Relationships between intraseasonal rainfall variability...

Theor. Appl. Climatol. 82, 153–176 (2005)DOI 10.1007/s00704-005-0129-0

1 Department of Oceanography, University of Cape Town, Rondebosch, South Africa2 Tanzania Meteorological Agency, Dar es Salaam, Tanzania

Relationships between intraseasonal rainfallvariability of coastal Tanzania and ENSO

A. L. Kijazi1;2 and C. J. C. Reason1

With 7 Figures

Received February 20, 2004; revised December 15, 2004; accepted December 16, 2004Published online March 31, 2005 # Springer-Verlag 2005

Summary

The influence of ENSO on intraseasonal variability overthe Tanzanian coast during the short (OND) and long(MAM) rainy seasons is examined. In particular,variability in the rainfall onset, peak and end dates aswell as dry spells are considered. In general, El Ni~nnoappears to be associated with above average rainfallwhile La Ni~nna is associated with below average rainfallover the northern Tanzanian coast during OND, and tolesser extent MAM. Over the southern coast, the ENSOimpacts are less coherent and this region appears to be atransition zone between the opposite signed impacts overequatorial East and southern Africa. The increased northcoast rainfall during El Ni~nno years is generally due to alonger than normal rainfall season associated with earlyonset while reduced rainfall during La Ni~nna years tendsto be associated with a late onset, and thus a shorter thanaverage rainfall season. Wet conditions during El Ni~nnoyears were associated with enhanced convection andlow-level easterly anomalies over the equatorial westernIndian Ocean implying enhanced advection of moisturefrom the Indian Ocean while the reverse is true for LaNi~nna years. Hovm€ooller plots for OLR and zonal wind at850 hPa and 200 hPa show eastward, westward propagat-ing and stationary features over the Indian Ocean. It wasobserved that the propagating features were absent duringstrong El Ni~nno years. Based on the Hovm€ooller results,it is observed that the convective oscillations over theTanzanian coast have some of the characteristic featuresof intraseasonal oscillations occurring elsewhere in thetropics.

1. Introduction

The economy of Tanzania depends significantlyon rain fed agriculture, which is highly vulnerableto the amount and distribution of rainfall. Bothspatial rainfall anomalies over the region and tem-poral variability within the wet season itself can beprominent and lead to significant deficiencies. Insome areas of Tanzania, agriculture is limited bythe length of the rain season while in others it isamount limited (Mhita, 1984). Intraseasonal rain-fall forecasts can assist farmers in decisionsregarding planting, fertilising, pesticide applica-tion, and irrigation demand. Preliminary work onthe onset and end of the rainfall seasons overTanzania has been done by Alusa and Gwange(1978) and Mhita and Nassib (1987) but there isa need for investigation into the possible impactson intraseasonal rainfall of the El Ni~nno SouthernOscillation (ENSO) and other large scale modes.

On interannual scales, it has long been estab-lished that ENSO modulates rainfall variabilityover East Africa (e.g. Ropelewski and Halpert,1987; Janowiak, 1988; Ogallo, 1988; Hastenrathet al., 1993; Hutchinson, 1992; Nicholson andKim, 1997; Indeje et al., 2000; Mutai and Ward,2000; Amissah-Arthur et al., 2002). Recently,

Hastenrath and Polzin (2004) have shown thatEast African rainfall during the OND season isrelated to the strength of the equatorial surfacewesterlies over the Indian Ocean and the zonalpressure gradient across the basin. Both of thesevariables are modulated during ENSO as well asduring other modes such as the Indian OceanZonal Mode (Webster et al., 1999) or tropicaldipole mode (Saji et al., 1999), which also im-pacts strongly on East African OND rainfall.Typically, rainfall tends to be enhanced (reduced)over equatorial East Africa during El Ni~nno (LaNi~nna) events (e.g. Camberlin, 1995; Nicholson,1996). Given the strong evidence of significantENSO signals in the Indian Ocean (e.g. Cadet,1985; Nicholson and Kim, 1997; Tourre andWhite, 1997; Reason et al., 2000) and over muchof East Africa, an ENSO influence on East Africanintraseasonal rainfall variability might also beanticipated.

Intraseasonal oscillations (ISO) in the atmo-sphere are generally defined as fluctuations withperiods longer than the synoptic scale but shorterthan the seasonal time scale and are a prominentfeature of tropical weather particularly in theIndo-Pacific region (Madden and Julian, 1971,1994; Vincent et al., 1991). Initially identifiedby Madden and Julian (1971) they have becomewidely studied in recent times. However, mostprevious work on East African climate associa-tions with ENSO has looked at seasonal anoma-lies. With the exception of Mpeta and Jury(2001), relatively little research has beenconducted on Tanzanian intraseasonal variabilityalthough some work has been done for othercountries in southern Africa (e.g. Matarira andJury, 1992 for Zimbabwe; Levey and Jury, 1996for South Africa). Previous work did not look athow ENSO influences the intraseasonal variabil-ity of Tanzania and therefore this study intends toaddress that gap by attempting to investigate theISO responsible for short-term rainfall variabilityin the Tanzanian coast during ENSO years. Inparticular, there is interest in the relationshipsbetween the ISO and ENSO, e.g. are El Ni~nnoand La Ni~nna years characterised by coherent dif-ferences in intraseasonal wet and dry spells overTanzania?

Studies of ISO which link tropical Africa andthe adjacent oceans include Zhu and Wang(1993) who observed two prominent action cen-

tres in the central Indian and western Pacificoceans for tropical 30–60 day convective vari-ability. Anyamba (1992) identified a 20–30 dayoscillation in the tropical outgoing long waveradiation (OLR) spectra in the western IndianOcean and found that while the 40–50 day ISOis characterized by an eastward propagatingwave in the Indian and West Pacific Oceans,the 20–30 day ISO has a much weaker zonalpropagation. Rui and Wang (1990) documentedthe development and dynamic structure of typicalintraseasonal convection anomalies using pentadOLR and ECMWF derived 200 and 850 hPawind divergence. Mpeta and Jury (2001) foundthat convective events over southwesternTanzania were associated with an influx of north-easterly Indian Ocean monsoon flow followed byincreased westerly flow from the Guinea=Congoregion.

The area of Tanzania studied herein isbounded by latitudes 4–12� S and longitudes38.5–41� E. Two main rainy seasons are experi-enced in the north coast (4–8� S), the long rains(March–May) and the short rains (October–December), and these are associated with thenorthward and southward movement of the ITCZrespectively. The south coast (8–12� S) experi-ences one rain season from November to Apriland appears to be a transition region in its ENSOresponse between the opposite signed rainfallimpacts of equatorial East Africa and tropical–subtropical southeastern Africa.

