Challenges in African Hydrology and Water Resources (Proceedings of the Harare Symposium, July 1984)'. IAHS Publ. no. 144.

The character of fractured rock aquifers in eastern Botswana

D. K. BUCKLEY* British Geological Survey (Hydrogeology Group), Wallingford, Oxfordshire OXIO 8BB, UK P. ZEIL+

Geological Survey Department, Private Bag 14, Lobatse, Botswana

ABSTRACT Hydrogeological studies for village water supply from basement rocks in eastern Botswana show that groundwater occurrence is restricted to linear structural features associated with faulting and intrusion. The fractured aquifer which develops in this situation is characteristically channel-shaped of large length and narrow width and is best regarded as a vertical feature rather than a horizontal one. The most promising location for borehole sites has proved to be at the major inter­sections of the separate channels. Standard exploration techniques can be used to identify these channel aquifers and a combination of magnetometer and resistivity methods proved useful to locate successful sites. Particular challenges faced by the hydrogeologist working with these aquifers lie in locating suitable structures, estimating possible recharge and analysing the aquifer response to pumping.

Les caractères des aquifères dans les roches fracturées au Botswana RESUME Des études hydrologiques pour l'approvisionnement en eau des villages à partir des roches du socle dans l'est du Botswana montre que l'existence d'eau souterraine reste limitée à des caractéristiques structurales linéaire associées avec des failles et des intrusions. L'aquifère de zone fracturée qui se constitue dans cette situation est caractérisé par des formes de canaux de grandes longueurs et très étroits et doit être plutôt considéré comme un système vertical plutôt que comme un système horizontal. Les implantations les plus prometteuses pour les forages se sont avérées être les intersections principales de diverses failles. Les techniques standard de prospection peuvent être utilisées pour identifier les aquifères de ce type et une combinaison des méthodes magnétométriques et par résistivité s'est avérée efficace pour localiser des sites donnant toutes chances de succès.

* Présent address: Geological Survey Department, Private Bag 14, Lobatse, Botswana . ^On contract from the Federal Institute of Geosciences and National Ressources, Hanover, FR Germany.

25

26 D.K.Buckley & P.Zeil

Les dé f i s p a r t i c u l i e r que do i t a f f ron t e r 1'hydrogéologue t r a v a i l l a n t sur ce type d ' a q u i f è r e sont la l o c a l i s a t i o n de s t r u c t u r e s convenables , l ' e s t i m a t i o n des p o s s i b i l i t é s de recharge e t l ' a n a l y s e de l a r é a c t i o n de l ' a q u i f è r e au pompage.

BACKGROUND

As p a r t of a programme of groundwater resources development, the Botswana Geological Survey and Department of Water Affa i rs r e c e n t l y c a r r i e d out an exp lo ra t ion study for Mochudi v i l l a g e water supply in the e a s t e r n pa r t of Botswana ( F i g . l ) . A supply of approximately

20° 21° 22° 23° 24° 25° 26° 27° 28° 29°

20'

22<

23'

24'

25'

26°-

27°k

"1 1 1—V r

ANGOLA \ ZAMBIA

Kasane \

/ Ota,»t<> D"«

Tsau ©

Toteng

® Orapa Francistownf v Shashe?

NAMIBIA ® Ghanzi

BOTSWANA

/Phikwe / O «

Serowe I - , . * Bobonong" v O / S e l e b l

Moropule © @ Palapye .. '

Mahalapye <

0 100 200 300 Km

Mmamabule < 2426A

Moiepolole

FIG.l Location map.

3 — 1

1700 m day was required in an area where existing borehole records indicated less than 10% probability of achieving a borehole yield of 8 m h~ , considered the minimum equipped yield for village supply. Careful borehole siting was thus necessary to meet the requirement unless a very large number of boreholes was to be drilled.

Fractured rock aquifers in eastern Botswana 27

Appro-ximately 60 boreholes exist in the village of which only the five with yields of more than 8 m h-1 are in use. This paper summarizes information from the existing boreholes and from additional drilling within a 10 km radius of the village and describes the aquifer system found to be present.

