ON THE ROLE OF RAINFALL DISTRIBUTION AND SOIL TYPE ON RUNOFF
GENERATION IN SEMI-ARID SOUTHERN ZIMBABWE
F.T. Mugabea, M.G. Hodnettb, and A. Senzanjec
aDepartment
of Lands and Water Resources Management, Midlands State University,
P. Bag 9055, Gweru, Zimbabwe
bCentre
cDepartment
for Ecology and Hydrology, Wallingford, Oxon OX10 8BB, UK
of Soil Science and Agricultural Engineering, University of Zimbabwe.
P.O.Box MP 167, Mount Pleasant, Zimbabwe.
Corresponding author: ftmugabe@yahoo.co.uk
ABSTRACT
This paper examines the effect of soil type and rainfall distribution on soil moisture and
runoff generation in the 5.9 km2 Mutangi catchment located in semi-arid Zimbabwe. Rainfall,
soil moisture and runoff were measured during the 1999/00 and 2000/01 rainy seasons, in
which 755 mm and 615 mm of rainfall were received. 58% and 69% of these rainfall totals
were recorded in February. The total catchment runoff was 102 mm and 63 mm, of which
52% and 49% was recorded over 6 and 4 days of 2000 and 2001 respectively. If there is a
sequence of daily events that completely fills the storage available in both the sodic and
transitional soils, and begins to saturate the ridge soils, subsequent events can produce very
large amounts of runoff (>50% of the daily rainfall). The occurrence of such runoff events
depends very heavily on the distribution of rainfall. Dry spells between rain events create
storage thereby reducing the risk of runoff from the next events. The sodic soils along the
stream channels appear to generate most of the runoff because of their small capacity to store
water before saturation. The ridge soils are coarse sands, with a large capacity to store
rainfall. The transitional (slope) soils have an intermediate capacity to store water.
Keywords: Semi-arid, rainfall distribution, soil moisture and runoff.
1
2.0
INTRODUCTION
The rains in semi-arid Zimbabwe generally commence in October/November and end in late
March or April with the bulk of the rainfall in February and early March. There is often a mid
season dry spell, usually in January. Such a rainfall distribution is likely to have an effect on
runoff generation because the infiltration capacity of a soil depends on antecedent soil
moisture content. Apart from antecedent soil moisture, runoff generation also depends on
rainfall quantity, intensity and duration, permeability of the soils and catchment relief and
geometry (Dubrueil, 1985,1986).
In semi-arid environments, the main process of runoff generation is usually Hortonian
overland flow (HOF) (Sandstrom, 1997), although saturation-excess overland flow (SOF)
may also occur. HOF occurs when rainfall intensity exceeds the infiltration capacity of the
soil (Horton, 1933). Saturation-excess overland flow occurs when the water table rises to the
ground surface and prevents infiltration, as there is no longer any storage capacity. In the
areas where it occurs, all the rainfall runs off. This process normally occurs where the soils
are close to saturation for much of the wet season (usually because the water table is close to
the surface), or in areas where there is an impending layer in the profile, and dry season soil
moisture deficits are small. The generation of runoff in the semi-arid areas exhibits both
similarities and differences to that found in humid temperate areas (Sandstrom, 1997). It is
spatially and temporally highly variable and closely related to the spotty nature of rainfall.
There are a number of runoff measurement stations in semi-arid Zimbabwe but these are
mainly in relatively large catchments. Only three small catchments have been extensively
studied in a semi-arid area in Zimbabwe. This data reports the studies that have been carried
out in one of these small catchments (Mutangi catchment) to determine the effect of rainfall
distribution and soil type on the amount of runoff generated in two seasons. These results are
2
presented and interpreted with soil moisture data to determine the effect of runoff generation
processes and the role of rainfall distribution and antecedent soil moisture in contributing to
catchment runoff.
2.0
STUDY SITE AND METHODS
2.1
Site, climate, soils and land use
Mutangi catchment lies in natural region V of Zimbabwe that is semi-arid. The rainfall
average is 550 mm with an inter-annual average of 200-1000 mm (Vincent and Thomas,
1960). The natural regions are a classification of the agricultural potential of the country, from
natural region I, which has the highest agricultural potential (the high altitude wet areas) to
natural region V, with the lowest agricultural potential, due to low rainfall and a high risk of
drought spells during the growing season. The monthly average daily maximum temperatures
vary between about 20oC and 32oC in winter and summer respectively. Minimum
temperatures are approximately 5oC in winter and 20oC in winter.
