The clime of Southern Africa strongly depends on seasonal fluctuation of synoptic graduated table characteristics over the part. Examples of these characteristics are cut-off depressions, deep tropical depressions, deep temperate depressions and Tropical Temperate Troughs ( TTTs ) . The characteristics have a clip graduated table of about 3-5 yearss, a average horizontal graduated table of about 1000-3000 kilometers, and normally propagate due west with a velocity of about 5-8 m s-1 ( Orlanski, 1975 ) over Southern Africa. In most instances, they are responsible for the utmost conditions conditions over the part. Although they are short lived, their complex interactions with the big graduated table circulation characteristics ( with a longer clip graduated table ) determine the clime of Southern African clime ( Hudson and Jones, 2002 ; Reason and Jagadheesha, 2005 ; Anyah and Semazzi, 2006 ) . It is hence indispensable that a planetary theoretical account for seasonal prognosis over the part gives a realistic simulation of the hereafters, at least their seasonal variableness. This survey investigates how good the characteristics are simulated in the Hadley Centre Atmospheric Model version 3 ( HadAM3 ) and the NCAR the Community Atmospheric theoretical account version 3 ( CAM3 ) .
These prevailing synoptic graduated table characteristics are good recognized at 500hPa tallness ( Tyson, 1981 ; Tyson and Preston-Whyte, 2000 ) . Therefore, it is of import to analyze the anomalousnesss in the geopotential height field at 500hPa and 850hPa for 30 old ages. Anomalies in this field aid in explicating how seasons vary from the one-year mean.
The frequence, continuance and strength of the synoptic characteristics either induce or suppress rainfall, therefore taking to rainfall variableness over Southern Africa ( Harrison 1984 ; Tyson and Preston-Whyte, 2000 ) . For illustration, Taljaard ( 1985 ) showed that the western perturbations and the formation of cut-off depressions induce rainfall. Cut-off depressions are low force per unit area systems that develop from the westerlies ( Hobbs et al. 1998 ; Harrison 1984a ; 1984b ; Todd and Washington, 1999 ; Tyson and Preston-Whyte 2000 ; Jury and Nkosi 2000 ) . They are seen in the upper troposphere as troughs, deepens until they form closed circulations ( Hobbs et al. , 1998 ; Tyson and Preston-Whyte, 2000 ; Browning and Mason, 1980 ; Fuenzalida et al. , 2005 ; Smith and Reeder, 1988 ; Garreaud, 2000 ) . These deep closed circulations that develop from upper western troughs are really intense synoptic characteristics over Southern Africa. They induce unstable troposphere at low degrees, produce terrible convective events that lead to heavy rainfall and inundations over big countries, and may trip terrible cyclonegenesis that induces strong air current ( Nieto et al. , 2005 ; Tyson and Preston-Whyte, 2000 ) .
Deep tropical depressions and deep temperate depressions are low degree circulation Centres in the Rossby moving ridges. Tropical depressions are caused by perturbations in the eastern moving ridge, driven largely by thermic warming of the continent. The deep tropical and temperate low can escalate into a cyclone and if more intense, can take to a storm. They help in tropical wet and energy transportation. This procedure significantly facilitates rainfall over the part ( Mason and Jury, 1997 ; Van den Heever et al. , 1997 ) . Besides, Miron and Tyson ( 1984 ) found that synoptic state of affairss responsible for rain-bearing air currents are consistent with easterly low perturbations. The tropical and temperate depressions are responsible for rainfall over the inside during summer. Harrison ( 1984 ) reported in his survey that tropical and temperate disturbances are major determiners of Southern African rainfall.
When the upper westerly wave coincides with an eastern moving ridge or depression in lower degrees, it consequences in the formation of a tropical temperate trough. This characteristic is associated with tropical wet and energy transportation and it has been known to significantly lend to summer rainfall over Southern Africa ( Harangozo and Harrison, 1983 ; Harrison, 1984b ) .