2. Data and methods

NCEP gridded data for 1970–1999 were used toprovide insight into the circulations associatedwith onset, peak and withdrawal of rainfall.NOAA OLR data, together with rainfall stationand CMAP data, were used to explore the influ-ence of ENSO on Tanzanian coastal rainfall.CMAP data, which are available from 1979onwards, are a merge of satellite, model and raingauge data at 2.5� resolution (Xie and Arkin,1997) whereas the rainfall station data usedconsists of monthly totals obtained from theTanzania Meteorological Agency. The latter in-cludes observations from 10 rainfall stations scat-tered throughout the coastal region of Tanzania.The period 1970–1999 was selected for analysissince all stations had continuous quality records.

154 A. L. Kijazi and C. J. C. Reason

In order to investigate the influence of ENSO onTanzanian coastal rainfall, the normalised andspatially averaged rainfall indices for the northand south coasts (Fig. 1) were developed using

the means and standard deviations of the stationdata. A departure of at least 1 standard deviationabove (below) the mean defines an anomalouslywet (dry) season. Consistency of these resultswere checked with CMAP data for the overlap-ping 1979–1999 period.

Moisture fluxes were derived from the productof specific humidity and the horizontal wind vec-tor using NCEP re-analyses. However, due to thefact that NCEP has probable errors in specifichumidity and there are relatively few tropicalAfrican observations for assimilation into themodel (Kalnay et al., 1996), this study will focusmore on the sign than on the magnitude of themoisture flux.

To remove diurnal variability, the analyses use5 day pentads beginning with pentad 1 (1–5January) of a given year and ending at pentad 73(27–31 December). Pentad 12 contains an addi-tional day to include February 29 in the case of aleap year. CMAP data is used for the pentad anal-ysis since the station data are only available asmonthly totals. Mean values and anomalies ofpentad rainfall were calculated separately forthe north and south coasts. Pentad area-averagedrainfall time series were inspected and rainfallonset peak and end dates derived as per Table 1.

Fig. 1. Time series of normalised station rainfall anomaliesfor a) north part (4–8� S) of coast MAM season, b) northcoast OND season, c) south part (8–12� S) of coast JFMseason, d) south coast OND season

Table 1. Criteria used to define onset, peak and end dates ofthe rainy season

North coast short rain season (OND)Onset: 5-day rainfall at least 7.5 mm followed by three con-secutive pentads having rainfall amount of not less than5 mm=pentad.Peak: The pentad with highest amount of rainfall in theseason.End: If three consecutive pentads have mean rainfall of2 mm=day or less, the preceding pentad is considered to bethe end of rain season.Dry spell: 5-day rainfall less than 7.5 mm.

North coast long rain season (MAM) and south coastNovember to May seasonOnset: 5-day rainfall exceeds 10 mm followed by three con-secutive pentads having rainfall amount of not less than10 mm=pentad.Peak: The pentad with highest amount of rainfall in theseason.End: If three consecutive pentads have mean rainfall�2 mm=day the preceding pentad is considered to be theend of rain season.Dry spell: 5-day rainfall less than 10 mm.

Note: The pentad preceding the onset pentad is considered asthe pre-onset

Rainfall variability of coastal Tanzania and ENSO 155

These criteria have been found by the TanzanianMeteorological Agency to be appropriate for theregion. These dates were used for analyses of theregional circulation associated with the pentadscorresponding to pre-onset (one pentad prior toonset, P� 1), onset (P0), peak (P1) and end(Pþ 1) of the rainfall season.

Longitude-time diagrams or Hovm€ooller plotsof OLR and zonal wind at 850 and 200 hPa areused to identifying zonally propagating ISO.These plots were constructed for various ENSOyears for both the short rain (OND) and the longrain season (MAM) starting from one monthbefore the season and ending one month afterthe season. To isolate the ISO, the data used inthese Hovm€ooller analyses were first subjected toa 30–50 day band pass filter using the Dolph –Chebyshev convergence window (Doblas-Reyesand D�eequ�ee, 1998).

3. Seasonal anomalies

In general, Fig. 1a–b indicates that El Ni~nno yearsare associated with above normal rainfall and LaNi~nna years are associated with below normalrainfall for the north coast. The OND signalsappear to be more robust and coherent than thoseduring the MAM wet season. The years 1972,1982, 1986, 1994 and 1997 are identified as ElNi~nno years with more than one standard devia-tion departure from the mean for the north coastwhile 1970, 1973, 1975, 1988 and 1998 are thecorresponding La Ni~nna years. OND 1997 standsout as the wettest season and coincides also witha positive Indian Ocean Zonal Mode event. Forthe MAM season, most mature phase El Ni~nno(La Ni~nna) years correspond to above (below)average north coast rainfall.

For the south coast, the OND season seems toshow similar ENSO impacts to the north coast,except for 1990 and 1994, and inspection alsosuggests a quasi-decadal signal in the anomalies(Fig. 1d). A less coherent signal is evident for theJFM season for the south coast (Fig. 1c) with1973, 1983 and 1995 representing mature phaseEl Ni~nno years with above average rainfall and1987, 1992 and 1998 years with below averageconditions. A similar mixed response is evidentfor the mature phase La Ni~nna years. This mixedresponse may arise as the south coast appearsto be a transition region between the different

ENSO signals in equatorial East and tropical–subtropical southeastern Africa. Given the mixedsignals over the south coast, focus here is placedon analysis of the ISO for the north coast.

4. Hovmoller analyses

Longitude time plots of 30–50 day filtered OLR,lower and upper level zonal wind anomalies werederived for the El Ni~nno years of 1982, 1986, 1994and 1997 (1972 is not included since OLR dataare not available) and reveal alternating positiveand negative features for each year. Figure 2ashows examples for the 1982=83 and the1986=87 seasons (the others are not shown forbrevity). This period of filtering was found to beoptimal (20–40, 20–60 and 40–60 days were alsotried) and roughly spans the 40–50 and near30 day peaks in tropical convection and upperair parameters derived by Hayashi and Golder(1993). Figure 2a often indicates eastwardspropagating anomalies between September andDecember over the central Indian Ocean (70�–100� E) while westward propagating anomalieswere sometimes observed between the Tanzaniancoast and 70� E. These propagating features maybe convectively coupled equatorial Kelvin, n¼ 1equatorial Rossby or mixed Rossby-gravity waves(Madden and Julian, 1994; Wheeler and Kiladis,1999; Wheeler and Weickmann, 2001). Stationaryfeatures dominated 1997, which was the wettestyear during the period of study and also occurredin 1982, particularly in the central Indian Ocean.This result suggests that propagating featuresoccur less often during particularly strong ElNi~nnos and wet OND seasons in Tanzania.