Mutangi catchment is about 5 km long and between 1 and 2 km wide along most of this
length, with an area of 5.9 km2. The altitude ranges from 878 m above sea level (masl) at the
catchment outlet (dam) to 930 masl at the head of the catchment. The average catchment
gradient is 0.8%. Four stream channels drain the mainly cultivated eastern end of the
catchment, converge within a short distance to join the main sand bed channel.
There are four soil types in the catchment, three of which follow a clearly defined
toposequence, grading from one to the other down the slope (Muzuwa and Gotosa, 1999). The
highest lying members of the catena are shallow to moderately permeable deep coarse sands
merging to coarse loamy sands, into soft weathering gneiss. Small areas of stony phase
3
(lithosols) occur associated with rock outcrops (not always on the crest). These soils are
described as having good permeability. These soils are infertile and have a very low available
water capacity. The lowest lying members of the catenary sequence, along the stream
channels, are strongly sodic and characterised by mopane schrub, which is often very stunted.
These soils are generally duplex in character, with course-grained loamy sands overlying
calcareous and saline sandy clay loams and sandy clays, below which is soft weathering
gneiss. The sandy clay lower horizons have an exceptionally high bulk density (~2.2 Mg m -3)
and permeability is correspondingly very poor.
Between the sandy, permeable ridge crest and the sodic soils, the soil properties are dependent
on slope position, with a gradual transition down slope to the highly sodic conditions. The
mid slope soils also have a duplex character; the topsoil properties are similar to those of the
ridge crest soils, but the lower horizons (below about 60 cm) are more clayey (clay content 15
– 22%) and beginning to show sodicity.
2.2
Experimental design
2.2.1 Rainfall
Seven manually read rain gauges and one automatic gauge were installed in the catchment
(Figure 1). The manual gauges were read every morning between 0600 and 0800 hours. The
automatic gauge was located near the centre of the catchment and is part of an automatic
weather station (AWS) and data is logged hourly.
4
Figure 1:
Map of Mutangi micro-catchment showing features and hydrological
instrumentation. Key: Rg- raingauge, A1 – C1 – soil moisture access tubes, BH – borehole, W
– shallow V weir, WLR – water level recorder, Ch1 – channel number, 880 – contour (m)
2.2.2 Soil moisture
Two transects of neutron probe access tubes were installed (figure 1). The first transect (A)
runs along the low ridge that forms the divide between subcatchments 1 and 2, starting close
to the confluence of the two channels. The second transect (B) starts on the ridge close to
transect A and runs at right angles to it, across channel 1 and onto the ridge that forms the
northern boundary of the catchment. This samples the catenary sequence from the sandy soils
on the ridge, through the transitional soils on the mid-slope, to the more clayey sodic soils,
which are the lowest members of the sequence, close to the streams. The access tubes in both
transects were located on different soil types, different land uses and different crops. The
tubes were installed to the maximum depth possible, which was usually limited by stones or
weathered rock that was too hard to penetrate. Depths ranged from 1 m to 2.5 m. A neutron
probe (IH II) was used to measure soil moisture at approximately weekly intervals. Readings
were taken at 0.1, 0.2, and 0.3 m and then at 0.2 m intervals for the remainder of the depth of
the tube using a counting rate of 16 s.
5
2.2.3 Catchment runoff
An automatic water level recorder was installed on a platform extending from the dam wall to
reach sufficiently deep water to record the lowest likely water levels. The recorder used a
float in a stilling well and was of the shaft encoder type; with an accuracy of 1 mm. Water
levels were recorded at intervals of ten minutes.
The rectangular weirs were constructed on the spillway to enable measurement of both high
and low flows. The water level data was used to calculate the runoff through the rectangular
weirs. Total catchment runoff was calculated from storage increases in the dam, and the
gauged spillway discharge.
3.0
RESULTS
3.1
Rainfall
Figure 2 shows the monthly rainfall totals for the 1999/00 and 2000/01 seasons. The rainy
season commenced on November 19 in 1999 and October 24 in 2000. Seasonal rainfall totals
were 755 and 615 mm respectively. February was the wettest month in both seasons, with
about 34% of the annual rainfall in both seasons. In the 1999/00 and 2000/01 respectively,
there were a total of 57 and 46 rain days and the maximum rainfalls of 72 mm and 65 mm.
Rainfall distribution was similar in both years with almost half the season’s rainfall falling on
just 6 or 7 days.
6
300
Monthly rainfall (mm)
250
200
150
100
50
0
Oct
Nov
Dec
Jan
Feb
Mar
Apr
1999/00 season
May
Jun
Jul
Aug
Sept
2000/01 season
Figure 2: Monthly rainfall distribution in the 1999/00 and 2000/01 seasons.