The survey reports the capableness of HadAM3 and CAM3 in imitating the average figure of cut-off depressions, figure of yearss with tropical depressions, temperate depressions and TTTs from 1971 through 2000. It besides attempts to discourse how the synoptic graduated table characteristics are related to the rainfall variableness over Southern Africa. HadAM3 is the atmosphere constituent of the Hadley Centre Coupled Model version 3 ( Gordon et al. , 2000 ; Pope et al. , 2000 ) , which was developed at the Hadley Centre for Climate Prediction and Research, UK. The theoretical account employs spherical polar co-ordinates on a regular latitude-longitude grid. The horizontal declaration is 3.75o longitude, 2.5o latitude, and 8 beds in the vertical, which are based on a intercrossed perpendicular co-ordinate system ( Simmons and Burridge, 1981 ) . The development and description of the HadAM3 theoretical account can be found in Gordon et al. , 2000 ; Pope et al. , 2000 ; Jones et al. , 2005 ; Murphy et al. , 2002. The HadAM3 has been used successfully in some surveies over the Southern Africa. Examples can be seen in Tennant, 2003 ; Reason et al. , 2003 ; Reason and Jagadheesha, 2005. The theoretical account is by and large able to capture the circulation kineticss of the Southern African clime. CAM3 is the atmosphere constituent of the Community Climate System Model, version 3.0 ( Collins et al. , 2004 ) , which was developed at the US National Centre for Atmospheric Research ( NCAR ) . The survey used the finite volume dynamic nucleus option of the CAM3, with horizontal declaration of 2.0o ten 2.5o and 26 perpendicular degrees. This theoretical account has non been used extensively to analyze the clime of Southern Africa. The survey hence reports the theoretical account ‘s capableness to concerned establishments in Southern Africa, which will make up one’s mind on the possibility of the theoretical account to be added in seasonal prediction over the part.
Few surveies over Southern Africa show that the rainfall seasonality varies greatly from one part to another which would explicate why the seasonal rainfall over all countries of the part is hard to calculate. Based on this trouble, in some parts of the survey, the part is divided into sub-regions. This is to capture the fluctuations in the rainfall and temperature seasonality over different parts of the part. The choice of the sub-regions has been explained in subdivision 2. Section 2 besides explains the techniques used in gauging the synoptic characteristics from the two planetary theoretical accounts. Following the techniques is the consequences and treatment for the average climatologies and the seasonal fluctuations of the characteristics and rainfall. The concluding subdivision is the decisions, which will foreground the of import results of this survey.
2. Techniques
Both HadAM3 and CAM3 were applied to bring forth 30 old ages ( 1971-2000 ) clime simulations. A five member ensemble is integrated frontward with ascertained day-to-day Reynold ‘s Sea Surface Temperature ( SST ) . The ensemble members were produced from unhinging the initial conditions of the SSTs as input to the theoretical accounts for each ensemble tally.
It has been shown that ensemble prediction is one of the best methods for cut downing mistakes associated with clime uncertainnesss over single theoretical account ensemble anticipation. Therefore, addition in the figure of ensemble members, straight affect the consequence positively. Merely five members are used for the survey because of deficient calculating infinite to imitate more ensemble members.
Statistical norms of these ensemble members ‘ appraisal of the characteristics were used in the analysis. For better comparing of the theoretical accounts with the reanalysis informations, the theoretical accounts consequences were interpolated to the declaration ( 2.5o x 2.5o ) of the NCAR reanalysis I data ( NCEP ) . In the probe of the theoretical accounts to reproduce the synoptic graduated table characteristics, the fake 500hPa and 850hPa geopotential highs were analyzed and the consequences compared with those from NCEP reanalysis.
Anomalies of the geopotential tallness from the climatological mean for 30 old ages are calculated and a relationship is created from the form with that of rainfall.
The standardised anomalousnesss calculated from the mean of day-to-day geopotential tallness from a clip series of 1971 through 2000 period were done to take attention of the seasonal fluctuations within the dataset. This removes scattering in the dataset and helps to acknowledge the magnitude of the anomalousnesss.
