-
Figure 1 shows the SST composite anomalies during strong positive SIOD events with respect to the 1981–2010 climatology. Based on the > 1.0 standardized SDI threshold, during the analysis period, 1981, 1982, 2006, and 2017, were selected as years with strong positive SIOD signal during JFM. It can be seen that the positive SIOD SST pattern is associated with warm SST anomalies south of Madagascar and cool SST anomalies off the western coast of Australia. Also, Fig. 1 indicates anticyclonic circulation anomalies over much of the South Indian Ocean, as also noted by Behera and Yamagata (2001) during strong positive SIOD events.
Figure 1. JFM composite of SST anomaly and surface wind anomaly for years with strong positive SIOD events. Anomalies were calculated as JFM mean for 1981, 1982, 2006, and 2017 minus JFM mean for the 1981−2010 period. Black contour lines and color-filled regions show where the SST anomalies exceed the 90% confidence limit based on the block permutation test. The contour interval is 0.2°C. Thick contours (and red shading) show positive SST anomalies; dashed contours (and blue shading) show negative SST anomalies. Green vectors represent wind anomalies. The wind vector scale is shown above the map. The selected positive SIOD years are those when the JFM standardized SDI exceeded the 1.0 threshold.
-
Figure 2 shows the JFM precipitation climatology from the CPC and NCEP-NCAR data sets. During austral summer, the western subtropical region in southern Africa is relatively drier due to a combination of its proximity to the cold Benguela current and to the anticyclonic circulation of the South Atlantic Ocean high-pressure, which drives moisture offshore. Also, the Botswana high that exists at mid-levels (e.g., near 500 hPa) over Botswana/Namibia contributes to the strongly stable atmosphere and subsidence in this region (Reason, 2016). However, the eastern region of southern Africa near the Mozambique Channel and the Agulhas current are relatively wet. Based on the spatial correlation coefficient (R=0.8), the NCEP-NCAR reanalysis mirrors the spatial variation of austral summer precipitation in the study region as obtained from the CPC data set. Nonetheless, there are marked differences in the magnitude of precipitation from both data sets, especially in the central equatorial region and the northeastern parts of South Africa where the NCEP-NCAR reanalysis overestimates precipitation. Also, this dataset does not adequately resolve the tight rainfall gradients or topographic influences on rainfall. It should be equally noted that according to Thorne et al. (2001), the CPC method incorporates real-time rain gauge data where sufficient rain gauges are available.
According to the JFM SLP climatology, the Mascarene high is located southward of its climatological position during austral winter (Fig. 3). Thus, it drives moist southeast trade winds that converge with the moist cross-equatorial northeast trade winds over the tropical South Indian Ocean, enhancing the convergence within the Inter-Tropical Convergence Zone (ITCZ). As a result, from about 50°E eastward, north of Madagascar, the vertical velocity at 850 hPa is negative (representing upward motion) and relative vorticity is negative (implying cyclonic vorticity) so that precipitation is enhanced in this region. Also, the flux of moisture in these regions is altered from easterly to westerly by the cyclonic circulation. To the northwest of Madagascar, the cross-equatorial northeasterly moisture transport penetrates both the southern African mainland and the Mozambique Channel where they become westerly and converge with southeasterly moisture transport. Thus, a region of enhanced cyclonic relative vorticity is evident in the Mozambique Channel, referred to as the Mozambique Channel Trough (MCT). The MCT plays a vital role in the hydroclimate of Madagascar and northern Mozambique. Also, the MCT can influence the inter-annual variability of rainfall in the southern African mainland (Barimalala et al., 2020). The MCT, in addition to the high topography of Madagascar, prevents direct moisture advection into the southern African mainland from the southwest Indian Ocean. This is because the topography of Madagascar deflects moist easterly winds from the Mascarene high towards the south of Madagascar and then the MCT alters the easterly moisture flux to the northwest (Barimalala et al., 2018).
Figure 3. JFM climatology for the 1981−2010 period for (a) vertical velocity (color), SLP (black contour lines), and moisture flux at 850 hPa (green vectors, vector scale shown at top), (b) divergence (color) and relative vorticity (black contour lines), thick contours indicate positive relative vorticity, dashed contours indicate negative relative vorticity (contour interval of 5 × 10−6 s−1), and (c) precipitation (mm d−1).
