mechanism is observed in Figure 7(b) over eastern Zambia and Mozambique. Further north over Tanzania and southern Kenya, there are statistically significant positive rainfall anomalies (Figure 7(a)) which are related to relative enhanced low level moisture convergence anomalies (Figure 7(b)). The latter results because the westerly Figure 7. Same as Figure 5 but for experiment 4.

10 1010 K. HANSINGO AND C. J. C. REASON anomalies over and north of Madagascar (Figure 7(b)) oppose the mean easterly flux in this region (Figure 1(b)) Warm South Indian Ocean SST anomalies shifted closer to the subcontinent: experiment 5 Experiment 5 (Figure 3(d)) is similar to experiment 4 except that the SST anomaly forcing in the subtropical southwest Indian Ocean is shifted closer to the subcontinent. Previous experiments with a numerical model suggest that the positive rainfall anomalies are further enhanced over southern Africa when the warm SST anomalies in the subtropical southwest Indian Ocean are close to the subcontinent (Reason, 2002). Therefore, one expects this experiment to generate more rainfall over the subcontinent than in experiment 4. Increased rainfall is now produced over a much larger area of southern Africa than in previous experiments (Figure 8(a)). The size of the rainfall anomalies over southern Africa and the southwest Indian Ocean is now also generally greater than in experiment 4, although the cyclonic moisture flux anomaly (Figure 8(b)) over Angola and the tropical southeast Atlantic is broadly similar. However, increased low level moisture convergence is observed extending southwest from the SST forcing in the tropical southeast Atlantic to the west of South Africa (Figure 8(b)). Strong low level moisture convergence over western South Africa and Namibia leads to a large increase in rainfall over these regions. Much of the increased rainfall over Botswana, Zambia, South Africa and Mozambique is likely due to the more tropically located SST anomaly in the southwest Indian Ocean, which is also much closer to the landmass than in experiment 4. Consistent with Figure 8. Same as Figure 5 but for experiment 5.

11 ATMOSPHERIC MODEL RESPONSE TO SOUTH ATLANTIC AND SOUTH INDIAN SST 1011 previous AGCM experiments (Reason, 2002), this leads to a stronger rainfall response over large area of southern Africa. Taken together, the results of experiments 2, 4, 5 suggest that, depending on their sign and location, SST forcing in the southwest Indian Ocean may augment or oppose the rainfall impacts over southern Africa during a Benguela Niño event. For both the southwest Indian and southeast Atlantic SST anomalies, a more tropically located pattern tends to produce a larger response in the model. 5. Summary and discussion Southern African rainfall exhibits strong variability on interannual and interdecadal time scales which is usually linked to SST variations in the neighbouring and remote oceans. Despite substantial evidence of variability in the tropical southeast Atlantic, relatively little work has been done to understand the relationships between variability in this ocean region and southern African rainfall. In this study, experiments with the UKMO HadAM3 GCM forced with various idealized SST anomalies are used to explore the sensitivity of the regional atmosphere to SST forcing in the tropical southeast Atlantic/southwest Indian Oceans and to provide further evidence that these anomalies may influence rainfall over southern Africa. The model s response to the SST anomaly forcing in the tropical southeast Atlantic Ocean (Figure 2) consists of positive rainfall anomalies over western Angola and northern Namibia that stretch southeastward over Namibia and South Africa (Figure 4(a)). Enhanced rainfall is also observed over parts of southeastern Africa and the southwest Indian Ocean (Figure 4(a)). This model response is consistent with the observations of Rouault et al. (2003). Positive anomalies are also evident over the Gulf of Guinea and neighbouring coastal West Africa consistent with the observations in Reason and Rouault (2006) who found a strong correlation between rainfall there and SST anomalies in the tropical southeast Atlantic. The positive rainfall anomalies over western Angola are associated with enhanced surface latent flux over the SST forcing near the SST forcing in the tropical southeast Atlantic Ocean. The relatively weak anomalies in low level moisture convergence (Figure 4(c)) suggest that the enhanced surface latent heat flux and local uplift contribute more to the positive rainfall anomalies over western Angola than do changes