Additionally, we use the Laplace equation to track cut-off depressions, temperate and tropical depressions from geopotential tallness. The designation of low force per unit areas was based on an 8 neighbor grid value of a two dimensional geopotential field at 500hPa for cut-off depressions and tropical temperate troughs ; 500hPa and 850hPa for deep tropical and temperate depressions. The 2nd derived functions from the Laplace equation allow the minimal value for the depression. The restriction used in sing the cut-off depression is the minimal geopotential tallness at a grid point and the closed circulation westerlies in the upper troposphere. On each twenty-four hours, a given grid point was identified as a geopotential lower limit ( gpm ) if it is within a lower limit of six out of the eight environing grid points. Once this set of cut-off low points was chosen, merely the grid points that showed a minimal geopotential tallness are retained. The tropical temperate trough signifiers when the upper westerlies coincide with the east winds wave in lower degrees. The algorithm records yearss with tropical temperate troughs on yearss where a nexus is established between a tropical depression and a western moving ridge trough ( van lair Heever, 1997 ) . The tropical part is defined in this survey as 0o – 25oS and 0o – 50oE and the temperate part as 25oS – 50oS and 0o – 50oE.
Correlation coefficient is calculated to mensurate the strength of the additive association between the synoptic characteristics and rainfall. The significance of correlativity coefficients is obtained from the tabular array of critical values of correlativity after ciphering the two-tailed trial. The assurance degree used is 95 % and with 12 months as the sample infinite, the grade of freedom is 10.
REG1 is between latitude 0, 20oS and longitude 0, 55oE. REG2 is between latitude 20oS, 40oS and longitude 0, 20oE and REG2 positioned at latitude 20oS, 40oS and longitude 20oE, 55oE.
The tropical temperate troughs form when a tropical depression is coupled to a temperate westerly wave via a semitropical trough ( Figure 5a ) , organizing a Northwest to southeast cloud set along the taking border of the western trough ( Harangozo and Harrison, 1983 ; Harrison, 1984c ; new waves den Heever, 1994 ) .
3. Consequences
a. Seasonal Rainfall form
Figure 1 compares the fake seasonal rainfall forms in the theoretical accounts with the ascertained ( NCEP reanalysis ) . By and large in NCEP for all seasons, entire rainfall is greater in the Torrid Zones. This bit by bit decreases due west such that most of the cardinal and western parts are semi-desert with low rainfall. Dryness to moo rainfall is observed over most cardinal and south-west of the part.
In general, the fake rainfall forms are closer to the observed in HadAM3 than in CAM3. CAM3, the rainfall form is different from that of NCEP. The theoretical account over-predicts rainfall over the cardinal in December-January-February ( DJF ) . In the same season, a zone of maximal rainfall lies along the western half and the north eastern portion of the Southern Africa. The most outstanding characteristic in the March-April-May ( MAM ) is the zone of maximal rainfall located between 5o and 15oS. Along the zone, rainfall lessenings from 8.0 mm/day at east seashore to about 4.0 mm/day. In HadAM3, the fake rainfall distribution is closer to that from NCEP than that from CAM3. In June-July-August ( JJA ) , the full part is dry having less than 1mm/day of rainfall in that season. Some sum of rainfall is merely found North of 10oS and South of 30oS. Both theoretical accounts estimate about no rainfall in JJA over most portion of the part. However, CAM3 has about 3-5mm/day of rainfall at the north-eastern part. The theoretical accounts capture the little sum of rainfall at the south-most portion of the part but CAM3 fails to capture the rains over Western Cape. In September-October-November ( SON ) , rainfall is observed to be maximal in the Torrid Zones and it extends to the cardinal portion of the part and a little part at the south-eastern portion can be seen with some sum of rainfall ( about 2mm/day ) . CAM3 simulates a rainfall part that extends from the Torrid Zones to the full sub-continent, doing the full part receive 1mm/day or more rainfall. HadAM3 simulations capture all the indispensable characteristics shown in NCEP, except that it by and large under estimations the rainfall forms in all seasons.