A low-pressure system is equally evident in southern Angola and northern Namibia. Figure 3b shows that the low-pressure system is characterized by enhanced low-level convergence and cyclonic relative vorticity. This is because the cross-equatorial moist northeast wind and moist westerly wind from the tropical southeast Atlantic Ocean feed into this system; hence by the nature of this cyclonic circulation, moisture is transported to the eastern parts of southern Africa (Reason and Smart, 2015). This system is called the Angola low. As the Mascarene high ridges into the northeastern parts of South Africa, southeasterly moisture fluxes also penetrate the southern African region through the warm Agulhas and Mozambique currents, so that low-level fluxes of moisture converge in the eastern regions. Over the southern African area, upward motion is evident, which can be attributed to a combination of continental heating during the austral summer season and low-level moistening of the boundary layer. The maritime regions with enhanced cyclonic relative vorticity and upward motion receive the highest rainfall while regions with anticyclonic relative vorticity and downward motion, such as over the semi-permanent high-pressure system and its ridging northeast of South Africa, receive less rainfall owing to enhanced subsidence.
-
Figures A2 and A3 show the classified CTs in the study region, and their dominant periods, respectively. The CTs are patterns of atmospheric circulation in Africa south of the equator, characterized by the different modes of SLP variability. The seasonal distribution of positive and negative phases of CTs is mostly asymmetric because each analyzed PC component can be designated as a mode of variability and the positive and negative phases are opposing states of the same atmospheric mode, i.e., their respective patterns can neither occur at the same time nor dominate in a given period (e.g., month), resulting from the asymmetry. Also, seasonality such as seasonal variations in diabatic heating and migration of the ITCZ contributes to the occurrence of the CTs over time and their mechanism. Thus, though the CTs can occur at any time of the year given the imprecise/fuzzy nature of atmospheric processes, their dominant periods can be documented as in Fig. A3.
CT1+, CT3−, CT4+, CT5+, CT6+, CT7− tend to dominate during the colder months in the southern hemisphere (i.e., May to August). Conversely, their inverse phases (i.e., CT1−, CT3+, CT4−, CT5−, CT6− and CT7+) tend to dominate during the warmer months in the southern hemisphere (i.e., November to March). During austral winter (JJA and the adjacent months), the ITCZ is positioned more equatorward. This can imply (i) a more northward track of the southern hemisphere mid-latitude cyclone, leading to enhanced westerly winds over large parts of southern Africa; and (ii) dominance of the subtropical high-pressure over the southern African region. Both scenarios lead to rainfall suppression in large parts of southern Africa. Such seasonal patterns are characterized by the SLP modes (i.e., CTs in Fig. A2). CT1+ and CT3− characterize the circulation pattern associated with the northward track of the mid-latitude cyclones during the winter season. CT4+, CT5+ and CT6+ and CT7− characterize the circulation pattern associated with enhanced anti-cyclonic activity over the landmasses. During the inverse phase of the CTs (i.e., in the warmer seasons when SST is also enhanced over the adjacent oceans as can be inferred from the enhanced cyclonic activity in the southwest Indian Ocean under the warm season CTs), diabatic heating is enhanced, the ITCZ and the subtropical highs are located more southward, and so asymmetric patterns (opposing their inverse phases) that enhance rainfall formation become plausible (as can be inferred from Table 1). Generally, as noted by Morioka et al. (2015) the northward (southward) migration of the subtropical high-pressure systems that is evident during the colder season (warmer season) dominant CTs is associated with the suppression (enhancement) of rainfall over southern Africa.
CT Temporal (mm d−1) Spatial (mm d−1) CT1+ −1.00 −0.34 CT1− 0.64 0.51 CT2+ 0.51 0.40 CT2− 0.05 0.38 CT3+ 0.48 0.45 CT3− −0.53 −0.55 CT4+ −0.46 −0.49 CT4− 1.22 0.09 CT5+ −0.47 −1.22 CT5−* 2.01* 1.61 CT6+ −0.35 −0.56 CT6−* 2.39* 1.76 CT7+* 2.63* 2.01 CT7− −0.77 −1.27 CT8+ 0.22 0.04 CT8− 0.65 0.56 CT9+ 0.09 −0.07 CT9− 1.45 1.31 Table 1. Temporal and spatial median anomaly values of precipitation under each of the classified CT. (*) implies CTs with values greater than 2 mm d−1.