It is seen that the largest OLR negative anom-aly values (below �20 W=m2) occur between 70and 120� E, indicating that more active convec-tion occurs towards the maritime continent overthe central and eastern tropical Indian Ocean asfound by other researchers (e.g. Rui and Wang,1990). Westward propagating negative OLRanomalies over the western Indian Ocean propa-gate with a mean speed of about 2 m=s and areassociated with strong convection over theTanzanian coast and increased rainfall during ElNi~nno years. Eastward propagating features werefound to have different phase speeds in differentyears and over different areas. They generallypropagate between 2–6 m=s with mean speed of


about 3 m=s or slightly slower than typical east-ward propagation in the Indian Ocean region(Madden and Julian, 1994).

Areas of large negative OLR anomalies(strong convection) couple well with areas ofstrong westerly anomalies at 850 hPa and strongeasterly anomalies at 200 hPa. Such an associa-tion has been found in previous studies of tropi-

cal intraseasonal oscillations (e.g. Rui and Wang,1990; Knutson and Weickmann, 1987). The com-bination of strong convection over the IndianOcean with westerly wind anomalies at lowerlevels coupled with easterly wind anomalies inthe upper levels implies a well-established IndianOcean Walker cell during the OND season, con-sistent with the findings of Hastenrath (2000).

Fig. 2. Hovm€ooller plots of band-pass filtered OLR, zonal wind at850 hPa and 200 hPa for 1982=83and 1986=87 El Ni~nno yearsduring a) OND and b) MAMseason. Contour intervals are5 Wm�2, 0.5 m=s and 1 m=s re-spectively. Shaded areas repre-sent negative anomalies


Longitude time plots of 30–50 days filteredOLR, 850 and 200 hPa zonal wind anomaliesfor mature phase El Ni~nno years 1983, 1987,1995 and 1998 are shown in Fig. 2b. Alternatingpositive and negative OLR anomalies are acommon feature of most years except in 1983between May and June when disorganised pat-terns were observed. Eastward propagating nega-

tive and positive OLR anomalies were observedin 1994 and 1997 with a mean phase speed of2 m=s. Westward propagating features, whichseem to originate from near 80� E, were mainlyobserved in 1983 and 1987.

Stationary features tend to dominate the850 hPa wind over the central Indian Ocean be-tween March and May with westward propagat-

Fig. 2 (continued)


ing features over the western Indian Ocean (lessobvious in 1983). At 200 hPa, the dominantaspect was eastward propagating positive andnegative wind anomalies except in 1987 wherewell-defined westward propagating features wereobserved between February and April. As for theOND season, the areas of large negative OLRanomalies (strong convection) couple well withareas of strong westerly anomalies in lowerlevels with strong easterly anomalies in upperlevels, which leads to increased rainfall overthe Tanzanian coast.

Filtered OLR and wind anomalies for La Ni~nnaand mature phase La Ni~nna years were also con-structed (not shown). As for the El Ni~nno seasons,alternating positive and negative OLR anomalieswere observed with both propagating and station-ary features occurring. Eastward propagating fea-tures were observed to be the dominant pattern inboth 850 and 200 hPa zonal winds except for1999 when westward propagating anomalieswere observed between March and June.

Two differences in the oscillations are appar-ent between El Ni~nno and La Ni~nna years. First, theeastward propagating OLR anomalies seem tohave high phase speed during La Ni~nna years, agood example is 1999 (7 m=s). Secondly, east-ward propagating OLR anomalies dominatedthe western Indian Ocean during La Ni~nna years,a good example is 1988=89 while for the El Ni~nnoyears the anomalies maintain westward propaga-tion (e.g. 1982=83, 1986=87). The presence ofeastward propagating OLR anomalies over thewestern Indian Ocean during La Ni~nna years leadsto suppressed convection and drier conditionsover the Tanzanian coast since it implies mois-ture convergence and unstable conditions signifi-cantly offshore rather than over the land.

5. Intraseasonal rainfallvariability and criteriafor rainfall onset, peak and end

The CMAP pentad rainfall data are averagedbetween 38.5�–41� E, 4�–8� S for the north coastand between 38.5� –41� E, 8�–11� S for the southcoast. CMAP data are used for this purpose asonly monthly data are available for the stations;however, this restricts the analysis to 1979 on-wards. Figure 3a represents the climatologicaltime series of pentad rainfall over the Tanzanian

north coast while El Ni~nno and La Ni~nna ONDcomposites and the following El Ni~nnoþ 1 andLa Ni~nnaþ 1 MAM composites for the northcoast are shown in Fig. 3b. Rainfall time seriesfor each individual El Ni~nno and La Ni~nna yearduring OND and MAM seasons and precedingmonths were also derived and plotted but arenot shown for brevity. The rainfall onset, peakand end (withdrawal) dates (Tables 2–4) wereextracted from each of these time series usingthe criteria listed in Table 1.

Table 2 indicates that for the north coast ONDseason, the rainfall starts almost a month earlierthan climatology during the El Ni~nno years, theseseasons reach their peak pentad about the sametime as the mean, and this peak receives a largeamount of rainfall (7 mm=day) (Fig. 3b). Onlyone significantly dry pentad occurs during ElNi~nno OND seasons. By contrast, the La Ni~nnayears start the rainy season two pentads later thanthe mean, reach their peak pentad over a monthlater than average with reduced values (around4 mm=day) and, as a result of a shorter seasonand reduced intensity, receive below averagerainfall. In addition, the La Ni~nna OND seasonsare characterised by three dry spells as comparedto only one on average for climatology or ElNi~nno seasons. Note that there is no differencein the end date of the OND season for any ofthe three categories in Table 2; thus, the majorENSO influence seems to be on the onset dateand the intensity of the OND rains.

During the north coast MAM season (Table 3),the onset for the El Ni~nnoþ 1 composite was threepentads late but the amount received was higherthan average with maxima of about 9 mm=day(Fig. 3b) as opposed to 5–6 mm=day in the cli-matological plot, and this peak was one pentadlater than average. For the La Ni~nnaþ 1 years,onset occurs at about the same time as the clima-tological mean but the amount received wasmore variable (maxima 7–8 mm=day) and thepeak pentad is substantially later than average.Neither the El Ni~nnoþ 1 nor the La Ni~nnaþ 1 sea-sons experienced any significant dry pentads andboth finished one pentad earlier than climatology.

Plots of individual ENSO OND and MAMseasons show that each event has its own char-acteristics in terms of onset, peak and end of therainy season and dry spells as listed in Table 4.With the exception of 1986, all El Ni~nno OND


seasons are found to have an early onset. Thisdoes not always translate into a longer rainy sea-son since some OND seasons end earlier than

average, and in fact, 1994 had a season thatlasted one pentad shorter than climatology and1986 was the same duration. However, in 1997,

Table 2. Rainfall onset, peak, end and major dry spell dates as derived from CMAP data for OND for the north coast

Events Climatology El Ni~nno composite La Ni~nna composite

Pre-onset(P� 1)

58th Pentad(13–17 October)

54th Pentad(23–27 September)


Onset(P0)


55th Pentad(28 Sept.–02 October)

61st Pentad(28 October–01 Nov.)