B. Seasonal Temperature form
In figure 2, the mean of the temperature for DJF, MAM, JJA and SON from 1971 through 2000 is shown from NCEP reanalysis, HadAM3 and CAM3. NCEP shows that during the stellar summer in DJF, temperature lessenings due souths from the equator. The lowest temperature ( 18oC ) is observed over the south most portion of the part in that season.
The theoretical accounts reproduce similar form but HadAM3 has 16oC and CAM3 has a lower limit of approximately 20oC. In MAM, NCEP shows a similar form but with a lower limit of 14oC over South Africa and upper limit of 28oC over north-eastern portion of Southern Africa. CAM3 reproduces a lower limit of 18oC and a upper limit of 26oC while HadAM3 reproduces a lower limit of 16oC and a upper limit of 26oC. In summer months ( JJA ) , a lower limit of 14oC is observed over South Africa and a upper limit of 24oC at the equator. The theoretical accounts capture these lower limit and the maximal temperatures values at about same places. During the SON season the Centre of the southern Africa experiences a temperature of 24oC. It decreases towards the South of the part. The theoretical accounts reproduce similar form in all seasons.
c. Anomalies in HGTs
The contour lines of Figure 3 depict the climatological anomalousnesss obtained from 30 twelvemonth norms of the 500hPa height Fieldss of DJF, MAM, JJA and SON from NCEP reanalysis, HadAM3 and CAM3 theoretical accounts. In DJF, weak anomalous depression is observed in NCEP over the oceans, which increases inland and a weak anomalous high over the Torrid Zones. In MAM, high geopotential is seen over southern portion and the oceans but a weak high over the Torrid Zones. Anomalous values of the geopotential tallness runing from -5 to -40gpm are concentrated over most cardinal and southern portion of the part in JJA. At 500hPa, semitropical high prevails during these seasons and a corresponding waterlessness to moo rainfall is observed in DJF, MAM and in JJA over most portion of the part. During SON, high anomalous values occur over the cardinal portion of the part. A high geopotential tallness appears merely over the oceans and over the southern portion of the part. The form here agrees with that of rainfall ( figure 1 ) ; maximum in the Torrid Zones and extends to the cardinal portion of the part except at the western portion of the part with some waterlessness.
HadAM3 seem to hold anomalous high in SON but right show the forms. Almost the same form and values are simulated for JJA. Although CAM3 have close values as that of NCEP, it simulates somewhat different form for MAM and SON.
Figure 4 shows the circulation anomalousnesss at the lower troposphere. In DJF, weak high is observed over the whole part. A high emerging from the southern Oceans extends into the cardinal portion of the part. Negative anomalousnesss are observed over the Torrid Zones. A strong high geopotential tallness is observed over every portion of the part in JJA. The associated circulation at the lower troposphere seems opposite to that at the upper troposphere. In JJA when the anomalous geopotential is low, in the upper troposphere, it is high in the lower troposphere and there is waterlessness over most portion of Southern Africa. When the semitropical high prevails at the lower troposphere, a corresponding waterlessness to moo rainfall is observed in DJF, MAM and in JJA over most portion of the part. In SON synoptic anomalous flow of low geopotential tallness emerge from the ocean to the southern Africa and extends to the Torrid Zones. Positive anomalousnesss are observed over the Indian Ocean near Madagascar and over the southern Atlantic Ocean. The low geopotential tallness experience in SON in most portion of the subcontinent and the Torrid Zones agrees with the rainfall form over the part. By and large, the depression over the ocean at 850hPa contributes to the rainfall over the most portion of the part and the corresponding high over the ocean contributes to the waterlessness in that season. A fleet alteration from high in JJA to moo in SON of the geopotential at the 850hPa connotes dryness in JJA and rains in SON. The nexus between the 850hPa geopotential tallness and rainfall has been reported in a related survey by Landman and Goddard ( 2002 ) utilizing a form analysis from Canonical Correlation Analysis. The theoretical accounts by and large capture most of these forms good for all the seasons. Particularly, HadAM3 shows really similar form as NCEP. However, it over predicts and shows a stronger gradient for both high and low geopotential over the part. CAM3 simulates somewhat different form, particularly for SON, a high is simulated over the cardinal portion of the part and a low shown over the south and in the Torrid Zones.