Further, some of the other CTs are dominant during the transition seasons, i.e., austral spring and autumn, for example, CT9+/CT9−; while some of the other CTs do not show the tendency of seasonality. Moreover, the imprint of climate drivers can be found in most of the CTs (e.g., Ibebuchi, 2021a, b) so that anomalies of climate drivers can equally control the occurrence and amplitude of the CTs over time.
-
For all the classified CTs, Table 1 shows the temporal and spatial anomaly values of precipitation. Subjectively using > 2.0 mm threshold to characterize rain-bearing CTs, CT5−, CT6−, and CT7+ (i.e., from Fig. A2) are further selected. CT9− is also included even though it falls short of the criteria; nonetheless, subsequent analysis indicates it can be associated with significantly above-average rainfall in specific domains. From Fig. A3, overall all the aforementioned selected CTs are dominant during late austral spring to early austral autumn. As already noted, since the landmasses are very large with heterogeneous climate types, the so-called rain-bearing CTs were subjectively selected because their mechanisms (as explained below) can significantly bring about above-average rainfall over considerable parts of the continent. Table 1 (and highlighted further figures) indicated that the selected CTs have more grid points associated with above-average rainfall, when the composite anomalies are computed and tested for statistical significance at each grid box.
Figures 4 and 5 present the composite patterns of the four selected JFM rain-bearing CTs in Africa south of the equator. Figure 6 shows the anomalous rainfall patterns that are associated with them. The general attributes of the CTs are (i) enhanced low-level cyclonic relative vorticity and moisture flux convergence in southern Angola and northern Namibia; and (ii) moisture fluxes into southern Africa mainland from the tropical South Atlantic Ocean, the tropical Indian Ocean, and the southwest Indian Ocean. Variability in the strength and location of the semi-permanent high-pressure systems and the MCT, coupled with the associated implications for the regional patterns of moisture transport, mark the differences in the large-scale atmospheric circulation patterns under each CT. The JFM composite of CT6− features (i) enhanced downward motion within the western branch of the Mascarene high south of Madagascar and (ii) enhanced anticyclonic relative vorticity within the western branch of the Mascarene high, ridging into the northeastern parts of South Africa. This results in enhanced low-level southeasterly moisture fluxes into the southern African mainland. Coupled with moisture fluxes from the tropical South Atlantic Ocean and the tropical Indian Ocean, under this CT6− pattern, enhanced and widespread moistening of the boundary layer can be generally expected in the study region. Indeed, the rainfall composite reveals a positive rainfall anomaly across most of Africa south of the equator, extending diagonally into the South Indian Ocean. However, within the western branch of the Mascarene high and its ridging into parts of eastern regions of southern Africa, rainfall is significantly reduced, possibly due to subsidence. Thus, CT6− can be associated with rainfall in large parts of the region during JFM, except for the eastern regions where the Mascarene high ridges into the northeastern parts of South Africa.
Figure 4. JFM rain-bearing circulation types in the study region. Black contour is SLP at 3 hPa intervals. Color is vertical velocity. Vectors represent moisture flux (scale at top of figure).
Figure 5. Same as Fig. 4 but for moisture flux convergence (color) and relative vorticity (black contours). Thick contours show positive relative vorticity and dashed contours show negative relative vorticity (contour interval is 5 × 10−6 s−1).
Figure 6. Rainfall composite anomaly for the JFM rain-bearing CTs in Fig. 5. Anomaly is calculated as the difference between the mean JFM precipitation of the days when the CT occurred and the JFM climatology of 1981−2010. Contour lines (mm d−1) show regions that exceed the 90% confidence limit. Dashed contours show regions with a negative anomaly and thick contours show regions with a positive anomaly. Contour interval is 0.8 mm d−1.