GEOLOGY

Geological maps (2426A and C) published by the Geological Survey in 1973 show only three main rock types in the area near the village. A grey granitic quartz-biotite gneiss ^2600 million years old under­lies the whole region forming a vast (late Tertiary) peneplain south of the village with little rock exposure. North of the village the gneiss is overlain by Waterberg sandstones and siltstones (̂ 1420-1950 million years old), which build higher land with a prominent basal sandstone escarpment. Finally dolerite dykes and sills (̂ 1200 million years old) extensively invade both rock types, though generally they are not well exposed.

A review of existing borehole records and reports (Wilson, 1956; Sir Alexander Gibb & Partners, 1975) revealed that dolerite was more extensive than suggested by its mapped outcrop, and it was evident that better yields and higher success rates were obtained from drilling close to dolerite intrusions. Nevertheless, many unsuccessful boreholes had been drilled in the area.

REMOTE SENSING

The main structural features affecting the region were identified from Landsat images. The main trends were northeast-southwest (occupied by main drainage), north-south, northwest-southeast and east—west. A comparison of satellite images during the wet and dry season proved valuable for identifying lineaments. This was done stereoscopically using 18.5 cm square colour transparencies. Several features not apparent on paper prints could be seen by this method. On the basis of existing borehole yields three main lineaments were judged to be of direct relevance to groundwater exploration including a long northwest-southeast dolerite dyke extending 8 km north of the village and 30 km south. Air photo interpretation revealed linear features with similar orientation to those identified on the satellite images; the most conspicuous lineations were northwest-southeast and northeast-southwest.

FIELD INVESTIGATION

Geophysics

Initially the standard geophysical methods, employing magnetometer measurements and resistivity soundings, were used to locate dolerite margins and zones of deeper weathering. Resistivity soundings were made at existing boreholes, both high and low yielding, to calibrate responses with geology, but it was concluded that this approach of

28 D.K.Buckley S P.Zeil

looking for horizontal features was not correct , because no such features associated with good borehole yields existed in the area.

Therefore, the identification of lateral changes and vertical structural features became important and the lineations interpreted from Landsat images, air photos, airborne magnetic and electro­magnetic data (from mineral surveys) were selected as targets for a more detailed geophysical study (Figs 2 and 3). Because many of the features had little or no surface expression, square-gridded magnet­ometer and closely-spaced resistivity measurements were made in the selected areas to locate anomalies. This approach revealed considerable changes in implied physical properties between geophysi­cal soundings only 50 m apart and indicated the scale of the structural features. For two areas resistivity data were plotted as depth-sections (Fig.3) and vertical features, interpreted as fractures and faulting, can be traced over the grid. At borehole 4258 a north-south, east-west junction of channel-type features was indicated by the resistivity data (Fig.4).

For the interpretation of resistivity data the ZOHDY algorithm was used: in contrast to the normal solution given in layers of constant resistivity, a step-like resistivity-depth function is automatically derived. This is probably nearer to the real physical subsurface conditions, and was found helpful in revealing changes in resistivity along profiles . The major disadvantage of this method is that, due to the close spacing necessary, it is very time consuming, even if using the half SCHLUMBERGER configuration. The application of electromagnetic methods, VLF-R in combination with VES, would be more rapid.

Relative magnetic amplitude (100 gammas) Aeromagnetic data reproduced by permission of the International Nickel Company

(20 gammas)

FIG.2 Geological and aeromagnetic data.

Fractured rock aquifers in eastern Botswana 29

SANDSTONE

FIG.3 Resistivity depth-sections.

Dr il ling

Borehole drilling demonstrated that three variable layers were present, consequent upon weathering of both fractured and non-fractured

30 D.K.Buckley s P.Zeil

areas : (1) Soft upper layer: soil and weathered zone up to 50 m thick. (2) Intermediate layer: broken and fractured layer at base of

weathered zone, 0-5 m thick. (3) Hard bedrock: solid bedrock, softer where fractured and

jointed.

MN S

0 1 2 3 U 5 km 1 I 1 ' ! =d KEY TO BORF.HnLES

contours i n ohm-metrGS

FIG.4 Resistivity map (70 m depth).