Peak(P1)

65th Pentad(17–21 November)

65th Pentad(17–21 November)

72nd Pentad(22–26 December)

End(Pþ 1)

4th Pentad(16–20 January)



Major dryPentads


63rd Pentad(07–11 November)

63rd, 68th–69th Pentad(07–11 Nov., 02–11 Dec.)

Fig. 3. a) Climatological pen-tad CMAP rainfall time seriesfor north coast of Tanzania, b)north coast pentad rainfall timeseries for El Ni~nno=La Ni~nnaand El Ni~nnoþ 1=La Ni~nnaþ 1composites


which was the wettest year during the period ofstudy, the rainfall started about three weeksbefore and the withdrawal was one month laterthan climatology. This season stands out as un-usually long (by comparison, 1982 was three pen-tads longer than average). Of these four seasons,all except 1986 were also an Indian Ocean ZonalMode year so it does not appear to be the com-bination of this mode with a strong El Ni~nno thatis responsible for 1997 being so unusual in thisrespect. OND 1997 and 1986 were also unusualin that wet spells were more closely spaced (1–2pentads apart) than for 1982 and 1994 whichshowed generally longer spacing, with one dryspell of seven and five pentads respectively.

In terms of OND totals after onset, 1982(310 mm) represents the wettest OND El Ni~nnoseason after 1997 (480 mm), and 1994 (210 mm)was the least wet.

Turning to the La Ni~nna OND seasons, thesehave a shorter rainfall season than averagewhich results from a late onset and a generallyearly withdrawal (except 1988 which ended onepentad later). Major wet spells also seem to bemore widely spaced in time than for the El Ni~nnoOND seasons. An extreme example is 1998,when the length of the rainfall season was onlyabout two weeks. None of the four La Ni~nna sea-sons correspond to a negative Indian OceanZonal Mode year. In order of increasing dryness

Table 3. As for Table 2 but MAM season for the north coast (MAM season)

Events Climatology El Ni~nno composite La Ni~nna composite

Pre-onset (P� 1) 13th Pentad (02–06 March) 16th Pentad (17–21 March) 13th Pentad (02–06 March)Onset (P0) 14th Pentad (07–11 March) 17th Pentad (22–26 March) 14th Pentad (07–11 March)Peak (P1) 24th Pentad (26–30 April) 25th Pentad (01–05 May) 27th Pentad (11–15 May)End (Pþ 1) 31st Pentad (31 May–4 June) 30th Pentad (26–30 May) 30th Pentad (26–30 May)Major dry Pentads 0 0 0

Table 4. As for Table 2 but for each ENSO (north coast OND season) and ENSOþ 1 (north coast MAM season) year. E and Lrepresent El Ni~nno and La Ni~nna year respectively with corresponding following year Eþ 1 and Lþ 1

Year Pre-onsetpentad (P� 1)

Onsetpentad (P0)

Peakpentad (P1)

End pentad(Pþ 1)

Major drypentads

1982 (E) 54th 55th 57th 3rd (1983) 61st–63rd, 70th,2nd (1983)

1986 (E) 58th 59th 67th, 69th

and 71st4th (1987) 61st, 63rd, 68th,

and 73rd

1994 (E) 57th 58th 69th 2nd (1995) 65th and 66th

1997 (E) 54th

23–27 Sept55th

28 Sep–02 Oct59th

18–22 Oct10th (1998)15–19 Feb

60th, 63rd, and5th (1998)

1983 (Eþ 1) 16th 17th 24th 33rd 22nd and 26th

1987 (Eþ 1) 19th 20th 26th 32nd 30th and 31st

1995 (Eþ 1) 19th 20th 30th 30th 26th

1998 (Eþ 1) 16th

17–21 March17th

22–26 March23rd

21–25 April26th

06–10 May0

1988 (L) 61st 62nd 73rd, 5th (1989) 5th (1989) 66th and 67th

1995 (L) 59th 60th 61st 2nd (1996) 63rd, 64th,68–70th and 73rd

1998 (L) SEE NOTE BELOW1999 (L) 62nd 63rd 64th 71st 68th and 69th

1989 (Lþ 1) 18th 19th 22nd 30th 01996 (Lþ 1) 13th 14th 28th 29th 26th

1999 (L) 12th 13th 24th 31st 28th

2000 (L) 15th 16th 24th 27th 22nd

Note: In 1998, the onset and end of rainfall were undefined as the whole season was dry except 65th, 66th and 67th pentads,which recorded rainfall


after onset, the seasons are 1988 (155 mm), 1995(140 mm), 1999 (105 mm) and 1998 (90 mm forthe whole season – the onset criterion was neversatisfied in this case). In general, the commonfeature for all La Ni~nna years is below averagerainfall during the season while El Ni~nno yearsshow above average rainfall during the northcoast OND season.

For the MAM season, all El Ni~nno years weredelayed in starting the rains but two finished lateand two early compared to climatology. Themajor wet spells are also more widely spacedthan for the OND season and all MAM seasonsexperienced at least one dry pentad. After onset,the wettest seasons were 1983 and 1998 and theleast wet 1995. By comparison, two of the LaNi~nna seasons started late and one early, and allbut 1999 ended rather earlier than average. Eachseason except 1989 had one dry pentad and themajor wet spells for this season were more close-ly spaced in time than for the other three cases.After onset, the wettest season was 1999 and thedriest 1989 and 2000. For the MAM season,ENSO rainfall impacts over the north coastappear more variable than those during OND.

The dates given in Table 4 are used to definethe periods for which NCEP re-analyses are com-posited in order to try and understand the circu-lation patterns associated with the onset, peakand withdrawal stages of the rainy season.

5.1 ENSO influenceon the short rains (OND)

Figure 4a–d show NCEP anomalies in OLR,850 hPa moisture flux, middle and upper levelwind for the El Ni~nno years and Fig. 5a–d the cor-responding La Ni~nna seasons. To qualify for com-positing, El Ni~nno and La Ni~nna years need to showa north coast station averaged rainfall anomaly ofat least 1 standard deviation during the ONDseason (Fig. 1b). The resulting El Ni~nno yearsfor the OND composite are 1972, 1982, 1986,1994 and 1997 with the corresponding La Ni~nnayears being 1970, 1973, 1975, 1988, 1998 and1999. The pentads used are the same as those

used above with the dates for the El Ni~nno andLa Ni~nna composites shown in Tables 2–4.

One pentad before the onset of the OND rainseason (P� 1) (Fig. 4a), much of the westernIndian Ocean is covered by positive OLR anoma-lies with the highest anomaly value of the orderof 40 W=m2 observed over the equatorial centralIndian Ocean. At 700 hPa, the easterly windanomalies over this region oppose the mean flow,implying convergence and subsidence downthrough the lower atmosphere. Most of thewestern Indian Ocean is characterized by an anti-cyclonic moisture flux anomaly that results innorthwesterly to northerly anomalies near theTanzanian coast. As a result, there is little pene-tration of moisture inland, which is consistentwith positive OLR anomalies over Kenya andnorthern Tanzania. At 200 hPa, easterly windanomalies exist over much of Tanzania, therebyopposing the background flow and implying con-vergence, and hence subsidence down throughthe atmosphere, suppressing convection.

By the onset of the OND rain season (P0)(Fig. 4b), a deep convective zone has developedover the equatorial western Indian Ocean, asindicated by negative OLR anomalies over theregion. Enhanced moisture flux convergence overthe Tanzanian coast, as indicated by northeast-erly anomalies over the Kenyan and northTanzanian coasts, which also imply less exportof moisture away from East Africa over the equa-torial western Indian Ocean, results in increasedrainfall over the north coast at the onset of therainfall season. In addition to the convective zonenear 55–65� E, a developing area of convection isnow apparent over coastal East Africa. Over theequatorial Indian Ocean, easterly anomalies areevident at both 850 and 700 hPa consistent witha weaker zonal pressure gradient across thebasin and enhanced rainfall on the African side(Hastenrath and Polzin, 2004). At 200 hPa,strong westerly wind anomalies are in the samedirection as climatology implying strong upperlevel divergence. This is consistent with the ini-tiating convective zone over coastal Tanzania,which suggests enhanced rainfall there. The

1

Fig. 4. El Ni~nno composite OND anomaly fields OLR in (W=m2), 850 hPa moisture flux in g=kg �m=s and 700 hPa and200 hPa wind in m=s for a) pre-onset (P� 1), b) onset (P0), c) peak (P1) and d) withdrawal (Pþ 1) pentads. Scale vectors areshown and shading represents negative OLR anomalies


Fig. 4 (continued)

164 A. L. Kijazi and C. J. C. Reason: Rainfall variability of coastal Tanzania and ENSO

Fig. 5. La Ni~nna composite OND anomaly fields OLR in (W=m2), 850 hPa moisture flux in g=kg �m=s and 700 hPa and200 hPa wind in m=s for a) pre-onset (P� 1), b) onset (P0), c) peak (P1) and d) withdrawal (Pþ 1) pentads. Scale vectors areshown and shading represents negative OLR anomalies

Fig. 5 (continued)


presence of an active convective zone over theequatorial western Indian Ocean with enhancedmoisture convergence results in above averagerainfall over the Tanzanian coast during El Ni~nnoyears.

At the peak of OND rain season (P1) (Fig. 4c),the convective zone over the western IndianOcean shows significant westward extent andhas merged with that initiated over coastal EastAfrica during the onset (Fig. 4b). Over coastalTanzania, negative OLR anomalies of about�15 to �20 W=m2 are apparent. The area be-tween 10� N and 5� S over the central IndianOcean shows low-level easterly moisture fluxanomalies and easterly anomalies at 700 hPaopposing the climatological export of moistureaway from East Africa by the equatorial westerlies,consistent with negative OLR anomalies overthe region and enhanced rainfall (Hastenrath andPolzin, 2004). Much of Tanzania and the nearbyocean indicates a westerly moisture flux anomaly,which converges with the easterly moisture fluxanomaly at about 55� E. Since the mean moistureflux over Tanzania during this time is easterly, thewesterly moisture flux anomaly tends to opposethe flow causing deceleration and enhanced mois-ture convergence over the coast. This convergenceleads to the peak of rain during this pentad ofabove average rainfall. Moisture convergenceover the Indian Ocean at about 55� E with stron-ger negative OLR anomalies over the regioncoupled with divergence at 200 hPa (not shown)indicates the position of the rising limb of IndianOcean Walker cell, which shows a significantwestward shift towards East Africa.

At the end of the OND rain season (Pþ 1)(Fig. 4d), the convective zone over the westernIndian Ocean is replaced by periods of clearweather as is evident from the positive OLRanomalies covering the tropical western IndianOcean, the Tanzanian north coast and Kenya.The moisture flux anomaly over much of theequatorial western Indian Ocean indicates a rela-tive divergent pattern as it flows in the samesense as the climatological flow, leading to moremoisture advected towards 50� E and increasedexport from this region back over the oceannorth and east of Madagascar. Relative conver-gence occurs over Zambia, southern Tanzaniaand northern Mozambique as the flow anomaliesare in the opposite direction to the mean flow.

This period marks the shift towards the summerrainy season over southern Africa as indicatedby negative OLR anomalies over this part ofthe landmass. Subsiding conditions and reducedclouds occur over the Tanzanian coast withdivergence in the lower levels as suggested bythe moisture flux anomalies and convergence inthe upper levels. The latter is implied by thewind flow pattern at 200 hPa that opposesthe mean flow. These circulation patterns markthe end of the rainy season over equatorial EastAfrica.

The corresponding anomalies for the variouspentad stages of the OND La Ni~nna compositeare given in Fig. 5a–d. One pentad before theonset of OND season (P� 1) (Fig. 5a), the LaNi~nna composite is more or less the reverse ofthe El Ni~nno composite with negative OLR anom-alies over the equatorial central Indian Oceancoupled with enhanced westerly moisture fluxwhich feeds the convection there. The presenceof offshore (westerly) moisture flux anomaliesover the Tanzanian coast suggests suppressedmoisture advection one pentad prior to the onsetof rainfall season. At 200 hPa, the winds repre-sent a strengthening of the mean, and hencedivergence, near the negative OLR anomaly, pro-moting convection and suggesting that the posi-tion of the rising limb of Walker cell is locatedover the equatorial central=eastern Indian Oceanconsistent with negative OLR anomalies appar-ent over this part of the ocean. As discussed inHastenrath (2000), an enhanced Indian OceanWalker cell in the OND season tends to occurduring La Ni~nna.

During the onset of the OND rain season (P0)(Fig. 5b), the negative OLR anomalies overthe central Indian Ocean are strengthened, andshifted further east, reaching approximately�40 W=m2 at 80� E. This location contrasts withthe El Ni~nno onset situation (Fig. 4b) where thedeveloping convection shifts west towards Africaand is located near 55–65� E. Most of Tanzaniashows positive OLR anomalies suggestingreduced cloud and convection compared to aver-age. At 850 hPa, westerly moisture flux anoma-lies exist over much of the Indian Ocean, whichfeed the convective zone centered at about 80� Eand increasing the climatological export ofmoisture away from East Africa. This offshoremoisture flux results in reduction of rainfall over


the Tanzanian coast during La Ni~nna years. Strongwesterly anomalies are particularly evident at850 hPa in the equatorial Indian Ocean consistentwith a strong zonal pressure gradient and reduced(enhanced) rains over East Africa (Indonesia)(Hastenrath and Polzin, 2004). The ascendinglimb of the Walker cell over the central IndianOcean strengthens as indicated by enhanced wes-terly moisture flux at 850 hPa with enhancedwesterly winds at 200 hPa and hence divergence(not shown).

At the peak of the OND rain season (P1)(Fig. 5c), positive OLR anomalies are observedover the Tanzanian coast with highest values of15–20 W=m2 indicating suppressed convectionover the coast. Negative anomalies are evidentover countries further south, consistent with theenhanced rainfall generally experienced overmuch of southern Africa during La Ni~nna. Themoisture flux anomaly plot is roughly the reverseof the El Ni~nno composite with ongoing westerlymoisture flux anomalies over the western equa-torial Indian Ocean and easterly anomalies northof Madagascar and near the Tanzanian coast.The northeasterly moisture flux anomaly overthe coast represents acceleration relative to cli-matology resulting in moisture flux divergenceand reduction of rainfall, which is consistentwith positive OLR anomalies over the region.At 700 hPa, there are strong westerly anomaliesover the Tanzanian coast enhancing the meanflow, implying divergence. Divergence at thesemiddle levels (not shown) weakens the verticalextent of the convection and results in reducedrainfall.

At the end of the OND rain season (Pþ 1)(Fig. 5d), negative OLR anomalies are apparentover most of the tropical Indian Ocean and south-eastern Africa indicating a strengthened Walkercell over the Indian Ocean and reduced uplift overKenya and northern Tanzania. In addition, strongnegative OLR anomalies are evident over theMozambique region consistent with the develop-ing wet conditions over southeastern Africa asso-ciated with La Ni~nna. Weak northeasterly tonortherly moisture flux anomalies exist near the

Tanzanian coast with relative divergence, markingthe end of the rainfall season here.

5.2 ENSO influenceon the long rains (MAM)

One pentad before the onset of the MAM rainyseason, (P� 1) (Fig. 6a), a convective zone is ap-parent over the equatorial western Indian Ocean(negative OLR anomalies). The other area of sig-nificant negative OLR anomaly east of Madagascaris too far south to be associated with the ITCZ.Westerly moisture flux anomalies over the equa-torial North Indian Ocean feed the equatorialconvective zone. The enhanced equatorialwesterlies at 700 hPa relative to climatology sug-gest that the mid-level moisture flux also contrib-utes to this developing convection. The upperlevel wind patterns show convergence (not shown)over much of the Tanzanian coast suggestingsubsidence, and hence reduced cloud one pentadbefore the onset of the rain season.

During onset (P0) (Fig. 6b), the negative OLRanomalies over the equatorial Indian Ocean ex-tend westwards towards the Tanzanian coastconsistent with the southeasterly moisture fluxanomaly just off the coast that opposes the meanflow. The ascending convective limb is over theequatorial western Indian Ocean as indicated bynegative OLR anomalies and moisture conver-gence over the region coupled with upper levelwind divergence (not shown). The enhancedmoisture convergence over the equatorial westernIndian Ocean results in increased rainfall overthe coast during MAM season of the maturephase El Ni~nno years.

At the peak of the season (P1) (Fig. 6c), nega-tive OLR anomalies over the western IndianOcean deepen reaching a minimum anomalyvalue of �25 W=m2 centered at 50–55� E andshow the westward extent of convection, whichcovers the entire north coast of Tanzania. Acyclonic moisture flux anomaly exists over theequatorial western Indian Ocean with westerlymoisture flux anomaly across southern Tanzania.This leads to enhanced moisture convergence

1

Fig. 6. El Ni~nnoþ 1 composite MAM anomaly fields OLR in (W=m2), 850 hPa moisture flux in g=kg �m=s and 700 hPa and200 hPa wind in m=s for a) pre-onset (P� 1), b) onset (P0), c) peak (P1) and d) withdrawal (Pþ 1) pentads. Scale vectors areshown and shading represents negative anomalies


Fig. 6 (continued)


Fig. 7. La Ni~nnaþ 1 composite OND anomaly fields OLR in (W=m2), 850 hPa moisture flux in g=kg �m=s and 700 hPa and200 hPa wind in m=s for a) pre-onset (P� 1), b) onset (P0), c) peak (P1) and d) withdrawal (Pþ 1) pentads. Scale vectors areshown and shading represents negative anomalies

Fig. 7 (continued)


over much of Tanzania (not shown) and in-creased rainfall marking the peak of rainfall sea-son over the Tanzanian coast. The 700 hPa windshows an anticyclonic anomaly centered overMadagascar with strong easterly anomalies overthe tropical western Indian Ocean and Tanzaniaopposing the background flow, hence reducingthe climatological export of moisture away fromEast Africa. This anticyclonic anomaly shows anequatorward shift with height and, at 200 hPa, itis located over the tropical western Indian Oceanthereby indicating the position of the ascendingconvective limb.

During withdrawal (Pþ 1) (Fig. 6d), the con-vective zone over the equatorial western IndianOcean shows an eastward shift leaving weak neg-ative anomalies over the Tanzanian north coast.The cyclonic moisture flux anomaly over theequatorial western Indian Ocean is less apparentand a south equatorial westerly moisture fluxanomaly, which feeds the maritime convectivezone, is evident. The descending limb of the con-vective cell is located over the equatorial westernIndian Ocean as indicated by positive OLRanomalies over this region with wind conver-gence at 200 hPa (not shown) implying reducedcloud cover. This marks the end of the rainfallseason over the Tanzanian coast.

The corresponding anomalies for the MAM LaNi~nnaþ 1 composite are now discussed. One pen-tad before the onset of MAM rain season (P� 1)(Fig. 7a), negative OLR anomalies cover easternTanzania and southeastern Africa and positiveOLR anomalies are located east and north ofMadagascar. An anticyclonic moisture flux anom-aly is evident over the Indian Ocean eastof Madagascar. The northerly to northwesterlymoisture flux anomalies over the Tanzanian coastare in the same direction as the mean flow imply-ing moisture flux divergence. The convectivezone over the Tanzanian coast is short-lived sincethere is upper level wind convergence (notshown), which implies descending motion overthe region.

At onset (P0) (Fig. 7b), a dipole pattern inOLR anomalies exists over the Indian Oceanwith positive anomalies south of 10� S and nega-tive anomalies to the north. Most of Tanzaniashows weak negative OLR anomalies. The anticy-clonic moisture flux anomaly east of Madagascarshows a slight westward shift, which allows

easterly moisture flux anomalies to dominatemuch of the western Indian Ocean between theequator and about 25� S. As a result, the mois-ture flux anomaly over the Tanzanian coast is thesame sense as the mean flow and thereforedivergent, leading to reduction of rainfall. At700 hPa, there are strong southwesterly anoma-lies over tropical Africa and the western IndianOcean that feed into the equatorial convectivezone offshore, but lead to mid-level convergence(not shown) over Tanzania, subsidence, and lessrain.

During the peak of the season (P1) (Fig. 7c),negative OLR anomalies are observed over theTanzanian coast consistent with low level mois-ture flux convergence over the region (notshown). However, the development of deep con-vective clouds is unlikely since the upper levelwind anomalies are relatively strong easterlies,opposing the mean flow, and leading to conver-gence at 200 hPa (not shown) and hence subsi-dence. Below average rainfall over the Tanzaniancoast is evident during this period.

Towards the withdrawal of the season (Pþ 1)(Fig. 7d), a convective zone develops over thenorthwest Indian Ocean with relatively strongwesterly moisture flux anomalies at 850 hPafeeding the convection. These suggest that thedeveloping monsoon over the northwest IndianOcean may be initially stronger than averageand the ITCZ has moved northwards morequickly than typical. The descending limb ofthe convective cell is located over East Africa,with reduced cloud there as indicated by positiveOLR anomalies near the coast. This marks theend of rainfall season over the northern coast ofTanzania.

6. Summary and conclusion

This study has outlined some of the importantdynamical patterns underlying the intraseasonalvariability of rainfall over the Tanzanian coast.Focus has been placed on the north part (4–8� S)of the coast which shows a coherent ENSOimpact; the signals over the south coast are lessclear and appear to be a transition betweenthe opposite responses over East and southernAfrica. The climatological onset, peak and rain-fall withdrawal plots confirm the association ofTanzanian coast rainfall with the migration of


the ITCZ. The peak of the rainfall season isfound to occur while the ITCZ is over theregion.

Hovm€ooller and pentad analyses revealed acoherent temporal and spatial relationship be-tween the enhanced and suppressed convectionassociated with intraseasonal oscillations duringENSO years. The rainfall time series for thenorth coast revealed that increased rainfall dur-ing El Ni~nno OND seasons was associated withan early onset, less dry spells and more intenserainfall whereas reduced rainfall during La Ni~nnayears was associated with a late start of the sea-son, more dry spells and less intense rains. ForMAM, the El Ni~nno seasons tended to start lateas did most of the La Ni~nna seasons. Rainfallimpacts were less clear in MAM than forOND. A possible explanation for this may bebecause the ENSO-induced oceanic Rossbywave that propagates the signal across the IndianOcean leads to a negative feedback between thiswave, the overlying winds and SST in the post-mature phase MAM season whereas for thepreceding OND season, the feedback with theatmosphere is positive and the SST signal ismore robust (Xie et al., 2002). Thus, Fig. 6c,which corresponds to the peak pentad of theMAM rainy season shows a low level cyclonicanomaly north of Madagascar implying a damp-ing of the El Ni~nno-induced SST warming viaEkman divergence and upwelling. Conversely,the low level anticyclonic anomaly south ofIndia during the peak phase of the OND rainyseason (Fig. 4c) implies Ekman convergence andhence amplification of the developing warm SSTanomaly during this phase of ENSO. This anti-cyclonic anomaly extends towards East Africaand is co-located with increased rainfall overthe western Indian Ocean. As a result, ENSOimpacts over Tanzania are more likely to dieout quickly in MAM, or not be coherent if thetiming of the event is shifted relative to theannual cycle, leading to a less obvious relation-ship between north coast rainfall and ENSO dur-ing this season.

Increased rainfall over the north coast duringEl Ni~nno years was observed to be associated withenhanced convection and moisture flux conver-gence over the Tanzanian coast and reducedexport of moisture away from East Africa overthe equatorial western Indian Ocean. La Ni~nna

years tend to show the reverse. Consistent withHastenrath and Polzin (2004), there are weak-ened (strengthened) equatorial westerlies overthe central Indian Ocean during El Ni~nno(La Ni~nna) with signs of weaker (stronger) south-easterly trades. The wet conditions of the ONDseason of the El Ni~nno onset years are extended tothe usually dry months of January and Februaryof the mature phase El Ni~nno year.

The OLR and zonal wind Hovm€ooller plotsrevealed eastward and westward propagating aswell as stationary features over the Indian Ocean.It was observed that the propagating featurestended to be absent during strong El Ni~nno yearswhereas for strong La Ni~nna years, they are oftenpresent. It is also found that the areas of largenegative OLR anomalies (strong convection)tend to couple well with areas of strong westerlyanomalies in lower levels with strong easterlyanomalies in the upper levels. Westward propa-gating negative OLR anomalies are found overthe far western Indian Ocean during El Ni~nnoOND seasons and are associated with strong con-vection over the Tanzanian coast and increasedrain while the reverse is observed during LaNi~nna. Further east over the Indian Ocean, east-ward propagating features are evident during ElNi~nno whereas the signals are less clear for LaNi~nna. For the MAM season, some years showwestward propagating features over the centraland western Indian Ocean (e.g. 1983, 1987)whereas others show eastward propagation (e.g.1995, 1998).

The onset and peak of rainfall during El Ni~nnoyears are found to be associated with an activeconvective zone over the equatorial westernIndian Ocean with enhanced moisture conver-gence, which results in above average rainfallwhile the reverse is true for La Ni~nna years. Inconclusion, ENSO has a significant influenceon intraseasonal rainfall over the north Tanzaniancoast, particularly for OND. For the south coast,the signals are similar to the north coast and therest of East Africa for OND, but for the other halfof the rainy season (JFM), mixed ENSO signalsare evident.

Acknowledgements

This study was done when the first author was at theUniversity of Cape Town on a Third World Organization


for Women in Science (TWOWS) fellowship. Dr. D.Jagadheesha offered useful support on computing andplotting during the preparation of a MSc thesis by the firstauthor, from which this paper derives. The NCEP=NCARdata obtained from their website http:==www.cdc.noaa.gov are gratefully acknowledged. Rainfall data werekindly supplied by the Tanzania Meteorological Agency.We thank the reviewers for helpful comments whichimproved the paper.

References

Alusa AL, Gwange PM (1978) The occurrence of dry spellsduring the East African long rains. Research report No.5=78, E.A Institute for Meteorology, pp 1–18

Amissah-Arthur A, Jagtap S, Rosenzweig C (2002) Spatio-temporal effects of El Ni~nno events on rainfall and maizeyield in Kenya. Int J Climatol 22: 1849–1860

Anyamba EK (1992) Some properties of a 20–30 dayoscillation in tropical convection. J Afr Met Soc 1: 1–19

Cadet DL (1985) The Southern Oscillation over the IndianOcean. J Climatol 5: 189–212

Camberlin P (1995) June–September rainfall in north-east-ern Africa and atmospheric signals over the tropics: azonal perspective. Int J Climatol 15: 773–783

Doblas-Reyes FJ, D�eequ�ee M (1998) A flexible bandpass filterdesign procedure applied to midlatitude intraseasonalvariability. Mon Wea Rev 126: 3326–3335

Hastenrath S, Nicklis A, Greischar L (1993) Atmospherichydrospheric mechanisms of climate anomalies in thewestern equatorial Indian Ocean. J Geophys Res98(C 11): 20219–20235

Hastenrath S (2000) Zonal circulations over the equatorialIndian Ocean. J Climate 13: 2746–2756

Hastenrath S, Polzin D (2004) Dynamics of the surface windfield over the equatorial Indian Ocean. Quart J RoyMeteor Soc 130: 503–517

Hayashi Y, Golder DG (1993) Tropical 40–50 and 25–30day oscillations appearing in realistic and idealized GFDLclimate models and the ECMWF dataset. J Atmos Sci 50:464–494

Hutchinson P (1992) The Southern Oscillation and predic-tion of ‘‘Der’’ season rainfall in Somalia. J Climate 5:525–531

Indeje M, Semazzi FHM, Ogallo LJ (2000) ENSO signalsin East African rainfall seasons. Int J Climatol 20:19–46

Janowiak (1988) An investigation of Inter annual rainfall.J Climate 1: 240–255

Kabanda TA, Jury MR (2000) Synoptic evolution of com-posite wet spells over northern Tanzania. Climate Res 15:239–248

Kalnay E, and Co-authors (1996) The NCEP=NCAR40-year Reanalysis Project. Bull Amer Meteor Soc 77:437–471

Knutson TR, Weickmann KM (1987) 30–60 day atmo-spheric circulations: Composite life circles of convec-tion and circulation anomalies. Mon Wea Rev 115:1407–1436

Levey KM, Jury M (1996) Composite intra-seasonal oscilla-tions of convection over southern Africa. J Climate 9:1910–1920

Madden RA, Julian PR (1971) Detection of a 40–50 dayoscillation in the zonal wind in the tropical Pacific.J Atmos Sci 28: 702–708

Madden RA, Julian PR (1994) Observations of the 40–50day tropical oscillation: a review. Mon Wea Rev 122:814–837

Matarira CH, Jury MR (1992) Contrasting Meteorologicalstructure of Intra seasonal wet and dry spells inZimbabwe. Int J Climatol 12: 165–176

Mhita MS (1984) The use of water balance models in theoptimation of cereal yields in seasonally-arid tropicalregions. PhD Thesis. University of Reading, p 211

Mhita MS, Nassib IR (1987) The onset and end of rain inTanzania. Proc. 1st Tech. Conference. Meteor Res Easternand Southern Africa, Nairobi, pp 101–115

Mpeta EJ, Jury MR (2001) Intra-seasonal convective struc-ture and evolution over tropical East Africa. Climate Res17: 83–92

Mutai CC, Ward MN (2000) East African rainfall and thetropical circulation=convection on intraseasonal to inter-annual timescales. J Climate 13: 3915–3939

Nicholson SE (1996) A review of climate dynamics andclimate variability in Eastern Africa. In: Johnson TC,Odata E (eds) Limnology, climatology and paleoclima-tology of the East African lakes. The Netherlands: Gordonand Breach Publishers

Nicholson SE, Kim J (1997) The relationship of the El Ni~nno-Southern Oscillation to African Rainfall. Int J Climatol17: 117–135

Nyenzi BS (1988) Mechanisms of East African Rainfallvariability. PhD. Thesis Florida State University, 184 pp

Ogallo LJ (1988) Relationship between seasonal rainfall inEast Africa and Southern Oscillation. J Climatol 8:34–43

Ogallo LJ (1989) The spatial and temporal patterns of theEast African seasonal rainfall derived from principalcomponent analysis. Int J Climatol 9: 145–167

Reason CJC, Allan RJ, Lindesay JA, Ansell TA (2000)ENSO and climatic signals across the Indian Ocean basinin the global context. Int J Climatol 20: 1285–1327

Ropelewski CF, Halpert MS (1987) Global and regionalscale precipitation patterns associated with the ElNi~nno-Southern Oscillation. Mon Wea Rev 115:1606–1626

Rui H, Wang B (1990) Development characteristics anddynamic structure of tropical intraseasonal convectionanomalies. J Atmos Sci 47: 357–379

Saji NH, Goswami BN, Vinayachandran PN, Yamagata T(1999) A dipole mode in the tropical Indian Ocean. Nature401: 360–363

Tourre YM, White WB (1997) Evolution of ENSO signalover the Indo-Pacific domain. J Phys Oceanogr 27:683–696

Vincent DG, Sperling T, Fink A, Zube S, Speth P (1991)Intraseasonal oscillation of convective activity in thetropical Southern Hemisphere: May, 1984-April, 1986.J Climatol 4: 40–53


Webster PJ, Moore AM, Loschnigg JP, Lebden RR (1999)Coupled ocean–atmosphere dynamics in the Indian Oceanduring 1997–98. Nature 401: 356–360

Wheeler MA, Kiladis GN (1999) Convectively coupled equa-torial waves: analysis of clouds and temperature in thewavenumber-frequency domain. J Atmos Sci 56: 374–399

Wheeler MA, Weickmann KM (2001) Real-time monitoringand prediction of modes of coherent synoptic to intrasea-sonal tropical variability. Mon Wea Rev 129: 2677–2694

Xie P, Arkin PA (1997) Global precipitation: A 17-yearmonthly analysis based on gauge observations, satelliteestimates and numerical model outputs. Bull AmerMeteor Soc 78: 2539–2558

Xie S-P, Annamalai H, Schott FA, McCreary JP (2002)Structure and mechanism of South Indian Ocean climatevariability. J Climate 15: 864–878

Zhu B, Wang B (1993) The 30–60 day convective seesawbetween the tropical Indian and western Pacific Oceans.J Atmos Sci 50: 184–199

Authors’ addresses: Agnes L. Kijazi�, C. J. C. Reason(e-mail: [email protected]), Department of Oceanography,University of Cape Town, Rondebosch 7701, South Africa.�Also at: Tanzania Meteorological Agency, P.O. Box 3056Dar es Salaam, Tanzania.


Relationships between intraseasonal rainfall variability...

Documents

Transcript of Relationships between intraseasonal rainfall variability...