The large-scale features of CT5− and CT7+ are similar in the sense that, unlike in CT6−, the western branch of the Mascarene high is relatively weakened. Thus, southeasterly moisture fluxes into the southern African mainland arise mainly through the activity of the South Atlantic Ocean high-pressure ridging south of South Africa, where enhanced anticyclonic circulation, downward motion, and low-level anticyclonic relative vorticity are evident during these CTs (Figs. 4 and 5). Southeasterly moisture fluxes by the Mascarene high, which is located further to the east, are deflected southward under these CTs by the Madagascar topography and are partly adjusted to westerly by the MCT. Given a weaker state of the Mascarene high, both CTs are associated with enhanced cyclonic activity in the southwest Indian Ocean. Under CT7+, a continental tropical low is evident in the western parts of southern Africa, and enhanced upward motion extends from Angola, off the southeast coast of South Africa. As a result, both the eastern part of southern Africa and the southwest Indian Ocean are significantly moist (Fig. 6). Under CT5−, enhanced upward motion is evident in the Oceans southeast of Madagascar; thus, this region is significantly moist. Recall CT5− is the same CT related to El Niño, as documented in Ibebuchi (2021c). Here, its composite patterns further reveal that in addition to disrupting moisture fluxes advected by the South Atlantic Ocean high pressure ridging south of South Africa, El Niño weakens the hydroclimate of southern Africa by enhancing upward vertical motion and wet conditions over the western branch of the Mascarene high (i.e., further weakening of anticyclonic circulation and moist southeast winds). A plausible reason why rainfall is mostly enhanced to the east compared to CT6− may be due to weaker advection resulting from the weaker state of the Mascarene high. Thus, areas close to the southwest Indian Ocean receive rainfall under CT5− and CT7+ and advective flow is through the activity of the South Atlantic Ocean high-pressure, ridging south of South Africa (Fig. 6).
Under CT9−, the MCT is relatively more active. This results in the lessening of moisture fluxes from the tropical Indian Ocean in penetrating the Angola low since cross-equatorial northeast winds are adjusted to westerly towards Madagascar by the MCT and southeast trade winds are more westerly. The direct implication is above-average rainfall in Madagascar and northern Mozambique but diminished rainfall in southern Africa (e.g., Barimalala et al., 2020). While CT9− is not explicitly rain-bearing in southern Africa (Table 1), it is associated with above-average rainfall in Madagascar.
The analysis of CT6− and CT9− indicates that due to the high topography of Madagascar and the MCT, the strength of the anticyclonic circulation within the western branch of the Mascarene high controls the extent of moisture advection into southern Africa by southeast winds, and also controls the degree with which cyclonic/convective activity might be widespread over the southwest Indian Ocean. When the western branch of the Mascarene high is weak, e.g., CT7+, enhancement of convective activity, as indicated by significant cyclonic and positive precipitation anomalies under CT7+, can be expected to be widespread over the southwest Indian Ocean.
-
Correlation analysis was used to relate the inter-annual variability in the amplitudes of the classified modes and the anomalies of the SDI, during JFM. Table 2 shows the results. It can be seen that within the analysis period, the SDI can be related to mode 4 (i.e., CT4+/CT4−) and mode 7 (i.e., CT7+/CT7−). The negative (positive) correlation under mode 4 (mode 7) implies that positive SIOD is related to CT4− (CT7+). Figure A2 shows that both CTs (i.e., CT4− and CT7+) have something in common: the western branch of the Mascarene high is weak, allowing enhanced cyclonic activity in the southwest Indian Ocean. This is more evident under CT7+, which accounts for why it is relatively more rain-bearing (c.f. Table 1). The correlations imply that positive SIOD can increase the amplitude of CTs associated with enhanced cyclonic activity in the southwest Indian Ocean. Since CT7+ is selected to be rain-bearing and the correlation between the SIOD is also stronger with mode 7, further analysis is focused on CT7+.
Mode Correlation coefficient (R) p-value Mode 1 −0.10 0.39 Mode 2 −0.03 0.78 Mode 3 0.04 0.81 Mode 4* −0.303 0.03* Mode 5 −0.10 0.46 Mode 6 0.29 0.06 Mode 7* 0.402 0.00* Mode 8 0.05 0.86 Mode 9 −0.24 0.05 Table 2. Correlations between the annual mean SDI and the annual mean PC loadings from the classified modes during JFM in the analysis period. (*) indicates correlations that are statistically significant based on the Kendall Tau at a 95% confidence level (i.e., p < 0.05).
A statistically significant relationship was found also between the frequency of occurrence of CT7+ and the SDI (R = 0.4), during JFM. On average, years with above-average SDI are likely to be associated with an above-average frequency of occurrence of CT7+ (Fig. 7). Thus, there could be a modulation of the positive SIOD climate signal over the occurrence of CT7+. According to Reason (2001), the major mechanism through which the positive SIOD brings enhanced rainfall to parts of southern Africa is enhanced warming and cyclonic activity in the southwest Indian Ocean; resulting in more moisture available to be advected into large parts of southern Africa by the anomalous southeast winds (c.f. Fig. 1). Under CT7+ the southwest Indian Ocean is significantly moister as a result of enhanced cyclonic activity in this region, making the relationship between CT7+ and the positive SIOD physically realistic.
-
Figure 8 shows the four classified JFM homogeneous regions of rainfall anomalies in Africa south of the equator. In region R1, a positive anomaly dominates over the western subtropical land area of south Africa. Region R2 shows a strong anomaly in the south-central regions of Africa south of the equator and adjacent oceans comprising the ocean waters in parts of the Mozambique Channel and the northern Agulhas current. Region R3 features a negative anomaly dominant in ocean waters south of Madagascar. Region R4 shows a positive anomaly dominant in Madagascar and the adjacent oceans.
Figure 8. JFM rainfall regions in Africa south of the equator. The region that is related to CT7+ is highlighted by the blue frame. The loadings cluster homogeneous regions of JFM rainfall anomalies. Opposing signs of the loadings under a given rainfall variability pattern indicate regions associated with positive or negative rainfall anomalies during the dominant period of the rainfall variability pattern.
Table 3 shows the correlations between mode 7 that is related to the SIOD and the rainfall regions. R2 (highlighted by the blue frame in Fig. 8) is statistically related to mode 7, i.e., CT7+/CT7−. The statistically significant relationship between R2 and mode 7 makes Region R2 the region of interest in this paper. The correlation revealed that CT7+ (CT7−) that is related to the positive (negative) SIOD is also related to the wet (dry) phase of R2.
Precipitation region Correlation coefficient p-value R1 0.23 0.05 R2 0.60 0.00* R3 0.08 0.49 R4 0.15 0.23 Table 3. Correlation analysis between the precipitation regions and the PC loadings of mode 7 during JFM. (*) indicates correlations that are statistically significant based on the Kendall Tau at a 95% confidence level (i.e., p < 0.05).
Figure 9 shows that the enhancement of relative cyclonic vorticity and low-level moisture flux convergence are associated with the time development of the wet phase of R2. Since R2 is influenced by the maritime regions over the adjacent oceans, Fig. 9 indicates that enhanced cyclonic circulation over the adjacent ocean waters from the Mozambique Channel towards northern Agulhas current can influence the time development of the wet phase of R2. Recall that under CT7+ enhancement of cyclonic activity and precipitation is also evident in the southwest Indian Ocean (c.f. Figs. 4, 5, and 6): physically, this can be a plausible reason why CT7+ and the wet phase of R2 are statistically related. Thus, during positive SIOD events, the circulation feature of CT7+ implies that the enhancement of convective (cyclonic) activity in the northern Agulhas current and the parts of the Mozambique Channel can induce rainfall in the south-central regions of Africa, south of the equator.
Figure 9. Physical mechanism associated with the time development of features over region R2. (a) Rainfall (shading, for regions with positive rainfall anomaly), SLP (black contour lines), and moisture flux anomaly (green vectors). Anomalies were calculated as the difference between the JFM months when R2 is most pronounced and the JFM climatology of 1981−2010. Only values exceeding the 90% confidence limit are plotted. The contour interval is 0.4 hPa, with thick contours showing positive anomalies and dashed contours showing negative anomalies (see scale at top of map). (b) The same as for (a) but for relative vorticity anomaly (contour lines) and moisture flux convergence anomaly (shading: blue indicates convergence and red indicates divergence). The contour interval for relative vorticity is 0.5 × 10−6 s−1.
CT | Temporal (mm d−1) | Spatial (mm d−1) |
CT1+ | −1.00 | −0.34 |
CT1− | 0.64 | 0.51 |
CT2+ | 0.51 | 0.40 |
CT2− | 0.05 | 0.38 |
CT3+ | 0.48 | 0.45 |
CT3− | −0.53 | −0.55 |
CT4+ | −0.46 | −0.49 |
CT4− | 1.22 | 0.09 |
CT5+ | −0.47 | −1.22 |
CT5−* | 2.01* | 1.61 |
CT6+ | −0.35 | −0.56 |
CT6−* | 2.39* | 1.76 |
CT7+* | 2.63* | 2.01 |
CT7− | −0.77 | −1.27 |
CT8+ | 0.22 | 0.04 |
CT8− | 0.65 | 0.56 |
CT9+ | 0.09 | −0.07 |
CT9− | 1.45 | 1.31 |