The spatial variability of rainfall, represented by the CV of the mean rainfall recorded at the
seven rain gauges, decreased with increasing rainfall. For events less than 10 mm, the CV
ranged from 8% to 96% (typically 30%), but for the longer events the consistency of rainfall
over the catchment was notable. For the 13 mean daily rainfall totals that exceeded 30 mm,
the CV ranged from 10 to 26%, with 11 CVs in the range from 10 to 17%.
3.2
Soil moisture
Figure 3 shows the extreme wettest and driest profiles from the access tubes at different
positions in the catenary sequence, from sodic soil close to the channels (A&B) through
transitional soils (C, D, &E) to coarse sandy soils on the ridge tops (F). The wettest data are
typically from late February to early March, and the driest from September to October, at the
end of the dry season. The differences in soil properties are clearly reflected in the shape of
the water content (Figure 3), and the seasonal changes in water content (Figure 4).
7
Depth (cm)
0.0
0
0.1
Vol. water content
0.2
0.3
50
50
100
100
150
150
26 Oct 99
10 Feb 00
0.0
0.1
0.2
0.3
0
Depth (cm)
0.0
0
A
200
0.4
0.0
0
50
100
100
150
150
26 Oct 99
10 Feb 00
0.1
0.2
0.3
200
0.0
0
0.4
E
50
50
100
100
150
150
200
Vol. water content
0.2
0.3
0.4
B
26 Oct 99
01 Mar 00
0.1
0.2
0.3
0.4
D
50
0.0
0
0.1
200
C
200
Depth (cm)
0.4
12 Oct 00
03 Mar 01
200
12 Oct 00
03 Mar 00
0.1
0.2
0.3
0.4
F
12 Oct 00
03 Mar 00
Figure 3: Wettest and driest water content profiles. See text for soil type and crop cover
description.
8
400
Sodic bare
Fallow sand
Mopani bush
Clay cultivated
0 - 1.10 soil moisture storage (mm)
350
300
250
200
150
100
50
0
28-Aug-99
06-Dec-99
15-Mar-00
23-Jun-00
01-Oct-00
09-Jan-01
19-Apr-01
28-Jul-01
Figure 4: Soil moisture storage at four access tubes representing different land use practices.
Except for the fallow, ridge top soils, there was not much difference between soils moisture
between the two seasons. These soil still had some capacity to store water during the 2000/01
season suggesting runoff from this land use type was less likely than the 1999/00 season,
when storage reached higher levels (water levels closer to the surface).
The fallow sodic profiles (A) showed very small seasonal changes in the water content,
limited to only the upper 0.4 m of the profile. These soils have a very high bulk density,
which limits their available water capacity. There was also little vegetation cover to abstract
water, and as a result the profile has only a limited capacity to store rainfall before runoff
occurs. The sodic soils under Mopane schrub near the stream channel (B) showed more
seasonal changes of water content than (A).
9
The cultivated transitional soils (C, D & E) have a duplex character, with sand overlying a
more clay rich (22%) layer with a high bulk density (2.1 Mg m-3) and a porosity of about
20%. This is reflected in the maximum water contents that were measured, and in the small
range of water content change observed in the lower profile. At site C, large water content
changes were limited to the upper 0.4 m. At site D there were changes of water content down
to a depth of 1.6 m. The upper slope site (E) is transitional, but has a greater depth of sandy
soil. The wettest profile occurred when the water table was almost at the surface. The largest
changes of storage in the profile occurred as the water table level fell at the end of the wet
season.
(F) is a ridge top site, entirely on deep coarse sand. The seasonal water content change is
much larger than at any of the measured sites. As at the other sites, the wettest condition
occurred when the profile was saturated almost to the soil surface, on 3 March 2000. The
wettest conditions at this site in 2001 were recorded in March, but the storage was much
lower, and the highest water table level observed was 80 cm below ground level (bgl).
Saturation excess runoff only occur from these sites once the storage available has been
entirely filled. These ridge top sites have a large amount of storage available in the sandy
profile, and will only generate runoff in very wet periods.
Clear differences in the seasonal storage changes in each of the soil type are reflected, with
the sodic soils showing the smallest seasonal change and the sandy, ridge soil showing by far
the largest change, almost twice that of the intermediate, transitional soils. In most cases, the
wet season measurements were made when the soils were saturated , so that these figures give
a good indication of the amount of storage available in the soils at the end of the dry season.
10
Table 1: Average maximum seasonal water content changes for each of the soil types.