d. Standard divergences in U and V wind constituents
Standard divergences of zonary air current ( fig.4 ) scope from 5ms-1 at approximately 25oS to a upper limit of 12ms-1 in the south-most portion of the part. Standard divergences of meridional constituent ( fig.5 ) are similar to those of the zonary constituent…
e. Regional Seasonal Variations
The seasonal fluctuations in the fake climatological rainfall from the theoretical accounts and NCEP reanalysis for REG1, REG2 and REG3 are shown in figure 4. In figure 4a, rainfall over REG1 is maximal in summer and lower limit in winter. The highest is seen in January from NCEP and the theoretical accounts. Rainfall decreases to a lower limit in August with NCEP, May-July with CAM3 and June-August with HadAM3. HadAM3 has about same sums of rainfall with NCEP from January boulder clay May and so underestimates it until November. CAM3 overestimates the summer rainfall and underestimates it in the winter months. REG2 ( figure 4b ) experiences a general low sum of rainfall. Harmonizing to NCEP, the part records the lowest rainfall in summer and highest in winter. HadAM3 captures the fluctuations right but CAM3 fails to acquire it. However, the differences are merely less than 0.5mm/day. Over REG3 ( figure 4c ) , rainfall fluctuation is similar to that over REG1 except that the sum is less than that over REG1. The theoretical accounts reproduced about the same rainfall fluctuations as that from NCEP reanalysis.
Fig. 5 shows temperature fluctuations over the 3 sub-regions. Temperature is high in summer and low in winter. The fluctuations are really similar in all the sub-regions. Over all the parts, the theoretical accounts over-simulate the temperature from February boulder clay June and so under-stimulate in July, August and September. REG1 ( figure 5a ) has a short scope of temperature fluctuations of above 21oC to below 25oC. Temperature over REG2 scopes from 15oC to 20oC and that over REG3 from 16oC to a value below 23oC.
The seasonal fluctuations of the standardised rainfall and temperature anomalousnesss over southern Africa for the 1971-2000 twelvemonth period are shown in figure 6. The standardised anomalousnesss are calculated from their day-to-day anomalousnesss and monthly estimations of the climatological criterion divergences are presented. In figure 6a, positive standardized anomalousness of rainfall about equal to 1 is observed, from NCEP, in January and it decreases to zero at the terminal of April. The negative anomalousness starts in May and extremums in July, it so rises through to October. Positive anomalousness starts once more at the beginning of November and it increases in December. The theoretical accounts simulate these anomalousnesss really near to the reanalysis. Both have a correlativity coefficient of 0.98 with NCEP reanalysis. In figure 5b, positive standardized anomalousness of temperature lessenings from 1 in January to zero at the terminal of April. The negative anomalousness starts in May at zero and decreases to approach -1.5 in mid July, it so increases gently from -1.5 in July to zero in October. The Positive anomalousness starts once more in October at zero to a small above 0.5 in December.
Figure 7 compares the theoretical accounts standardized anomalousnesss of their day-to-day geopotential tallness at 500hPa ( figure 7a ) and near the surface at 850hPa ( figure 7b ) with NCEP reanalysis. In figure 7a, NCEP shows a positive anomalousness that increases from January boulder clay April and in December. The positive anomalousness extremums in March at 1.5 and lessenings to zero in May. The negative anomalousness is between 0 and -1.5, within this scope, the standardised anomalousness lessenings from zero in May to -1.5 in August but rises once more between August and November. Similarly, both theoretical accounts simulate a similar form in the scope of their divergences. In peculiar, CAM3 reproduces the positive anomalousnesss from January through May and negative anomalousnesss between June and November. December has positive anomalousnesss. The extremum of the positive anomalousnesss in this instance is in April. Like in NCEP,
CAM3 simulates the negative anomalousnesss from June through November. Furthermore, CAM3 shows a correlativity coefficient of 0.89 with the NCEP. Interestingly, HadAM3 reproduces a similar form as in NCEP with a correlativity coefficient of 0.94. In figure 7b, negative anomalousnesss at 850hPa are observed from January through May and from October through December. The winter months get downing from May hold positive anomalousnesss up to about 1.5. The scope is between -1.5 and 1.5. The theoretical accounts simulate the scope of the standardised divergence as in NCEP. However, CAM3 simulates the passage from negative to positive anomalousness more than a month earlier, sometime in March while HadAM3 has a closer passage clip as NCEP.
