DISCUSSIONS 551.577.2 (676.1 : 676.2 : 678.2) : 551.577.3

Rain in East Africa

By D. H. JOHNSON

(Read 17 October 1962. See Q.J., 88, p. 1)

Dr. R. C. RAINEY (partly communicated) : Is the author’s implication that weather at one season in Uganda may be ‘ inherently unpredictable ’ possibly another way of saying that in that particular country and season ‘ weather consists of little more than climate ’ - with the absence of significant serial correlations perhaps attributable to day-to-day departures from average conditions having been relatively small ?

Mr. Johnson’s evidence that the more arid areas, on the other hand, in fact receive a relatively large proportion of their total precipitation in the form of widespread heavy rain is of considerable significance in relation to the biology of these extensive areas, and I should like to ask whether this finding is also likely to be true of arid regions elsewhere. In Arabia (for example) Jeddah, with a mean annuai rainfall of about 70mm, averages about two days with more than 10mm of rain in every year, and during two Decembers, out of eight, experienced a rainfall of more than 120 mm in a single day.

Mr. J. W. POTHECARY : In dealing with the detailed rainfall data was there any evidence for a relation between rainfall distribution and preferred positions for the formation of cloud ‘ streets ’ ? The distribution of cloud ‘ streets,’ shown in the slides illustrating vegetation types, suggested that the resolution of the raingauge network might have been sufficiently fine for such a relation to be detected.

Dr. H. L. PENMAN (President) : I know that hydrologists and agriculturalists in East Africa will welcome this valuable addition to knowledge of rainfall in the region. Mr. Johnson’s time and motion study has emphasized one of the great gaps in our knowledge of rainfall physics - where does the water get into the air ? We are all agreed that local evaporation is not the source of local rain, but for his storms a source area of about ten times the size of the sink area could be adequate. If so there might be some correlation between the size of the sink area and the duration of rain over it. Is this so ?

Dr. F. H. LUDLAM : Mr. Johnson has done very well to make some sense of the behaviour of the rains in East Africa, in spite of the apparently complete absence of data on the scale of the rainclouds themselves. This might be remedied by the use of radar, or even by attentive visual observations. It is surprising that no attempt seems to have been made to introduce ‘ sferics ’ techniques. With the help of data from these sources to determine more accurately the character of the convection, one could hope to relate it more closely to the topography and the large-scale situation.

Mr. D. H. JOHNSON (in reply, partly communicated) : On many occasions the air which feeds East African rainfall systems at low levels is drawn directly or indirectly from the east, where the Indian Ocean monsoon and tradewind flows provide virtually inexhaustible supplies of moist air, and it seems rather unlikely that the durations of large-scale rainfalls are limited by moisture supply, as suggested by Dr. Penman. However, since the rainfall patterns have a cellular structure, one might plausibly suppose that for a given large-scale convergence the rates of rainfall within the cells would be inversely porportional to their size. We have checked this possibility for the areas of general rain (10-areas) with the results given in Table 1. The three largest 10-areas were of sizes 33&, 283 and 253 one-degree squares, and gave mean daily rainfalls of 79, 75 and 103 in. respectively. Thus there was no significant tendency for rainfall amounts to decrease with increasing sink area. It was mentioned in the paper that cases of high rainfall over areas of order 100 km radius were usually associated with much larger areas of widespread rain.

TABLE 1. RELATION BETWEEN RAIN-AREA SIZE AND MEAN DAILY RAINFALLS FOR 10-AREAS

Area size (one-degree squares) 1 -2 24-34 4-8 86-123 13

Mean rainfall in.) 67 71 72 76 75

Number of areas 189 169 166 61 32

291

292 DISCUSSIONS

The serial correlation to which Dr. Rainey refers, applied to the months August-September in Region A. Table 2 gives the percentage frequencies with which the regional indexes, calculated for region A, D and F in August-September, fell into specified ranges. The relevant serial correlations and averages of the indexes are also shown. The spread of values for region A was not particularly small and was not less than the spread for region D in which there was an appreci- ably higher persistence effect. I think it would be misleading to say that the weather consisted of little more than climate even in region A during August-September, but hope of discriminating in advance between the relatively wet and the relatively dry days must depend upon the possi- bility of relating the rainfall with effects (such as for example, the total low-level convergence over Uganda as a whole) which can be measured at the outset of the day; the old mixture of persistence and climatology will not serve this purpose.

TABLE 2. PERCENTAGE FREQUENCIES OF REGIONAL INDEXES IN AUGUST-SEPTEMBER

Serial correlation Mean Region in daily indexes Ranges of index index

0 0 1.0 2.0 3.0 4.0 5.0 6 0 7.0 - 0.9 - 1.9 - 2.9 - 3.9 - 4.9 - 5.9 - 6.9 - 7.9

2.0 F 0.74 27.9 32.8 16.4 14.8 1.6

D 0.65 3.3 14.8 34.4 262 18.0 3.3 2.9

1.6 1.6 3.3

A 0.29 66 21.3 23.0 24.6 23.0 1.6 3 8

Particularly vigorous disturbances must be necessary to break down the persistently stable systems which prevail over arid regions. O n that basis, the finding that a large proportion of the total precipitation falls as widespread rain in the semi-desert parts of East Africa seems reasonable and likely to apply more generally, as Dr. Rainey’s Jeddah figures suggest.

By choice of data and technique we have deliberately filtered out the convective-scale varia- tions in an attempt to establish the existence and broad nature of the ‘ large-scale situations,’ to use Dr. Ludlam’s phrase : ‘ one must be able to recognize the large-scale situation before one can relate the convection to it.’ We do not doubt the importance of the convection cells in characterizing the structure and energy conversions of the broader-scale weather systems. The visual and radar studies of the distribution and nature of the rain clouds, advocated by n r . Ludlam, are essential.

There may have been evidence of cloud streets in the plotted rainfall data, but we did not specifically look for it. Streets, both of fair-weather cloud and precipitating cumuli or cumulonimbi, are common in East Africa and from visual observations in Nairobi I would say that for given wind regimes, there were certainly preferred positions for them. Mr. Pothecary’s question would best be answered by studies of the kind suggested by Dr. Ludlam.

551.509.313 : 551.511.32 : 551.558.21

The introduction of topographic and frictional effects in a baroclinic model

By P. GRAYSTONE

(Read 17 October 1962. See Q.J., 88. p. 256)

Dr. R. C. SUTCLIFFE : The treatment of surface friction is ingenious. As I see it the idea is that frictional divergence in the surface layer is balanced (almost) by divergence of opposite sign in the column above. The frictional flow is, however, almost a steady state, frictional retardation being balanced by pressure gradient and Coriolis terms. Thus almost the whole effect is given by the divergence in the free atmosphere corresponding with the vertical velocity at the top of the friction layer, treated as a boundary. This seems dynamically acceptable.