in convergence. The model results suggest that a Benguela Niño is capable of forcing anomalously wet conditions over western Angola/Namibia on its own. However, it is less clear as to how influential it is on the rainfall further inland. More rainfall is generated over western Angola/northern Namibia and farther east over the subcontinent when the southeast Atlantic SST forcing is shifted slightly northward (experiment 3). The resulting changes in the atmospheric circulation lead to changes in low level moisture convergence over interior southern Africa that appear to contribute to positive rainfall anomalies there. Therefore, this result suggests that a Benguela Niño is capable of influencing anomalous rainfall both over western Angola and much further inland when the maximum SST anomalies are located further north in the southeast Atlantic. Consistent with observational studies (Hirst and Hastenrath, 1983; Rouault et al., 2003), these model results imply that the western Angola/northern Namibia region is susceptible to flood events during Benguela Niños. Since the onset of these events is typically around December/January, monitoring the tropical southeast Atlantic Ocean could provide an early warning system that could possibly be beneficial to the Angolan and Namibian economies. Positive rainfall anomalies are also generated over southern Africa in experiments 4 and 5 in which warm SST anomalies in the subtropical southwest Indian are imposed in addition to that in the tropical southeast Atlantic. As indicated by Behera and Yamagata (2001) and Reason (2001, 2002), positive SST anomalies in the southwest Indian Ocean act as a source of surplus moisture. Enhanced lower tropospheric easterlies transport surplus moisture to eastern South Africa/Zimbabwe/southern Mozambique and enhanced moisture convergence and convection over this region lead to anomalous rainfall there. Conversely, in experiment 2 in which cold SST anomalies are imposed in the southwest Indian Ocean, the resulting model rainfall anomalies over western Angola and elsewhere in southern Africa are generally smaller than in the other experiments. Taken together, the results of experiments 2 5 suggest that, depending on its sign and location, SST forcing in the southwest Indian Ocean may augment or oppose the rainfall anomalies that occur over southern Africa during a Benguela Niño. These results also suggest that the 1995 observed rainfall anomalies over southern Africa were not solely due to regional Atlantic or South Indian Ocean SST forcing, but were also contributed to by other SST forcing (e.g. associated with the ENSO event) or by other factors not included here. The question of what proportion of these rainfall anomalies is due only to SE Atlantic SST forcing and what is due only to SW Indian Ocean forcing remains to be quantitatively addressed. This study has only considered seasonal anomalies. The issue of whether regional SST anomalies lead to changes in the frequency or intensity of dry and wet spells during the summer rainy season or in the onset and cessation dates of this season have not been considered. These factors are very important for the agriculture-driven economy of most of the countries in the region (Usman and Reason, 2004; Tadross et al., 2005; Reason et al., 2005c; Hachigonta and Reason, 2006). Therefore, there is considerable motivation for a better understanding of the relationships between southern African rainfall variability and regional SST forcing.

12 1012 K. HANSINGO AND C. J. C. REASON Acknowledgements This work was jointly supported by the National Research Foundation (NRF) of South Africa and the global change SysTem for Analysis, Research and Training (START). We thank Dr Mark Tadross for providing the C code (later converted to Fortran 90 by the first author) for manipulating dates in the initial dump files of the model. References Behera SK, Yamagata T Subtropical SST dipole events in the southern Indian Ocean. Geophysical Research Letter 28: Cook C, Reason CJC, Hewitson BC Wet and dry spells within particularly wet and dry summers in the South African summer rainfall region. Climate Research 26: D Abreton PC, Lindesay JA Water vapour transport over southern Africa during wet and dry early and late summer months. Journal of Climatology 13: Florenchie P, Lutjeharms JRE, Reason CJC, Masson S, Rouault M The source of Benguela Niños in the South Atlantic Ocean. Geophysical Research Letter 30: DOI: /2003GLO Florenchie P, Reason CJC, Lutjeharms JRE, Rouault M, Roy C, Masson S Evolution of interannual warm and cold events in the Southeast Atlantic