Figure 8 shows the seasonal average figure of cut-off depressions estimated from NCEP and the two-models. A high average figure of cut-off depressions is observed from NCEP during the oncoming of the austral winter season in March and April from NCEP. Singleton and Reason ( 2007 ) besides found that cut-off depressions over southern Africa are most common in the March-May season. A smaller extremum is present in October. Both theoretical accounts capture the extremum of the average figure of cut-off depressions in March but they represent the 2nd extremum ( in October for NCEP ) a spot earlier in September. Although, both theoretical accounts by and large under estimation the average figure of cut-off depressions over the part, HadAM3 simulates the average figure closer to that of NCEP reanalysis with a correlativity coefficient of 0.63 with NCEP.
In figure 9a, the ascertained figure of yearss with deep tropical depressions from NCEP varies between 17 and 30 per month. The highest figure of yearss with tropical depressions from NCEP is near to 30 in February. The figure of yearss with deep tropical depressions decreases as winter attacks. May, June and July recorded the lowest figure of yearss with tropical depressions and 17 yearss is the ascertained lower limit. The figure of yearss additions from July through December and from January through February. Both theoretical accounts underestimate the figure of yearss with deep tropical depressions although they show similar form to that of NCEP. For CAM3, the months January, February, March and December have 25 yearss of deep tropical depressions. The figure of yearss with tropical depressions decreases to 15 in April and May so increases from 15 to 20 through June. CAM3 has a correlativity coefficient of 0.87 with NCEP. In the instance of HadAM3, the figure of yearss with deep tropical depressions is in the order of 5 to 20 yearss from January through December. It besides reproduces less figure of yearss with tropical depressions from the oncoming of winter season in May till the extremum of winter season in July but the correlativity coefficient is still high ( 0.83 ) .
In add-on, the figure of yearss with deep temperate depressions from NCEP reanalysis and the theoretical accounts is shown in figure 9b. By and large, lower figure of yearss with deep depressions is seen from NCEP over the temperate part than over the tropical part ( see figure 9a ) . The figure of yearss with deep temperate depressions increases from September through December and besides from January through March. The highest figure of yearss with deep temperate depressions from NCEP is 8 in March. The NCEP reanalysis shows the lowest figure of yearss with deep temperate depressions to be 4 per month during the austral winter season. The theoretical accounts highly underestimate the figure of yearss with deep temperate depressions, although the form is clearly represented. CAM3 peculiarly estimates about 0 figure of yearss per month of temperate depressions in winter and shut to 1 twenty-four hours per month in summer. HadAM3 estimates about 2 yearss of temperate depressions in summer and winter months have between nothing and 1 twenty-four hours of temperate depressions per month. The highest estimated figure of yearss of these depressions is less than 2 in April. The theoretical accounts have a correlativity coefficient of 0.59 and 0.47 for CAM3 and HadAM3, severally.
Figure 10 shows the seasonal fluctuation of yearss with TTTs from 1971 to 2000 over the same sphere. It is apparent that the figure of yearss with TTTs estimated from NCEP additions from September through December. It peaks in March. The figure of yearss with TTTs from NCEP is between 4 and 8 throughout the seasons. The form is reproduced reasonably good in both theoretical accounts except with CAM3, which overestimates the figure of yearss with TTTs throughout the twelvemonth and more significantly in January, February, March and December. HadAM3 reproduced the figure of yearss with TTTs between 4 and 8, as seen from NCEP reanalysis.
4. Discussion
Most of the synoptic graduated table characteristics presented here are rainfall bring oning systems, which are noticeable at the 500hPa. The sum of geopotential tallness variableness, as measured by standard divergences, is compared between the seasons of rainfall and temperature.
As seen in the spacial forms, the average geopotential tallness at 500hPa is similar to that of the mean rainfall and temperature over the survey part. Negative anomalousnesss in the day-to-day geopotential tallness at 500hPa correspond to low climatological rainfall between May and November.
In contrast, at the 850hPa ( figure 7b ) , positive anomalousnesss in the day-to-day geopotential tallness correspond to low climatological rainfall between May and October. The nexus between the 850hPa geopotential tallness and rainfall has been reported in a related survey by Landman and Goddard ( 2002 ) utilizing a form analysis from Canonical Correlation Analysis.
Furthermore, from figures 8 and 4a, as the average figure of cut-off depressions increases, there is an addition in the mean rainfall and when the average figure of cut-off depressions decreases, the mean rainfall decreases. However, a correlativity coefficient of 0.44 non important at 95 % assurance degree is shown for cut-off depressions from NCEP and NCEP rainfall ( table 1 ) . CAM3 besides shows a low correlativity coefficient for cut-off depressions and rainfall, which is non important. In table 1, HadAM3 shows a high correlativity coefficient for cut-off depressions and rainfall which is important at 95 % degree of assurance.
In add-on, the months with a high figure of yearss of TTTs correspond to months with high rainfall. The nexus between TTTs and rainfall over southern Africa has been explained through wet convergence by Todd and Washington ( 1999 ) . Besides, low figure of yearss of temperate depressions agrees with low winter rainfall over southern Africa. It can be seen in table 1, that there is a important correlativity coefficient from NCEP for tropical depressions, temperate depressions, TTTs and NCEP rainfall at 95 % assurance degree. Similarly, CAM3 besides shows high correlativity coefficients which are important at the same assurance degree for the same characteristics and rainfall. However, HadAM3 shows a strong correlativity coefficient for tropical depressions and for TTTs but no relationship at all for temperate depressions. The form of the characteristics discussed above can be confirmed from related surveies ( Harrison 1984a ; Miron and Tyson 1984 ; Tyson 1986 ; Mason and Joubert 1997 ; Todd and Washington 1999 ) .
5. Decision
The theoretical accounts have shown their capablenesss in reproducing the synoptic graduated table characteristics over southern Africa in comparing with the NCEP reanalysis. The fluctuations in the geopotential tallness has been studied with the theoretical accounts and compared with NCEP reanalysis. The variableness of the synoptic graduated table characteristics is associated with the place and frequence of rainfall over the part. The theoretical accounts correlate good with NCEP in the standardised geopotential tallness anomalousnesss at both 500hPa and 850hPa. The fluctuations in the geopotential tallness have been linked with rainfall and temperature through their agencies and standardised anomalousnesss. The standardised anomalousnesss in the geopotential tallness at 500hPa have been shown to be in stage with the standardised anomalousnesss of rainfall and that at 850hPa to be out of stage with that of rainfall and temperature. In the summer months ( December-January-February ) , when the part experiences maximal rainfall, synoptic graduated table characteristics like cut-off depressions, tropical depressions and TTTs show increased strength. Besides during winter, low rainfall corresponds to low strength of the synoptic graduated table characteristics. The seasonal fluctuation in tropical depressions, temperate depressions and TTTs has besides been shown to associate with the average rainfall with strong correlativity coefficient, which are important at 95 % assurance degree.
By and large, the theoretical accounts are able to reproduce the synoptic graduated table circulation characteristics and have estimated the relationship between them and rainfall. These characteristics are important for dependable seasonal prognosis over southern Africa. Furthermore, the truth of prognosis produced from these planetary theoretical accounts will depend on the ability of these GCMs to imitate the synoptic graduated table circulation characteristics that play dominant functions in finding the clime over the part.