Soil type
Max storage change to 1.1m
Max storage change to max.
depth (mm)
depth of tubes (mm)
Sandy (ridge)
266.5
278.5
Transitional (slope)
118.8
144.9
Sodic (valley bottom)
82.1
86.4
3.3
Runoff
Figure 4 shows the cumulative total runoff from the catchment, in the 1999/00 (a) and
2000/01 (b) seasons respectively. The results show that the total runoff in the 1999/00 and the
2000/01 seasons was 102.2 and 63.0, or 13.5% and 10.3% of the total rainfall received in
these two seasons respectively. In both seasons there was a very large amount of runoff in late
February. Most of the runoff occurred in discrete rapid events, and there was little or no
baseflow, particularly following the early wet season events. Later in the season, there was
some indication that flows continued for longer, but the amounts were small and the flow
could have been generated by small rainfall events falling on the still saturated sand within
the channel bed, and surrounding contributing areas.
11
200
100
180
total runoff
160
rainfall
80
Cumaulative runoff (mm)
140
120
60
100
80
40
60
40
20
20
0
1-Nov-99
0
1-Dec-99
31-Dec-99
30-Jan-00
29-Feb-00
30-Mar-00
29-Apr-00
29-May-00
Date
200
100
180
total runoff
160
rainfall
80
120
60
100
80
Daily rainfall (mm)
Cumulative runoff (mm)
140
40
60
40
20
20
0
2-Oct-00
0
1-Nov-00
1-Dec-00
31-Dec-00
30-Jan-01
1-Mar-01
31-Mar-01
30-Apr-01
30-May-01
Date
Figure 4: Cumulative total runoff and daily rainfall during the 1999/00 and 2000/01 seasons.
12
4.0
DISCUSSION
The catenary sequence is very important in influencing the distribution of shallow
groundwater and the hydrological response of the catchment. On the ridge crest, the soils are
highly permeable but overlie almost impermeable weathered rock at only a few metres depth.
In the wet season, infiltration will pond on the impermeable horizon and move laterally, down
slope. This lateral flow is impeded by the dense, low permeable sandy clay horizons of the
transitional mid-slope soils, causing shallow groundwater to approach the surface. The
transitional soils have less storage available, and may saturate due to rainfall inputs alone.
During the wet period in February 2000, when a large amount of runoff was generated, there
were large increases in soil moisture storage in the sandy soils on the ridge. Between 10
February and 3 March (a few days after the main rainfall events), soil moisture storage
increased by between 48 mm and 164 mm (mean 123 mm) while on the sodic soils with
mopane shrubs, the storage increase was only 2 mm. The sandy, ridge soils were able to
absorb significant amounts of rainfall in this period, but the sodic soils absorbed almost none,
indicating that most of rainfall was lost as runoff from such areas. The sodic soils under
mopane wetted up in the early part of the wet season and dried down again in April and May,
but during the dry season, when mopane had shed off their leaves, further drying was very
limited. The fallow areas on the sandy ridge soils showed increases of moisture storage up to
304 mm during the wet season. Most of the rainfall is retained because the high storage
capacity, although locally HOF was occasionally observed during intense rainfall. Very high
rates of water loss during dry spells in the wet season (>10 mm d-1) suggest that at some
locations, some of the stored water is lost by down slope drainage, as well as through soil
evaporation and transpiration by grass. The highest amounts of runoff and highest runoff
percentages occurred in February, when soil moisture storage was at its maximum.
13
Rainfall distribution strongly affected the amount of runoff generated by given rainfall events.
Most of the runoff occurred in February in both years when rainfall events were closely
spaced. Between 22 and 28 February 2000, the total rainfall was 111 mm and in the same
period, 53 mm of runoff generated (48% of the total rainfall). 52% of the total runoff for the
season was generated within these six days. A similar trend was observed between 22 and 25
February 2001, when 31 mm of runoff (49% of the seasonal total) was generated from a total
rainfall of 101 mm. In both years, the closely spaced series of rainfall events had brought the
water table close to the soil surface over much of the catchment, resulting in little capacity for
the soil to store more water. Runoff was then generated as saturation-excess overland flow
(SOF). It is also of note that between 22 and 28 February 2000, the rainfall intensity did not
exceed 10 mm h-1 and only exceeded 5 mm h-1 in one occasion. These data indicate that HOF
is unlikely to have been an important process in this period.
On 4 May 2000, the mean rainfall over the catchment was 55 mm (CV 14%), which included
the highest hourly rainfall (28.9 mm h-1) recorded during the study. In contrast to the events in
February, this event produced only 4.5 mm of runoff (8% of rainfall). In the six weeks
preceding 5 May 2000 there had been almost no rainfall, and water levels had fallen as a
result of drainage, soil evaporation and uptake by vegetation. As a result, the soil had
significant capacity to store water and little runoff was generated, despite the high rainfall
intensity.
The 1999 wet season started with three consecutive days of rainfall (34 mm, 12 mm and 70
mm), the last day of which included an intensity of 27 mm h-1. This event produced 14 mm of
runoff (20% of rainfall). The two main factors that may have contributed to this relatively
14
large percentage early in the wet season are: (i) when the rains commence, the soil surface is
very flat and prone to capping in heavy rains. In addition, there is little vegetation cover to
protect the soil surface from slaking action of raindrop impact. These factors increase the
likelihood of HOF, which was observed on the sandy ridge soils in November 2000; (ii) the
46 mm of rainfall on the days before the 70 mm event would have filled up the storage in
some of the areas with sodic soils, leading to the generation of SOF in these areas.
Cultivation only takes place after the first rains, so that later in the wet season, after ploughing
has taken place, the greater roughness of the soil surface will increase the acceptance of
rainfall. Later in the season, vegetation cover also protects the soil surface. These factors, and
the availability of storage capacity in the soil following a dry period, may explain the lower
percentage runoff after the high intensity 55 mm event on the 6th of May 2000.
4.0
CONCLUSION
The dominant process of runoff generation in this study was SOF. Rainfall intensities were
generally low and HOF only appears to have played a role in runoff events in the early wet
season, before ploughing had created a rough soil surface, and annual vegetation cover
(annual crops, grasses and weeds) protected the soil from the raindrop impact. This is
unusual, as HOF is generally regarded as the dominant runoff generation process in semi-arid
regions. SOF is more important because of the very shallow depth of the weathered zone
(often less than 2.5 m) above the gneiss bedrock, and because the presence of the clayey sodic
soils, which are common in this part of Zimbabwe.
The distribution of rainfall is critical to runoff generation; if the last event of the closely
spaced series of events in late February (in both years) had not occurred, the seasonal rainfall
15
totals would have been about 15% lower, but the seasonal runoff totals would have been
halved. If events are well spaced, evaporation and drainage create storage capacity to store
water, but a series of closely spaced rainfall events will fill the storage to the surface such that
the following events cause large amounts of SOF. The soil water data show that the area
contributing to SOF increases progressively as the catchment wetted up, starting with the
sodic soils and progressing through the transitional soils to the sandy ridge soils, which have
the largest storage capacity. The variation in the size of the saturated contributing area during
rainfall events and its contraction due to evapotranspiration and redistribution agrees with
previous studies (Hewlett, 1961; Hewlett and Hibbert, 1963; Lane et a.l, 1978).
ACKNOWLEDGMENT
This publication is an output from a project funded by the UK Department of International
Development (DFID) for the benefit of developing countries. The views expressed are not
necessarily those of DFID. The research was partially supported by an African Doctoral
Fellowship provided by START and the Pan-African Committee for START. The authors
would particularly like to thank the people of Mutangi for their assistance and participation in
the research.
REFERENCES
Dubrueil, P.L. 1985. Review of field observations of runoff generation in the tropics. Journal
of Hydrololgy 80: 237-264
Dubrueil , P.L. 1986. Review of relationships between geophysical factors and hydrological
characteristics in the tropics. Journal of Hydrology 87: 201-222
16
Hewlett, J.D. 1961. Some ideas about storm runoff and baseflow. In: Process studies in hillslope
hydrology, ed. M.G. Anderson and T.P. Burt. John Wiley and Sons, England, pp 314.
Hewlett, J.D and Hebbert, A. 1963. Moisture and energy conditions on a sloping soil mass during
drainage. Journal of Geophysical research 66: 1081 – 1087.
Horton, R.E. 1933. The role of infiltration in the hydrological cycle. Transactions of the American
Geophysical Union 14: 446-460.
Lane, L.J., Diskin, M.H., Wallace, D.E. and Dixon, R.M. 1978. Partial area response on small semiarid watersheds. Water resources bulletin 14(5): 1143-1158.
Muzuva, J. and J. Gotosa. 1999. Field report on Geology and Soil surveys Mutangi MicroCatchment, Masvingo, Zimbabwe.
Sandstrom, K. 1997. Ephemeral rivers in the tropics. Hydrological processes and water resources
management: A review and pathfinder. Research Programme on environmental policy and
society. Institute of Tema, Linkopint University. Research report No. 8 from EPOS.
Vincent, V. and Thomas, R.G. 1960. Agricultural survey of Southern Rhodesia. Part 1. AgroEcological survey. Salisbury: Government Printers.
17