Logs of drilling penetration rate confirmed the progressive increase in hardness with depth and also helped to identify litho-logical changes, intrusive junctions, fractures, joints and water strikes (Fig. 5). Water strikes in the weathered zone were normally encountered in the broken and fractured zone close to bedrock. Water strikes in bedrock were related to fracturing or intrusion. A somewhat thicker weathered zone was recognized over fracture zones but generally proved to be a local, narrow development (less than 50 m wide) necessitating closely-spaced resistivity measurements for successful identification. The weathered zone was normally cased off with blank casing and bedrock supplies were developed "open hole". Occasionally where a good supply was encountered in the weathered zone wellscreen or slotted casing was used to develop these supplies as well.

INTERPRETATION

The aquifer system

The aquifer system in both the granitic gneiss and the Waterberg rocks was found to be governed by vertical or sub-vertical rock

Fractured rock aquifers in eastern Botswana 31

Penetration rate (minutes/metre)

4235 4324 4234 0 5 10 15 20 0 5 10 15 20 0 5 10 15 20 • • i 1 1 i 1 1 1 1 i i l i i

4346 (inclined) 0 s 10 15 20

0 - ,

10

2 0 -

30

4 0 -

f 50

£ 60

£ 70 a co tJ 8 0 -2? o 90 CO o 100-03

110

120.

130

140-

150

WG

FG

GG

-Water strike

KEY

WG

FG

GG

Weathered granite-gneiss

Fractured granite-gneiss

Granite-gneiss

WG

GG

:;p:-;

:D

BC

ÏK

:D::

Colluvium

Weathered dolerite

Slightly weathered dolerite BC

Fresh dolerite

Sandstone, arkose, grit

Baked contact

FIG.5 Geological and penetration-rate logs.

fractures. Major groundwater flow paths occupy the main fracture and crush, zones and, on a larger scale, these act as the drainage conduits of th.e rock mass. The aquifer system can therefore be conceptualized by zones of vertical fracturing (Fig.6) separated by large blocks of relatively unstructured rock. Away from the main fractures, a system of minor joints may have small groundwater circulation but

Q = 50 » I

local fl.ow direction

gional flow d i rect ion

surface drainage d i rect ion

FIG.6 Schematic representation of the aquifer system.

32 D.K.Buckley S P.Zeil

further away there is negligible groundwater circulation. Boreholes 3 — 1

located in the main channels show sustained yields up to 20 m h , 3 — 1

those in the system of minor fractures 3-5 m h and those penetrating 3 — 1

close to dolerite intrusions 5-8 m h . Boreholes located at the intersection of the main channels show the highest yields, up to

3 — 1 3 — 1

50 m h and up to 90 m h is known from elsewhere. The weathered zone provides some measure of (horizontal) hydraulic continuity between these essentially vertical fractures and may be capable of storing water and transmitting it to the vertical zones. Groundwater strikes are more often made in the weathered zone; however, the general lack of recharge leads to a more important circulation in the underlying fracture zones of the bedrock which is recharged at out­crop. There is some evidence from the coincidence of fracturing, jointing, slickensides, platy secondary minerals with water strikes at the surfaces of dolerite intrusions, that these intrusions create openings (parasitic fractures) now occupied by groundwater. The frequent sill-like form of the intrusions with their associated horizontal openings leads to higher borehole success rate near their margins.

Inclined drilling was carried out at one site where the geophysical soundings suggested a narrow zone of closely-spaced vertical fractures. The penetration rate of this borehole was unlike those of vertical boreholes at the same site (Fig.5), providing good evidence for vertical fracturing and demonstrating the higher efficiency of inclined exploration boreholes in such aquifers. In general, however, the drilling results were poor; of the boreholes drilled five were

3 — 1

moderate to high yielding (12-50 m h ), 20 were low to moderate 3 — 1

yielding (0.1-9 m h ) and 12 were dry. There is evidence from the drilling results and from work done

elsewhere in Botswana (VIAK AB, 1983) that the main groundwater circulation is not conducted through the most obvious lineaments seen on air photos and satellite scenes (generally shear zones) but through a system of less obvious joints and fractures of tension origin related to the main shear lineations. The high sustained

3 — 1

yields (50-90 m h ) recorded at some fracture intersections suggest that these intersections must lie within a tension zone and that an analysis of the tectonics in terms of stress directions is nessary for successful siting.