Ocean. Journal of Climate 17: Haarsma RJ, Campos EJ, Molteni F Atmospheric response to South Atlantic SST dipole. Geophysical Research Letter 30: DOI: /2003GL Hachigonta S, Reason CJC Interannual variability in dry and wet spell characteristics over Zambia. Climate Research 32: Hansingo K, Reason CJC Sensitivity of the atmospheric response to sea-surface temperature forcing in the southwest Indian Ocean: a regional climate modeling study. South African Journal of Science 102: Hirst AC, Hastenrath S Atmosphere-ocean mechanisms of climate anomalies in the Angola-Tropical Atlantic Sector. Journal of Physical Oceanography 13: Mason SJ, Jury MR Climatic variability and change over southern Africa: a reflection on underlying processes. Progress in Physical Geography 21: Pope VD, Gallani ML, Rowntree PR, Stratton RA The impact of new parameterizations in the Hadley Centre climate model: HadAM3. Climate Dynamics 16: Reason CJC Subtropical Indian Ocean SST dipole events and southern African rainfall. Geophysical Research Letter 28: Reason CJC Sensitivity of the southern African circulation to dipole sea-surface temperature patterns in the South Indian Ocean. International Journal of Climatology 22: Reason CJC Tropical cyclone Dera, the unusual 2000/01 tropical cyclone season in the southwest Indian Ocean and associated rainfall anomalies over Southern Africa. Meteorology and Atmospheric Physics 97: Special issue on tropical cyclones, DOI: /s Reason CJC, Jagadheesha D. 2005a. Relationships between South Atlantic SST variability and atmospheric circulation over the South African region during austral winter. Journal of Climate 18: Reason CJC, Jagadheesha D. 2005b. A model investigation of recent ENSO impacts over southern Africa. Meteorology and Atmospheric Physics 89: Reason CJC, Mulenga H Relationships between South African rainfall and SST anomalies in the southwest Indian Ocean. International Journal of Climatology 19: Reason CJC, Rouault M Sea surface temperature variability in the tropical southeast Atlantic Ocean and West African rainfall. Geophysical Research Letter 33: L DOI: /2006GL Reason CJC, Hachigonta S, Phaladi RF. 2005c. Interannual variability in rainy season characteristics over the Limpopo region of southern Africa. International Journal of Climatology 25: Reason CJC, Jagadheesha D, Tadross M A model investigation of interannual winter rainfall variabilityâ over southwestern South Africa and associated ocean-atmosphere interaction. South African Journal of Science 99: Reason CJC, Landman W, Tennant W. 2006b. Seasonal to decadal prediction of Southern African climate and its links with variability of Atlantic Ocean. Bulletin of the American Meteorological Society 87: Reason CJC, Florenchie P, Rouaullt M, Veitch J. 2006a. Influences of large scale climate modes and Agulhas system variability on the BCLME region, Chapter 10 in Large Marine Ecosystems, Volume 14 Predicting a Large Marine Ecosystem, Shannon V, Hempel G, Malanotte-Rizzoli P, Moloney C, Woods J (eds). : Benguela; Reynolds WR, Smith TM Improved global sea surface temperature analyses using optimum interpolation. Journal Climate 7: Richard Y, Trzaska S, Roucou P, Rouault M Modification of the southern African rainfall variability/enso relationship since the late Climate Dynamics 16: Robertson AW, Farrara JD, Mechoso CR Simulations of the atmospheric response to South Atlantic sea surface temperature anomalies. Journal of Climate 16: Rouault M, Florenchie P, Fauchereau N, Reason CJC Southeast tropical Atlantic warm events and southern African rainfall. Geophysical Research Letter 30: DOI: /2002GL Shannon LV, Boyd AJ, Brundrit GB, Taunton-Clark J On the existence of an El Nino-type phenomenon in the Benguela system. Journal of Marine Research 44: Tadross MA, Hewitson BC, Usman MT The interannual variability of the onset of the maize growing season over South Africa and Zimbabwe. Journal of Climate 18: Tennant W An assessment of intraseasonal variability from 13-yr GCM simulations. Monthly Weather Review 13: Usman M, Reason CJC Dry spell frequencies and their variability over southern Africa. Climate Research 26: Washington R, Preston A Extreme wet year over southern Africa: role Indian Ocean sea surface temperatures. Journal of Geophysical Research 111: D DOI: /2005JD Williams CJR, Kniveton DR, Layberry R Influence of South Atlantic sea surface temperatures on rainfall variability and extremes over Southern Africa. Journal of Climate(accepted). Zhang GJ, McPhaden MJ The relationship between sea surface temperature and latent heat flux in the equatorial Pacific. Journal of Climate 8: