Diagnosing the Droughts and Floods in Equatorial East Africa during Boreal Autumn 2005–08 (original) (raw)

1. Introduction

An earlier report in this journal (Hastenrath et al. 2007) diagnosed the circulation mechanisms leading to the 2005 drought in equatorial East Africa, the failure of the boreal autumn “short rains.” During January 2009 repeated media reports were again received of severe drought conditions in many parts of Kenya. The failure of the boreal autumn short rains of 2008 and associated impacts on food, water, and other resources has been declared a national disaster in Kenya. This invited the present endeavor to explore the underlying circulation mechanisms. At the same time the circulation and climate anomalies of the intervening 2006 and 2007 boreal autumn rainy seasons were also considered. For a comprehensive review of earlier publications, reference is made to the previous report (Hastenrath et al. 2007). On this basis the present short note is limited to the presentation of the most crucial evidence. Section 2 describes the data sources, section 3 reviews the essential background, section 4 presents the findings for the boreal autumn short rains of 2005, 2006, 2007, and 2008, and a synthesis is offered in the closing section 5.

2. Data

As in our previous report for 2005 (Hastenrath et al. 2007) the data sources for the present study consist of rain gauge measurements and global surface and upper-air fields from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis (Kalnay et al. 1996; Kistler et al. 2001). The spatial resolution in NCEP–NCAR is 2.5° latitude–longitude squares. Of interest here are surface wind, sea level pressure (SLP), and 500-mb omega vertical motion, as indicated on Fig. 1. For sea surface temperature (SST) we used the extended reconstructed sea surface temperature (ERSST.v3) dataset (Smith et al. 2008), with a spatial resolution of 2.0° latitude–longitude squares.

As in the previous report (Hastenrath et al. 2007), the precipitation of 2005, 2006, 2007, and 2008 was compared to the 1948–87 reference mean for two groups of four and of five stations in Kenya, as illustrated in Fig. 2 and listed in Table 1. In earlier work (Hastenrath and Polzin 2004, 2005; Hastenrath 2007; Hastenrath et al. 2007) we also used an index RON of October–November rainfall at the coast of East Africa, compiled from 7 stations for the period 1958–97.

As for the previous report (Hastenrath et al. 2007) various index series for October–November were compiled from these data as detailed in Fig. 1. UEQ (5°N–5°S, 60°–90°E) is an index of the zonal component of surface wind over the central equatorial Indian Ocean. The pressure index PW and the SST index TW are for a domain in the west (ocean portion of the quadrangle 7.5°N–7.5°S, 40°–50°E), the pressure index PE and the SST index TE are for a block in the east (7.5°N–7.5°S, 90°–100°E), and PWE = PW − PE and TWE = TW − TE represent the zonal pressure and SST gradients along the equator. SIW (5°–12.5°S, 60°–90°E) is an index of the total wind speed in the downstream portion of the south Indian Ocean trade winds. W5_ω_ (2.5°N–2.5°S, 30°–50°E) and E5_ω_ (2.5°N–2.5°S, 100°–120°E) are indices of the 500-mb omega vertical motion. The analyses of October–November of 2005, 2006, 2007, and 2008 are compared to the 1958–97 reference period. The index of the phases of the Southern Oscillation (SOI), the difference of the pressure anomalies Tahiti minus Darwin, was used, as in earlier studies (Hastenrath et al. 1993; Hastenrath 2007; Hastenrath et al. 2007).

Surface ship observations in the Indian Ocean from the Comprehensive Ocean–Atmosphere Data Set (COADS; Woodruff et al. 1987, 1993), in addition to the aforementioned NCEP–NCAR and rain gauge data, were used in Fig. 3 based on earlier work (Hastenrath and Polzin 2004; Hastenrath 2007).

For illustration of the hydrological conditions of Lake Victoria (Fig. 4), data on water levels were drawn from a U.S. Department of Agriculture (USDA) Web site (http://www.pecad.fas.usda.gov/cropexplorer/global_reservoir).

3. Background

A compact review of the annual cycle of circulation and climate in the equatorial Indian Ocean and pertinent literature is offered in the earlier report (Hastenrath et al. 2007). For comprehensive documentation, reference is made to a series of atlases, a book, and journal articles (Hastenrath and Lamb 1979a,b; Hastenrath and Greischar 1989, 1991; Hastenrath 1995, 57–66, 186–197; Hastenrath 2000; Hastenrath et al. 2002; Hastenrath and Polzin 2003, 2004, 2005). With reference to Figs. 1 and 3, the present brief synopsis focuses on the October–November core of the short rains season in East Africa.

An equatorial zonal circulation cell (Fig. 1) features ascending motion over Indonesia, divergent westward flow in the upper troposphere, and subsidence over the western Indian Ocean and East Africa. This feeds into the surface westerlies, which are forced by the strong zonal pressure gradient, with higher pressure in the west and lower pressure in the east. The westerly surface winds drive the eastward equatorial jet or Wyrtki jet in the upper ocean (Wyrtki 1973).

Turning to interannual variability, the chain of causalities in the coupled atmosphere–ocean system merits attention. The surface equatorial westerlies UEQ form the backbone of the system. Plausibly, UEQ is strong with steep eastward pressure gradient along the equator, large PWE; the PWE depends on both the pressure in the west PW and in the east PE. The UEQ is further favored by weak wind speed in the downstream portion of the south Indian Ocean trade winds SIW. The wind exerts forcing on the SST field and this can affect the pressure field by hydrostatic forcing. Subsidence, high pressure, and cold SST are not conducive for precipitation. Figure 1 gives an orientation on the equatorial zonal circulation cell and the areas of pertinent indices. Figure 3 shows the close associations within this finely tuned coupled system: the tight correlations between UEQ and the vertical motion in the west W5_ω_ and east E5_ω_; between UEQ and PWE, TWE, and less significantly with SOI; and the correlation of East African rain RON with equatorial westerlies UEQ, vertical motion W5_ω_, and less significantly with SOI.

Since submission of the present note, another article on related subject has appeared (Ummenhofer et al. 2009), reporting on model experiments with SST forcing. As advised by a reviewer of the previous report (Hastenrath et al. 2007), attention is called to some confusion in the literature. As evidenced before (Hastenrath and Polzin 2004, 2005; Hastenrath 2007), there is no dipole/seesaw between west and east in either pressure or temperature and no indication of local forcing of temperature on pressure. Fast UEQ enhances the zonal temperature gradient, thus tightening the inverse relationship between the zonal gradients of pressure and temperature. A new frontier is defined in the question: what controls the pressure pattern over the Indian Ocean basin?

4. Short rains of 2005–08

Rain gauge measurements for the short rains of 2005, 2006, 2007, and 2008 in Kenya are listed in Table 1 for two groups of stations identified in Fig. 2. The concomitant departures of pertinent circulation indices are listed in Table 2.

As discussed in the previous report (Hastenrath et al. 2007), during the 2005 drought the equatorial westerlies UEQ were strongly accelerated, along with enhanced pressure in the west PW and zonal pressure gradient PWE, and consistent with the weakened southern trade winds SIW; further subsidence in the west W5_ω_ and ascending motion in the east E5_ω_ were enhanced, all manifestations of a strong equatorial zonal circulation cell. During the 2006 floods UEQ was slow, the zonal pressure gradient PWE slack, the southern trade winds SIW fast; further the subsidence in the west W5_ω_ and the ascending motion in the east E5_ω_ were greatly reduced, all indications of a weak zonal circulation cell. The 2007 autumn rainy season was near average to moderately dry, the westerlies UEQ and zonal pressure gradient PWE near average, subsidence in the west W5_ω_ somewhat enhanced; SIW was weaker. So, overall, 2007 featured no strong anomalies in either rainfall or circulation. Then severe drought occurred again in 2008, with fast westerlies UEQ and slow southern trade winds SIW, although the pressure data bear out little enhancement of the zonal pressure gradient PWE and little increase in the subsidence in the west W5_ω_.

Complementing Table 2 are maps of SST departures in October–November of the four years, not reproduced here. Thus in the dry regime of 2005, for which SST map is presented in Fig. 4g of the previous report (Hastenrath et al. 2007), waters were anomalously warm at the eastern extremity to the south of the equator (off the coasts of Sumatra and Java), consistent with the reduced wind stress forcing related to the weak SIW; the weak SIW further favored the development of strong UEQ, which is pivotal for the drought. By contrast, in the anomalously wet regime 2006 extreme cold departures were found in the waters to the southwest of Sumatra and Java, consistent with the forcing by the enhanced southeast trade winds SIW; the strong SIW further hindered the development of the equatorial westerlies UEQ, which resulted in the extreme floods. In the near-average short rains of 2007, plausibly, the SST departure patterns were not pronounced. The dry regime 2008 featured cold anomaly at the western extremity of the equatorial zone.

Thus overall in these 4-year seasons, the anomalies in East African rainfall were closely related to the intensity of the surface equatorial westerlies UEQ, consistent with earlier findings (Hastenrath et al. 1993, 2007; Hastenrath and Polzin 2004, 2005; Hastenrath 2007). The relation of the westerlies UEQ to the zonal pressure gradient PWE and the vertical motion W5_ω_ and E5_ω_ was particularly pronounced in 2005 and 2006. The zonal pressure gradient PWE was in the 2005 drought mostly controlled by PW and in the 2006 floods by PW and PE. For 2007 and 2008 the data yield weak PWE, but the weak SIW was favorable for the UEQ.

The Southern Oscillation is pertinent only to the extent that pressure variations in the Australasian region, Darwin, contribute to anomalies in the zonal pressure gradient along the Indian Ocean equator. Thus the SOI phase was helpful during the droughts of 2005 and 2008 and the floods of 2006.