In all the boreholes groundwater rose 10-20 m after being struck to within the weathered zone, approximately 20-30 m below ground level. This is not taken as evidence of a fully confined aquifer. In some instances the full aquifer thickness has to be drilled before suffi­cient hydraulic connection is established to give a discharge, so that drilling can pass the first water strikes unnoticed. More often it is due to striking groundwater circulating in discrete bedrock fractures which are hydraulically connected to elsewhere in the system. The aquifer as a whole should be regarded as unconfined, capable of receiving diffuse recharge at the land surface, but some features of confinement are apparent due to the localized circulation. The groundwater table reflects the topography; whilst very flat on a regional scale it conforms with the surface drainage showing an overall gradient towards the basin outlet. Locally, however, the groundwater flow directions are controlled by fracture orientat­ion.

Fractured rock aquifers in eastern Botswana 33

Boreh<ole pump testing

Standard analysis of pumping test data showed the transmissivity (T) of the fractured aquifer to be very variable; values ranging from

2 — 1

less than 5 to 60 m day were obtained. Part of the variation can be ascribed to the fractured nature of the aquifer but some tests, particularly those at high yielding sites in the Waterberg rocks, appear to show a non-uniform response radially, and the normal methods of analysis are inappropriate.

A feature of these tests has been a continuing drawdown with time leading to curves on semi-log plots and straight lines on log-log plots . The T-values obtained by normal analysis are therefore time-dependent. For example at borehole 4345, T after 100 minutes = 600 m2day_:L, after 48 h = 108 m2day-1 and after 5 days = 45 m2day~J. A more appropriate analysis is given by a plot of drawdown vs. t5 which gives a straight line indicating a linear flow response. A drawdown equat ion can be obtained from the straight line either graphically or by linear regression of the raw data and can be used to predict when water levels will approach critical dewatering (Fig.7).

This drawdown response can be explained by the nature of the

10

I 15 o

T3

5 20-ro

Q

25-

3 0 -

10

tSmins)

100 10000

• ( Observation well

Borehole 4345

(pumpwell)

" - • - . . .

^t(mins) 10 20 30 40 50 60 70 80 90 100 110 120

I i l I I I I I I l I l

' ^ ^ e h o / e 4 2 S 7

i^Mole, Q = s

"1 1 r — - r

FIG.7 Plots of time (t) vs. drawdown and t2 vs. drawdown.

34 D.K.Buckley & P.Zeil

fracturing. For boreholes well-located in a main fracture channel the initial drawdown follows a radial-flow pattern but as the "trough of depression" expands the drawdown is controlled by the release of water from the minor fractures to the major fractures and ultimately from matrix to minor fractures. The changeover from radial to linear flow occurred at about 400 minutes in several tests which explains why short term tests (e.g. 100 minute step-drawdown tests, and short constant rate tests) appear to indicate radial flow.

Recharge

Estimating the recharge to these aquifers is not straightforward. A water balance approach for short period (10-day) data was attempted, using soil moisture/unsaturated zone storage estimates (100-200 mm) and calculated actual évapotranspiration (Morton, 1978, 1983), but no reliable runoff measurements were available and thus, the method could not be taken to balance. Moreover, in a fractured channel-form aquifer it is theoretically possible, and indeed likely, that the quantity of water recharged could exceed local rainfall due to "run-in" effects. Under such circumstances the water balance approach is invalid.

An estimate of recharge was therefore made by comparing water level changes in the aquifer supplying Mochudi village (some 10 km x 0.5 km in extent) with pumped abstraction. This led to a 1982 abstraction equivalent of 50 mm recharge over the entire aquifer area. This represents 10% of the rainfall, and groundwater level records in the village wellfield suggest that only about 30 mm was removed from the aquifer storage, so that the recharge could total up to 20 mm year- . An independent calculation based on chloride mass balance suggested 7 mm year- .

Evidence that rainfall recharge does, in fact, occur comes from monitoring groundwater quality. Serious nitrate and faecal bacteria contamination of groundwater in Mochudi village has occurred from pit latrines in an area of thin soil cover and rock outcrop (Foster et al., 1978). Routine sampling of pumped water supplies from this area demonstrates a marked reduction in chloride content after rain­fall events (Fig.8).