Complementing the rain gauge measurements listed in Table 1, Fig. 4 updates the documentation of Lake Victoria water level in Fig. 6 of the previous report (Hastenrath et al. 2007). Following the 2005 drought water levels stayed low; after the floods of autumn 2006 they rose again to a highest level in May 2008, decreased into the dry season, with little rise into the deficient boreal autumn rainy season of 2008, and further drop thereafter.

5. Conclusions

Equatorial East Africa has two rainy seasons centered on March–April and October–November. During these short time intervals surface westerlies sweep the central equatorial Indian Ocean. In boreal autumn these westerlies become part of a powerful zonal–vertical circulation cell along the Indian Ocean equator (Hastenrath et al. 1993; Hastenrath and Polzin 2004, 2005; Hastenrath 2007). The intensity of this circulation cell, with subsidence in the west, controls the interannual variability of the boreal autumn rains.

Following up on the earlier report on the 2005 drought (Hastenrath et al. 2007), the present short note promptly diagnosed the circulation mechanisms of the 2008 drought, taking advantage of the timely availability of the NCEP–NCAR reanalysis and Kenya rain gauge data. At the same time conditions during the 2006 and 2007 years were also examined. During the 2006 floods the surface equatorial westerlies were slack, during the near-average rains of 2007 they were near average, and during the 2005 and 2008 droughts the westerlies were anomalously fast.

Acknowledgments

This study was supported by the Variability of Tropical Climate Fund of the University of Wisconsin Foundation. We appreciate the timely availability of the NCEP–NCAR reanalysis and Kenyan rain gauge data, which made this prompt appraisal of the recent circulation and climate anomalies possible. We thank the anonymous reviewers for helpful comments.

REFERENCES

Fig. 1.

Fig. 1.

Fig. 1.

(b) Orientation map and (a) zonal–vertical cross section detailing October–November circulation indices. In the map, solid lines delineate the domains of the indices of surface zonal wind UEQ (5°N–5°S, 60°–90°E), of pressure PW and sea surface temperature TW (7.5°N–7.5°S, 40°–50°E), and of pressure PE and sea surface temperature TE (7.5°N–7.5°S, 90°–100°E). Dashed lines delineate the domains of the indices of 500-mb omega vertical motion W5_ω_ (2.5°N–2.5°S, 30°–50°E) and E5_ω_ (2.5°N–2.5°S, 100°–120°E) and of the total wind speed in the downstream portion of the south Indian Ocean trade winds SIW (5°–12.5°S, 60°–90°E). Dotted lines indicate the domain of the rainfall index RON for the coast of East Africa, and the dotted-line quadrangle in East Africa encloses the domains of the rain gauge stations shown in Fig. 2. Of further interest are the indices for the zonal gradients of pressure PWE = PW − PE and temperature TWE = TW − TE, as well as the SOI (pressure difference Tahiti minus Darwin). Shading in the zonal–vertical section (a) illustrates schematically the equatorial zonal circulation cell in October–November.

Citation: Journal of Climate 23, 3; 10.1175/2009JCLI3094.1

Fig. 2.

Fig. 2.

Fig. 2.

Orientation map of rain gauge stations in Kenya, with domain indicated by dotted lines in Fig. 1. Stations available for 1948–87 and 2005–08 are at the coast Lamu (LM), Malindi (MD), Mombasa (MB), and Voi (VO); and in the interior highlands Nanyuki (NK), Nyeri (NY), Meru (ME), Embu (EM), and Machakos (MK).

Citation: Journal of Climate 23, 3; 10.1175/2009JCLI3094.1

Fig. 3.

Fig. 3.

Fig. 3.

Diagram of correlations between the circulation indices in Fig. 1, for October–November 1958–97. Correlation coefficients are in hundredths and significant at the 1% level. Shading illustrates schematically the equatorial zonal circulation cell in October–November.

Citation: Journal of Climate 23, 3; 10.1175/2009JCLI3094.1

Fig. 4.

Fig. 4.

Fig. 4.

Lake Victoria annual cycle: (a) water level from satellite, 1993–2002 mean, in m; (b) same as (a) for 2005; (c) same as (a) for 2006; (d) same as (a) for 2007; (e) same as (a) for 2008; (f) same as (a) for 2009.

Citation: Journal of Climate 23, 3; 10.1175/2009JCLI3094.1

Table 1.

Rainfall totals in mm, at station groups shown in Fig. 2, in October–November (ON) and September–December (SOND), during 2005, 2006, 2007, 2008, as compared to long-term reference mean 1948–87.

Table 1.

Table 1.

Table 2.

Circulation indices identified in Fig. 1 for ON of 2005, 2006, 2007, and 2008. With reference period of 1958–97, values listed to the left are dimensionless in terms of standard deviation and to the right as departures in units of mb for PW, PE, PWE; in m s−1 for UEQ and SIW; in 10−4 mb s−1 for W5_ω_ and E5_ω_; in °C for TW, TE, TWE.

Table 2.

Table 2.