1980 1981 1982 1983

FIG.8 Chloride concentration and rainfall at Mochudi boreholes 1980-1983.

Fractured rock aquifers in eastern Botswana 35

Significantly the largest variations in chloride content are shown by borehole 2784 which has the highest nitrate concentrations (up to 250 rag 1_ ). The borehole is located in an area of thin soil and rock outcrop where there is easy and rapid connection to the water table. Two recharge mechanisms appear to exist: a slow diffuse rechaxge through the weathered zone and a more concentrated and rapid recha rge through areas of fractured rock outcrop having no or only thin soil. It is easy to see that in times of low rainfall diffuse recharge through the weathered zone will rarely occur because the accumulated soil moisture deficit will practically eliminate all infiltrating water and only concentrated recharge through rock out­crop has a chance of reaching the water table. Recharge through fract ured rock outcrop is therefore the dominant recharge process and this explains why water levels normally rise after penetrating the weathered zone and explains the presence of faecal bacteria together with low nitrate in some polluted groundwaters (rapid access, thin or absent soil nitrate reservoir). It seems probable in fact that for areas of very low rainfall having significant soil cover, and lacking fossil groundwater supplies, potable groundwater occur­rence will be limited to areas where concentrated recharge can take place .

CONCLUSIONS

(a ) Groundwater in an area of fractured basement rocks of eastern Botswana is shown to be largely restricted to certain narrow, vertical to sub-vertical, fracture zones related to faulting and intrusion.

(b ) The groundwater circulation is controlled by these fracture zones and distant from them there may be little or no groundwater circulation with large blocks of dry country.

(c ) The most promising location for water-supply boreholes is at the intersections of the main fractured zones, the zones themselves, and what are believed to be tension-zones.

(d ) Very detailed geophysical survey may be necessary to locate these features precisely. A combination of gridded magnetometer and closely-spaced resistivity measurements was applied successfully to locate such features, but the method was time-consuming. Electro­magnetic (VLF-R) methods may well prove more convenient.

(e ) An automatic resistivity interpretation and compilation of resistivity-depth sections proved useful for identifying lateral variations and the vertical structures.

(f ) Drilling of inclined exploration boreholes is considered a suitable exploration method in such aquifers, and was used success­fully . It is expected that inclined boreholes will generally have a higher success rate.

(g ) The response to borehole pumping from these aquifers may depart significantly from classical theory and show a continuing drawdown with time consistent with linear flow.

(h ) Aquifer recharge is believed to occur through two main processes: diffuse rainfall recharge through the weathered zone and concentrated recharge through areas of thin soil or rock outcrop. Diffuse recharge occurs only rarely and the dominant process is

36 D.K.Buckley & P.Zeil

concentrated recharge through rock outcrop.

ACKNOWLEDGEMENTS The authors would like to acknowledge the assistance and interest in this work given by the Principal Hydro-geologist, Dr M.von Hoyer, and the Deputy Director, M.Sekwale, of the Geological Survey, and also to acknowledge the assistance provided by the senior hydrogeologist of the Department of Water Affairs for providing the drilling facilities.

The paper is published with the authorization of the Ministry of Mineral Resources and Water Affairs, Republic of Botswana, and the permission of the Director of the British Geological Survey, Keyworth, Nottingham, UK.

REFERENCES

Foster, S.S.D., Farr, J.F. & Lewis, W.J. (1978) A detailed evaluation of the pollution hazard to village water-supply boreholes in eastern Botswana. Geological Survey Report GSlO/4. Sir Alexander Gibb & Partners (1975) Feasibility study for water supplies to Molepolole and Mochudi villages. Report to Department of Water Affairs, Republic of Botswana.

Morton, F.I. (1978) Estimating évapotranspiration from potential evaporation: practicality of an iconoclastic approach. J. Hydrol, 38, 1-32.

Morton, F.I. (1983) Operational estimates of areal évapotranspiration and their significance to the science and practice of hydrology. J. Hydrol. 66, 1-76.

VIAK AB (1983) Lineament studies in the area between Molepolole and Letlhakeng. Report to Geological Survey, and Department of Water Affairs, Government of Botswana.

Wilson, P.T. (1956) Borehole siting at Mochudi. Geological Survey Report PTW/2/56, Government of Botswana.