5.

WATER CONSERVATION
This chapter deals with methods to increase the amount of water stored in the
soil profile by trapping or holding rain where it falls, or where there is some
small movement as surface run-off. Schemes which are more concerned with
catching and storing run-off for use later or at a distance from the storage are
discussed in Chapter 6. The division of methods into these two types is
arbitrary, so cross-reference signposts have been used liberally. There are
also links with the efficiency of water use discussed in Section 4.2 on
Biological Soil Conservation.

Some writers have used the terms internal and external catchments. Internal
catchments are where the run-off producing area is within the cropped area
and are discussed in Chapter 5; external catchments are run-off areas outside
the cropped area and are discussed in Chapter 6.

5.1 PRINCIPLES
5.1.1 Choice of Method

There is no simple way of classifying methods of water conservation. One

suggestion is to do it by comparing rainfall with crop requirements (Narayana
and Ram Babu 1985), giving three conditions:

i. Where precipitation is less than crop requirements; here the strategy

includes land treatment to increase run-off onto cropped areas,
fallowing for water conservation, and the use of drought- tolerant crops
with suitable management practices.

ii. Where precipitation is equal to crop requirements; here the strategy is

local conservation of precipitation, maximizing storage within the soil
profile, and storage of excess run-off for subsequent use.

iii. Where precipitation is in excess of crop requirements; in this case the

strategies are to reduce rainfall erosion, to drain surplus run-off and
store it for subsequent use.

The weakness of this approach is that the main feature of rainfall in semi-arid
regions is that it is very erratic and completely unpredic- table. There can be
wide variations of moisture shortage and surplus, both within and between
seasons. A drought year whose total rain is well below the long-term average
may still include periods of excessive rain and flooding, while a high rainfall
season may include periods of drought.

This makes the choice of method difficult, because the desired objective may
change from one season to another. In a dry area it may be sensible to
increase surface storage to improve crop yield in most years, but in a wet year
this could cause waterlogging and reduce the yield. On the other hand, a
drainage system may have the objective of increasing the run-off but also
have the undesired effect of exaggerating the effect of a drought. It is
therefore not practical either to classify methods according to average
conditions, or to design strategies based on averages. The art or science of
water management is to reduce the problems caused by non- average events
of flood and drought.

Sometimes it may be passible to have dual purpose methods which can be

changed mid-season, for example by opening up the ends of contour bunds to
shed surplus water after a wet start to the season, or to block outlets for the
opposite effect. But not many methods allow this flexibility, and we showed in
Section 2.3.1 that little progress has been made in developing methods of
predicting the later part of the season on the basis of the earlier part.

In addition to the vagaries of rainfall, there are a host of other variables, the
soil, the land use, the farming system, and the social patterns. Throughout the
semi-arid regions of the world there is a wide range of indigenous systems
and methods, but experience has shown that they do not always transfer well
from one set of conditions to another. In a good discussion of this subject,
Pacey (1985 Chapter 6) points out that apart from the technical differences,
methods must also be compatible with local life styles, social systems, and
patterns of administration. Transferring what appear to be simple techniques
requires not only the dissemination of information but also the adaptation to
local conditions. In this Bulletin, when presenting information on available
methods, we have not tried to prescribe a detailed treatment for a particular
problem, but rather to set up a large array of possible methods, from which
the person on the spot can choose methods to test locally.

When choosing a method or technique we must be aware of the physical and

personal difficulties. Because of the variable rainfall we must expect a low
success rate. The odds will be slightly better for grass, trees, and shrubs than
for grain crops which are more demanding in their moisture requirements.
They need rain at the beginning of the growing season, then a number of
storms big enough to generate run-off and at reasonably regular intervals,
since run-off events too close together will be wasted if they exceed the
storage. The soil must therefore have sufficient soil moisture storage capacity
to keep the crop growing between run-off events. There will often be
difficulties in keeping stock away from improved areas. Wire fencing is
expensive, live hedges need time and care to establish them, and thorn
hedges need regular maintenance. Maintenance is also required on any
structural works like banks or furrows.

Personal constraints are that people naturally do not like to put a lot of effort
into schemes which have a low success rate, nor if their tenure is insecure. In
dry areas the people are frequently mobile or partly nomadic and so may not
live permanently near sites suitable for run-off schemes. There are many
former schemes which are now abandoned, and knowing more about the
reasons would help us plan new schemes.

5.1.2 Some Design Principles

The term "rainfall multipliers" describes methods where the run-off from an
uncultivated part of the land is diverted on to a cultivated part, thus giving it
the benefit of more water than it receives directly as rain, hence rainfall
multiplier.

The object of designing such methods is to get the best ratio of area yielding
run-off to the area receiving run-on. Because of the variations of soil and crop
requirement, and above all the variability of rainfall, this can never be a
straightforward mathematical calculation. A formula which can be used to
provide a starting point for field trials is given by Finkel (1985), together with
some empirical values for Turkana in Kenya, to be used in the formula:

Ratio = Catchment area = Crop water requirement - Design rainfall

Cultivated area Design rainfal x Run-off Coeft x Efficiency
· Crop water requirement is estimated from evaporation data with a crop
factor applied;

· Design rainfall is probably less than long-term average rainfall in order to

give a better crop in low rainfall years, while the excess can overflow in wet
years;

· Run-off coefficient is the percentage of rainfall which becomes run-off;

· Efficiency reflects the difference between rainfall distribution and the water
requirements of the crop.
Schemes with internal catchments require high energy inputs for surface
modification and so tend to be more common on crop land and in areas with
annual rainfall of 250 mm or more. The ratio of catchment area to cultivated
area is likely to be from 1:1 to 5:1. Schemes using external catchments may
have much higher ratios, for example up to 30:1 in the case of the run-off
farming in the Negev desert (Evenari et al. 1982). Such systems may be
practical at levels of rainfall down to less than 100 mm, which is lower than is
possible for in-field schemes for cropland.

Two examples of rainfall multiplier:

a) Internal catchment, ratio 2:1, design rainfall 250 mm

2 = 500 - 250               =  250

1     250 x 60% x 80%    120

b) External catchment, ratio 20:1, design rainfall 100 mm

20 = 500 - 100 =  400

    100 x 25% x
1     20
80%

The run-on is greater in b), so the soil moisture storage must be greater, or
run-off events better distributed.

In developed economies it is reasonable to look at the long-term returns from

schemes which require investment, but this is less helpful in semi-arid areas.
Even when long-term rainfall records are available, and probabilities can be
calculated, subsistence farmers or nomadic pastora- lists work to very short-
term plans. Several examples are reported from Kenya by Finkel (1984) when
soundly planned schemes were abandoned after poor results in the first year
to two.

5.1.3 The Effect of Scale

The size of catchment areas has a bearing on the yield of run-off. Under the
same hydrological conditions, small areas may have up to 50 percent,
whereas river basin run-off is only 5 percent (Stern 1979). There are two
reasons for this. Large catchment areas are more likely to have places where
there is temporary storage of run-off in surface depressions. Also small
catchment areas may yield run-off from short showers which do not produce
surface run-off from larger areas. The hydrological explanation is that small
areas have a shorter gathering time (i.e. time of concentra- tion) and so give
run-off from shorter storms (with higher intensities) than storms which need a
longer gathering time to produce run-off from larger catchment areas. Shanan
and Tadmor (1979) quote the following average run-off yield from 100 mm
annual rainfall:

Less than 0.02 ha 10-30 mm/ha/year

  3-10 ha 4-10 " " "
  300-500 ha 1"""

Procedures for estimating rate and amount of run-off are given in Hudson
(1981) and FAO (1981).

5.2 METHODS FOR CROP LAND

The techniques described in this section all involve some re-shaping of the
soil surface, and so require substantial inputs of energy. In many cases it is
not practical to use only hand labour, and animal-drawn imple- ments are
necessary, or tractors. Some of the methods show promise of increased crop
yields so a major feature of experimental trials is the search for cheap, simple,
low energy methods and machines. Figure 5.1 shows the main features of the
methods described in this Section.

5.2.1 Broad Bed and Furrow System (BBF)

The Broad Bed and Furrow system has been mainly developed at the
International Crops Research Institute for the Semi-arid Tropics (ICRISAT) in
India (Krantz 1981, Pathak et al. 1985). A comprehensive research
programme was carried out on-station for eight years before being taken for
on-farm adaptive research at Tadthanapalle in Medac district (ICRISAT
Annual Reports 1975-1984). It is a modern version of the very old concept of
encouraging controlled surface drainage by forming the soil surface into beds.
In medieval times in Britain this was used for improving pastures and called
"rigg and furrow"; it has also been used in North America (Phillips 1963) and
in Central Africa (Cormack 1954). A variation known as the camber-bed
system was used in Kenya (Robinson et al. 1955). An experimental layout at
ICRISAT is shown in Plate 5.1 and the bed cross-section in Plate 5.2.

The recommended ICRISAT system consists of broad beds about 100 cm

wide separated by sunken furrows about 50 cm wide. The preferred slope
along the furrow is between 0.4 and 0.8 percent on vertisols. Two, three, or
four rows of crop can be grown on the broad bed, and the bed width and crop
geometry can be varied to suit the cultivation and planting equipment. Some
examples are shown in Figure 5.2.

In India the system has been used mainly on deep vertisols (heavy black clay
soils sometimes called cotton soils); wide beds are used on a gentle grade
and they are formed by ox-drawn wheeled tool carriers (Plate 5.3). There are
extensions of this work on similar soils in Ethiopia (ILCA 1985) with an
interesting development of a very simple ridging implement shown in Plate
5.4. Another variation using smaller ridges is being developed by the
Agricultural Research Centre for the Semi-arid Tropics (CPATSA) of the
Brazilian Enterprise for Agricultural Research (EMBRAPA) located at
Petrolina, Brazil (Lal 1985 and 1986); it is illustrated in Plate 5.18.

All these schemes have, in varying proportions, the following objectives:

i. to encourage moisture storage in the soil profile. Deep vertisols may

have soil moisture storage up to 250 mm, which is sufficient to support
plants through mid-season or late-season spells of drought. The
possibility is also increased of double cropping by means of inter-
cropping or sequential cropping. The large water storage capacity of the
soil supports growth more easily during the subsequent dry but cooler
post-rainy season.

ii. to dispose safely of surplus surface run-off without causing erosion.

Plate 5.5 shows the flooding which can occur on vertisols.

iii. to provide a better drained and more easily cultivated soil in the beds.
There is only a narrow range of moisture conditions during which the
soil can be efficiently tilled or planted, and timeliness is a key factor.
Only about 20 percent of the deep vertisols in India are cropped during
the rains, mainly because of poor workability when wet. The situation is
similar in Ethiopia (Plate 5.6). If a crop can be established during the
early rains, the profile is usually near saturation only for short periods
during the latter half of the season, water is more efficiently utilized, and
there is less need for run-off collection and storage. The possibility is
also increased of double cropping by means of inter-cropping or
sequential cropping (Kampen 1979 and 1981). Tillage of the raised
beds may be possible before the rains, introducing the possibility of dry
seeding ahead of the rains in areas where the start of the rainy season
is fairly reliable, and there is a good chance of follow-up rains to ensure
the establishment of the germinating crop. The difficulty of preparing a
 seed bed during the dry season in these hard clay soils has been
greatly improved by the use of broad beds and animal-drawn
equipment.

iv. the possibility of the re-use of run-off stored in small tanks (Plate
5.1). Small amounts of life-saving irrigation applications can be very
effective in dry spells during the rains, particularly on soils with lower
storage capacity than the deep vertisols.

The BBF system is particularly suitable for the vertisols. The technique works
best on deep black soils in areas with dependable rainfall averaging 750 mm
or more. It has not been as productive in areas of less dependable rainfall, or
on alfisols or shallower black soils - although in the latter cases more
productivity is achieved than with traditional farm- ing methods. Other
methods, with more emphasis on storage and irrigation within a package
which includes BBF, are more likely to be viable for the alfisols (Ryan et al.
1979). It is also stressed throughout the ICRISAT research that the BBF
system should not be considered in isolation, but only as part of an improved
farming systems package.

An important component of the system is an ox-drawn wheeled tool bar, which

can be used with ridgers to form the raised bed and also later for carrying
precision seeders or planters (Plate 5.7). The tool carrier is thus used for the
initial forming of the beds, the subsequent annual reshaping, and for all tillage,
planting, and inter-row cultivation. Even the cost of the simplest factory made
tool bar is beyond the resources of most peasant farmers, but the really cheap
wooden ridger shown in Plate 5.4 shows promise.

5.2.2 Ridging and Tied Ridging

This method is also known in the USA as furrow blocking, furrow damming,
furrow diking, and basin listing. The principle is to increase surface storage by
first making ridges and furrows, then damming the furrows with small mounds,
or ties (see Plate 4.14). Graded ridges alone will usually lead to an increase of
surface run-off compared with flat planting, while tied ridges will decrease the
run-off and increase the storage. In different seasons either of these two
effects may be prefer- able. And so the design question is when to go for
drainage and when for storage. An aid to designing what is best in the long-
term by modelling has been developed (Morin et al. 1984), but designing the
system which will give the best effect in the long term will not prevent the
system giving the opposite effect from that which is required in a particular
season. The possibility has been suggested by Ahn (1977) of hedging the bet
by tying alternate furrows. This is not a very satisfactory solution. One
argument is that this would reduce the amount of damage by too much run-off
or too much retention, but one could equally well argue that it will ensure that
half of the land will always be under the wrong treatment.

An interesting possibility is to use tied ridging in connection with sprinkler

irrigation to allow higher application rates at low pressure (Stuart et al. 1981).
Tied ridging is usually associated with mechanized farming. There have been
some attempts at achieving it with ox-drawn implements, but the system really
needs high draught for speed and preci- sion, which is required if the ridges
are to be re-ridged or split in subsequent years.

Either ridging alone or tied ridging has occasionally been practised using hand
labour, but the high labour requirement usually makes this unpopular with
subsistence farmers. Anyway hand-made ridges are usually less efficient.
They are more likely to depart from a true contour and to have variations in
the height of the ridge, both of which will increase the risk of overtopping. A
variation in the theme, which is suitable for hand work, has been tried on a
limited scale in Burkina Faso and is called tied-mounding shown in Plate 5.8.

i. Varying evidence of the results of tied ridging

There is an extensive literature reporting trials of tied ridging in many

countries. A few of the reports indicate problems and failures but the great
majority claim such outstanding success for the system that one wonders why
the system has not been more widely used. A possible solution to this
conundrum may be that the inconsistency, and unreliability of good results,
prevents it becoming more widely adopted.

The experience in Africa has been well summarized by El-Swaify et al. (1985)
as follows:

"Under certain circumstances the system has been beneficial not only for
reducing run-off and soil loss, but also for increasing crop yield (Lawes 1961,
1963; El-Swaify 1983; Dagg and McCartney 1968). However, during high
rainfall years or in years when relatively long periods within the rainy season
are very wet, significantly lower yields were reported from systems with tied
ridges than from graded systems which disallowed surface ponding ef water
(Lawes 1963; Dagg and McCartney 1968). Under such conditions tied ridging
enhanced waterlogging, developed anaerobic conditions in the rooting zone,
excessive fertilizer leaching, and water table rise in lower slope areas."
In other countries there are conflicting reports, with a majority of successes.
McCartney et al. (1971) reported that tied ridging in Tanzania gave higher
maize yields not only in low but in high rainfall years as well. However, reports
of success are more common in low rainfall years, for example Njihia (1979)
reported from Katumani in Kenya that tied ridging resulted in the production of
a crop of maize in low rainfall years when flat-planted crops gave no yield.
Jones (1985) goes on to say:

"Similarly on a sandy soil at Lusitu in the Zambesi valley, tied ridges increased
mean crop yields (maize, sorghum, and millet) over those on flat land by 168,
159 and 16 percent under seasonal rain- falls of 587, 623, and 724 mm
(Honish, 1973); and on vertisols at Big Bend, Swaziland, mean increases for
maize, cotton, and sorghum were 64 percent in a year of 508 mm, and 308
percent in a year of 310 mm (Warwick 1979, 1980). Clearly responses can be
dramatic, but recent work under very harsh conditions in Botswana has shown
that there may be also negative effects. Higher soil temperatures within the
ridge can be detrimental to seed germination, and where showers are light the
penetration of moisture into the soil may be shallower than that in the flat soil
(DLFRS, 1984)."

Another report from Botswana indicates a positive response to sorghum yields

from tied ridging in both a very dry season (1972/73) and a very much wetter
one (1973/74) (DLFRS, 1973; 1974).

In the Negev desert in Israel, Rawitz et al. (1983) reported that tied ridging
greatly increased the effi- ciency of a fallow before cotton, but that when the
system was tried for wheat crops it was less successful because of
waterlogging in the basins (Plate 5.9). The use of wide beds with untied
furrows was therefore pre- ferred and gave increased yields, but the climate of
the Negev is special since it has winter rainfall and a very low mean annual
rainfall.

From India also, broad ridges gave better yields of sorghum than either flat
planting or narrow ridges(ICRISAT 1976) and a similar result was reported for
sorghum and castor in Gujarat by Brahmbatt and Patel (1983). In Texas, USA,
Clark and Jones (1981) reported substantial increases in sorghum yield from
1420 kg/ha to 1650 kg/ha for tied ridging compared with flat planting. In
another American study in Texas diked furrows were found to be effective at
holding and absorbing high intensity storms (Billbro and Hudspeth 1979).
Similar successful applications of tied ridging in the Great Plains of the USA
are also reported by Stuart et al. (1985), and again one is inclined to wonder
why, in the face of so much positive experimental evidence, one does not find
the whole of Texas and the whole of Tanzania covered with tied ridging.

One possible reason for the low adoption of the system is the temperature
effect reported by Jones. Another is the danger of waterlogging in wet years.
This can be worsened when there is compaction of the furrow bottom by
tractor traffic, and one must also consider whether the machinery and
implements are as efficient as the reports would suggest.

ii. Conditions where tied ridging is suitable

There is a danger of soil erosion if the ridges are overtopped and break so
that the water temporarily stored in the depressions is suddenly released. This
will not happen if the combination of surface storage plus the amount which
infiltrates into the soil surface is less than the storm rainfall. This implies a
high value of soil storage, usually deep soils with good infiltration and
permeability. In some systems the infiltration is increased either by mulching
in the furrow bottoms or by subsoiling or cultivating.

Three safety back-ups are required to minimize the risk of damage by erosion:

· the furrows should be on a gentle grade to assist run-off if the ties fail;

· the ties should be lower in height than the ridges so that the ties fail
along the furrows before the ridges fail down the slope;

· there should be a back-up system of conventional graded channel

terraces to prevent damage if the ridges do overtop or fail.

iii. Implements for building tied ridges

Making an implement to form the ridges is straightforward, it is interrupting the

ridging process to leave a tie that is difficult. Possibilities are:

· intermittent lifting by hand if the ridger is pulled by tractor or by oxen

(Plate 5.10);

· automatic lifting devices based on an eccentric wheel (Plate 5.11);

· intermittent hydraulic lift either manual or triggered by rotation of

tractor wheels.
The design of implements in the USA is described by Lyle and Dixon (1977),
and in the USSR by Tregubov (1981). An earlier review by Boa is useful but
out of print (FAO 1966).

iv. Simpler systems for non-mechanized farming

The construction of tied ridges is mainly associated with mechanized farming

because of the high labour requirement, but it is quite possible to do it by hand
labour. Plate 5.12 shows an example from Ethiopia where the ridges and ties
are made by piling up lines of sorghum residues and roots and soil.

5.2.3 Conservation Bench Terraces (CBT) (also known as Zingg terrace,

and flat channel terrace)

This is another type of rainfall multiplier, using part of the land surface as a
catchment to provide additional run-off onto level terraces on which crops are
grown. The method is particularly appropriate for large- scale mechanized
farming such as the wheat/sorghum farmlands of the south- west of the USA,
where the method was pioneered by Austin W. Zingg in 1955. Experimental
results were first reported in 1959 (Zingg and Hauser) and after detailed
studies at Bushlands Experiment Station in Texas, and Hayes in Kansas, a
technical and economic evaluation was made by Hauser and Cox (1962). This
led to extensive trials in the six western States with low rainfall, and these
were reported in detail (ASAE 1968). CBTs were compared with the
conventional practice of level terraces (i.e. level along the length, but the
original slope is left between terraces), and also with all over bench terracing
(Figure 5.3). There is a soil erosion hazard during high-intensity summer
storms, but CBTs were as effective at controlling erosion as the other two
practices, and more effective at reducing the overall run-off. The data from
these trials provided general guidelines on the method, but standard designs
should be avoided because of the wide variation in conditions of soil, rainfall
and farming system. The best way of applying the system in a particular
situation should always be investi- gated locally.

The features required for CBTs are:

· Gentle slopes of 0.5 - 1.5 percent are most suitable although the
system has been used up to 6 percent in North Dakota. As with all
terracing, a steeper slope requires more earth moving.

· A deep soil is required, both to provide sufficient soil moisture storage,

and also to lessen the effect of cutting during the con- struction of the
terraces. Several USA workers report a reduction of yield from the
disturbance of the soil during terracing for up to six years. Some even
went to the length of stockpiling the topsoil and replacing it after
levelling. Good permeability is also required so that the contained flood
water can be absorbed quickly.

· Smooth slopes are an advantage where large mechanized farming can

be made more convenient by constructing all the terraces parallel and of
equal width.

· Precise levelling of the bench terrace is important to ensure uniform

build up of soil moisture.

· If there is a risk that run-off from the catchment area will be greater
than can be absorbed and stored on the terrace, there must be outlets
at the ends of the terrace, which can discharge into grass waterways or
other safe disposals. As with tied ridging, if there is going to be surface
run-off it must be along the terrace, not over the edge of the terrace and
down the slope with consequent risk of erosion. For CBTs a common
method is to make the level of the outlet at the end of the terrace one
half of the height of the bund on the edge of the terrace, ensuring a
freeboard of half the bund height.

· This is a high-cost system which has to be paid for by high yields and
so a high level of inputs is required, particularly seed, fertilizer, and crop
management. Adequate levels of fertilization are required to avoid this
being the constraint after the moisture limitation has been removed.

· The main design factors are the width of the levelled terrace and the
ratio of the area of the terrace to the catchment area. The width of
terrace is governed mainly by the machinery to be used. As the width of
planters and drills continues to increase, so does the need for wider
terraces, but this is restricted by the slope and the depth of soil. Typical
widths are from 10 metres on land of 5 - 6 percent, to 30 metres at 2
percent, and 50 metres or more at one percent. Jones (1981) reports
mini-terraces 9 m wide with a 1:1 ratio and different rotations on the
catchment and the terrace. Total yield was doubled on the terraces.

· The most usual practice is for the bench terrace to be level along its
length, but there have been a few trials with the terraces on a gentle
gradient of 1:400 to encourage safe disposal of excessive run-off.
 · The ratio of catchment to terrace may be from 0:1 (in which case the
whole land surface is converted to level bench terraces) but typical
values are 1:1 or 2:1. In general, the lower the mean annual rainfall the
larger will be the required catchment area to provide sufficient moisture
for the crop. Another consideration will be the type of rainfall since run-
off will only be produced from storms where the intensity exceeds the
infiltration on thecatchment area. The treatment of the catchment area
will also effect the amount of run-off. In very dry areas the catchment
may be left uncultivated to promote run-off, while in the south western
USA the usual practice is to have continuous cropping on the bench
terrace with a rotation on the catchment area. A common rotation on the
catchment is wheat/sorghum/fallow, giving two crops in three years.
Table 5.1 gives examples of the rainfall multiplier effect.

Results of the American experience show that the method is suitable and
economic for the high-input large-scale mechanized farming of the region. The
costs of land preparation can be expected to be recovered in about 10 years.
Complete bench terracing of the whole land surface some- times gives the
greatest cash return since the whole of the land is cropped, but the cost of
terracing is much higher. With a ratio of 1 terrace:2 catchment the terracing
costs are only one third, and in several studies this was the most economic
system (Jones and Shipley 1975).

TABLE 5.1

THE RAINFALL MULTIPLIER EFFECT OF CONSERVATION BENCH

TERRACES

Assuming 500 mm of water is required for a sorghum crop*

  Catchment Ratio Run-off Coeft C% Run-on (MARxRatioxC) Total

MAR = 250 mm 2 40 200 450
  2 50 250 500
  3 30 225 475
  3 40 300 550
  4 30 300 550
MAR = 350 mm 2 20 140 490
  2 25 175 525
  3 10 105 455
  3 15 157 507
* The actual amount will depend on the water use efficiency, i.e. allowing for
losses from seepage and evaporation.

Note: An example of CBT being used for growing winter wheat comes from
Arizona. The average winter rainfall is 180 mm and assuming a moisture
requirement of 585 mm and run-off of 33%, a catchment ratio of 6.8 was used
(See Section 5.1.2):

( 3
180 + x 180 x 6.8 ) =~ 585
3
100

The calculated bench was 2.4 m wide, and the catchment 16.3 m (data from
Flug 1981).

These calculations assume 100% efficiency, compared with Finkel's method.

Because of the high cost of installation of CBTs, it is not appro-priate at very

low rainfalls. The main application of the system is to increase the yield and
reliability of yield where rainfall is nearly sufficient for crop production, in the
range 300-600 mm. Improving the probability of getting a reasonable crop
may be more important than numerical increase of yield. For methods more
suitable in lower rainfalls see Section 5.2.4 for simple methods, and Section
5.3 for methods which use water spreading.

5.2.4 Contour furrows (also known as contour bunds and desert strip
farming)

These are variations on the theme of surface manipulation which require less
soil movement than conservation bench terracing, and are more likely to be
used by small farmers, or in lower rainfall areas. The crop- ping is usually
intermittent on strips or in rows, with the catchment area left fallow. The
principle is the same as Conservation bench terraces, that is to collect run-off
from the catchment to improve soil moisture on the cropped area. Contour
bunds as used in Kenya, are shown in Figure 5.4. Studies were made of the
soil moisture profile (Smith and Critchley 1983) with a catchment ratio of 2:1; a
satisfactory sorghum crop was grown on only 270 mm of rainfall. It was
estimated that run-off from the catchment was 30 percent giving 166 mm of
run-on, and 432 mm available to the plants:

( 3 x 2 x 270 ) + 270 =

0 432
100
On heavier soils, contour bunds may be less effective because of the lower
infiltration. Studies on vertisols in India showed that yields were lower near the
bunds, both upslope and down, as a result of waterlogging (ICRISAT 1976).

Contour bunds are also used in Ethiopia (Plate 5.13) for a combina- tion of
soil conservation and water conservation. The bunds are built on level grade
with ties in the basin. A stone wall is built on the lower side of the earth bund
in an attempt to reduce damage if the basin is over- topped, but this has not
been very successful, and it would be better to put the bunds on grade if it is
likely that they will not be able to hold all the run-off (Hurni 1984). Contour
bunds are also discussed in Section 4.3. Another successful example, from
Ethiopia, is shown in Plate 5.14.

If the contour furrows are not laid out precisely on the contour, or are built with
some irregularities, there may be a danger of uneven depths of ponding
behind the bank. This can be reduced by smaller bunds at right angles, but as
with tied ridging, these bunds should be lower in height than the main ridges
so that if there should be any overtopping it will be laterally along the contour
and not over the bund and down the slope. Sometimes the emphasis is on the
excavated furrow which collects water, so that in exceptional storms the run-
off can overflow without damage.

A disadvantage of contour bunds at intervals down the slope is that the crop
tends to be uneven, reflecting the soil moisture profile in Figure 5.4, and only
a small part of the field is cropped. But the point is that it enables some crop
to be grown where otherwise the rainfall would be inadequate for any
cropping. The energy input for building the ridges is still considerable but
several ox-drawn ridgers are available, and more expensive but more efficient
tool bars, as discussed in Section 5.2.1. Even if the whole operation has to be
done by hand labour this may be acceptable if it is replacing a traditional
system which also has a high labour requirement. This was the situation in the
Kenya example reported by Smith and Critchley, where the traditional method
consisted of deep digging by hand to increase the infiltration and moisture
storage.

Another approach to localized surface storage on part of the land is the use of
small semi-circular catchments or larger rectangular or trape- zoidal structures
(see Figure 5.5). The large number of separate semi- circular or trapezoidal
bunds spreads the risk of damage from overflow in exceptionally heavy rain,
or if one bund fails. The semi-circles are usually constructed by hand, and are
larger versions of range pitting which is used to promote vegetation on grazing
land (see Section 7.1 and Plates 7.5 and 7.6). In several semi-arid climates in
Africa the farmers have shown more interest in using these techniques for
crops than for range improvement (Smith and Critchley, 1983, describing the
work in Baringo, Kenya, and Wright, 1984, working in Burkina Faso). The
trapezoidal bunds were developed by Finkel (1985)in Turkana, Kenya, and
can be built by hand or machinery. Semi-circular hoops built of stones for tree
planting are discussed in Section 7.2.2 and illustrated in Plate 7.21.

The trapezoidal bunds can be from 0.25-2 hectares, and so are more suitable
for cultivation. They can be levelled within the bund to ensure more uniform
spreading of the retained water, but this may not be a good idea if there is a
concentration of fertility in the topsoil. One tip of the bund may be slightly
lower to serve as an overflow spillway, which can be protected with simple
stone paving if there is any danger of erosion. Both semi-circular and
trapezoidal bunds can be staggered on alternate rows, so that the overflow
from one row will run into the next downslope, and ensure that all are filled
before there is any run-off from the field. Both types are best on gentle slopes,
especially trapezoidal bunds which are best limited to slopes of less than 2
percent, because at steeper slopes the bunds have to be built up higher and
so need a high labour input.

Small basins are used to catch and store water in many countries, sometimes
called pitting, rather like the tied ridges discussed in Section 5.2.2. Plate 5.15
shows the use of pits to store water on a tea estate in Malawi.

Where the whole of the soil surface is shaped, the ratio of run-off area to run-
on area is governed by the configuration as shown in Figure 5.6 and Plates
5.16 and 5.17. The land may be shaped by ploughing, as in Plate 5.17, or by
ridgers used with oxen, as in Plate 5.18, or tractors as in Plate 5.19. An
important advantage of small run-off producing areas compared to larger ones
is that they will yield run-off from smaller rain storms. This was discussed in
Section 5.1.3. At the Central Arid Zone Research Institute in India, using
narrow strips with a ratio of run-off to run-on of 0.75 (catchment) to 1.0
(cropped), threshold values of rainfall for inducing run-off were between 3 and
15 mm for moist surfaces and 7 to 9 mm for dry surfaces. The possibility of
increasing run-off through surface applications of bentonite or sediment from
tanks was possible but unlikely to be economic. Another possibility which
could have practical significance is changing the configuration to increase the
slope of the run-off area (Jain 1980). A system of creating furrows on sloping
land using asymmetric ridgers was shown to be successful for maize
production on an experimental basis in Mexico (Garduno 1981), see Figure
5.7.
A configuration suitable for flat or very gently sloping land, is shown in Plate
5.20 and comes from studies in Mexico. A similar cross- section but on a
small scale was shown to be successful in Colorado by Fairbourn (1975). Low
cross-ties can be used to help obtain a better distribution of water since they
reduce accidental run-off if some of the lines have a fall. Fairbourn also
showed that infiltration could be increased by burying crop residues in a
narrow trench in the cultivated area; he called this vertical mulching but the
extra complication of trenching and filling make this practice unlikely to be
adopted on a wide scale (Figure 5.8). It bears some relation to a simpler
variation in which crop residues are thrown into the bottom of furrows and the
ridges are split onto the furrows to form a new ridge for cropping the following
year. Subsoiling or deep ripping in the furrow may also be used to increase
infiltration, particularly when there is a clay pan or a laterite layer.

Treatment of the run-off area with additives, as in Plate 5.20, is unlikely to be

economic, but it may be possible to take advantage of naturally occurring soil
properties to increase the run-off. A case in point is Botswana where the
sandy soils known as the 'hardveld' become hard and crusted after heavy
rain. A method proposed there by Nilsson consists of forming permanent low
ridges about a metre wide which are left to develop this hard crust, while
ploughing, cultivating, and planting are restricted to narrow strips of 30 cm
(Figure 5.9). This system has the advantage that after it has been established,
no further maintenance is required.

The concept of strip tillage on the contour with ripping to encourage infiltration
was tested in Tanzania by Macartney et al. (1971), and in Botswana,
Willcocks (1981) showed that precision strip tillage was more effective than
other cultivation methods, particularly if the strips could be consistently
aligned along the previous crop rows.

The practice of cultivating to leave ridges and furrows on the contour without
any additional land shaping was discussed in Section 4.2. It appears to be
more common for maize and sorghum to be planted on the ridges, and small
grains in the furrows. The practice of drilling wheat, barley, and rye into small
furrows is reported from Poland by Wollen (1974) and is common in the wheat
growing areas of Kazakhstan, USSR.

In low rainfall areas a large ratio of catchment to cropped area is required, but
it is not easy to design a system which will give the best result for all variations
of annual rainfall. In some experiments in North America the ratio of 33:1 was
tried, and with 190 mm of rain gave 530 mm of run-on to the farmed area,
which was more than required and more than could be absorbed. The next
year the ratio was reduced to 15:1, but the rain was greater at 246 mm and
with a higher run-off coefficient gave 390 mm of run-on, about the right
amount. Retaining the ratio of 15:1, the next year had 140 mm rain, giving 220
mm run-on which was not sufficient. The authors (Morin and Matlock 1975)
suggest that mathematical modelling is an appropriate tool to design the
optimum ratio for varying conditions of soil and climate, but as we have
discussed previously, any system is going to have widely varying degrees of
success or failure from year to year. In a region of low winter rainfall, using
saw-tooth ridges shaped by motor grader, Shanan and Tadmor (1979) found
that the best ratio of catchment to planted strip could vary from 4-20. The very
wide range suggests that trial and error may be as effective as mathematical
modelling.

5.3 WATER SPREADING (THE USE OF RUN-ON

AREAS)
Here again we have a wide variety of methods for different conditions. The
soil type influences the method in several ways. Clay soils have low infiltration
rates and high moisture storage capacity, so they are suitable for deep
flooding with subsequent cropping, e.g. the tanks in India described in Section
5.3.3. Sandy soils have quicker infiltration and lower storage which tends to
go with the diversion schemes of Section 5.3.2. A deep soil can absorb larger
amounts of water, while a shallow soil may need the provision of overflow
outlets to avoid drowning the crop.

The climate too; affects the method. Mediterranean climates have winter
rainfall and low evaporation which may make possible run-off farming with
mean annual rainfall (MAR) down to 100 mm as in the Negev desert in Israel
(Evanari et al. 1982) and in Tunisia (Damagnez 1979). But with tropical
summer rainfall and high evaporation, water harvesting is most likely to be
useful in areas with more than 250 mm.

The probability of high-intensity rain influences the method. In the Negev most
rain falls in light showers so the objective is to maximize the run-off (discussed
in Section 6.2). In tropical Africa the problem is to handle the sudden floods
from violent summer thunderstorms.

If the rain and the growing period do not coincide, the object will be to store
water for later use. If the rainfall and growing period are the same, the desired
result will be several applications of water during the period of crop growth.
5.3.1 Natural Run-off

There are many examples of traditional use of naturally occurring run-off to

augment rainfall in areas where the rain alone is not sufficient for growing
crops. Some authors have described this as "exploiting micro-niches in the
environment".

In North America the culture of many Indian tribes has depended for centuries
on simple methods of floodwater farming. For example, the Navajo in Arizona
use run-off from sandstone outcrops to water alluvial soil at the base of the
hills (Billy 1981). Good crops of maize, squash, and melons are produced
where the annual rainfall is only 300-400 mm (Figure 5.l0a). This type of
scheme usually does not use a large investment of labour to manage the
water, but advantage may be taken of any favourable feature, such as a road
which can act as a collector drain (Figure 5.lOb), and to some extent the run-
off may be managed by leaving the upper part of the field unplanted (Figure
5.10c), or increasing the storage by a simple impounding dam (Figure 5.10d).

Several other tribes use floodwater farming extensively in the south-west of

North America, for example it is the practice of the Hopi tribe on three quarters
of their cultivated land, although the area so used today is only one third of
what it was in former times (UNEP 1983).

Four situations where simple, unimproved run-off farming may be useful are:

· alluvial or colluvial deposits at the base of ridges or escarpments;

· alluvial fans where streams leave hill land and flow onto flatter plains;

· stream banks which are intermittently flooded naturally;

· where run-off collects in naturally occurring depressions, as the

'cuvettes' in West Africa.

In Africa these conditions are all exploited, for example a case in Turkana,
Kenya, reported by Morgan (1974) and Hillman (1980). In a mainly pastoral
community with annual rainfall of less than 200 mm, small patches of sorghum
are grown in all the first three situations. Local traditional practice and
familiarity with local conditions may be useful when consider- ing possible
sites for the development of more ambitious run-off farming schemes, and this
was the case in the Turkana example (Cullis 1985).
Where the rainfall is sufficient for general cropping (and this usually means
more than about 500 mm), it may still be possible to make use of run-on areas
by increasing the cropping intensity. In the use of con- servation bench
terraces (Section 5.2.3), separate rotations are used for the run-off area and
the run-on terrace. The same principle is applied to natural run-off in Morocco
(Kutsch 1982 reported in Pacey 1986). When growing wheat or barley, a plant
density of 70-90 plants/m  is used on land receiving rain only, but the plant
2

density is increased to 90-120 plants/m  in depressions which collect run-on.

2

This does entail an element of chance, described by Pacey as "the deliberate

reduction of plant density results in some sacrifice of yield in seasons of
ample rainfall, but it makes for greater certainty that there will be useful grain
production in a drought year."

Two examples are reported by Kovda (1961) of the use of naturally occurring
run-off in arid areas of the USSR. 'Kair' farming is the name given to cropping
on flood terraces of the large rivers of central Asia where the soil moisture is
partly the result of surface flooding, but also by lateral seepage from the river.
Another example is 'khaki' farming in Turkmenistan where run-off from rain on
higher mountain slopes inundates gentle slopes in the plains. The flooded
fields are left to dry out and then cultivated and planted as soon as is
practicable. A similar practice is reported from Ethiopia by Carr (1979). In the
Woito and lower Omo valleys, as soon as the annual flood recedes, the flood
plains are cleared and planted to maize and sorghum, using heavy seeding
rates which are thinned later.

5.3.2 Collected and Diverted Run-off

The previous Section describes some uses of natural or unimproved run-off,

while this Section is about schemes where there is some element of
manipulation or management of the land or the run-off. The difference is not
important, because some systems use only very simple works. For example,
the floodwater farming of the Hopi and Navajo Indians of northern Arizona,
described in Section 5.3.1, is very similar to that of the Papago Indians of
southern Arizona who make simple barriers of wooden posts and woven
brush to spread natural floodwaters more evenly (Nabhan 1979).

It is not important whether we consider run-off farming as a type of irrigation.

In the case of terraced wadis and other water management schemes in valley
bottoms, man sets out to manage both the land and the water. Farming valley
bottoms is a very ancient practice and has been well documented in Tunisia
and in the Negev desert in Israel. In the Matmatas of Southern Tunisia the
system is called 'Gessours'. In an area with only 100-150 mm of rain, rows of
olive trees are planted across the width of the valley bottom near a minor
barrage, which slows down the floods, and result in deposition of silt with a
terracing effect (Plate 5.21). Cereal crops are grown on each terrace
upstream of the olive trees. This is similar to the terraced wadis in the
southern Negev in Israel (see Plate 5.22). In some cases these ancient
schemes are partly cultivated today by the mainly pastoral Bedouin
inhabitants (Plate 5.23).

Another example of carefully controlled water use in valley bottoms comes

from Colorado (Mickelson et al. 1965). Level pans of one to three hectares are
formed in broad valleys where slopes are less than 3 percent. The rainfall is
400 mm, sufficient for some rather unreliable cropping of grain or forage
sorghum. The cropped land yields 5 to l0 percent of the rainfall as run-off from
the heavy summer storms. The run-off from 150 ha of cropland is spread onto
the first level pan, and when the depth reaches more than 100 mm the surplus
spills over to the second pan and so on. During the season an additional 200
mm of water results in good crops on the pans, so that the installation costs
can be repaid in three to five years. In most schemes involving flooding of
crops, it is usual to provide outlets for emergency overflow, and possibly for
draining as well, because growing crops appear to be less affected when
there is a slow continuous passage of water than if the water is stationary.
Also most crops only tolerate complete saturation for a short time.

Some of the ancient Negev valley floor cultivation systems used only spate
run-off down the valley, but most of them depended also on the collection of
run-off from the surrounding hills. Some examples are shown in Plates 5.24a
to d. Some had sophisticated methods for inducing run-off, collecting it in
channels and distributing it to the various terraced fields. A fascinating
programme of research has been conducted for twenty years by the
universities and research institutes of Israel and has been splendidly recorded
(Evenari et al. 1982). The surface management to induce run-off included
collecting and piling all the loose surface stones into mounds, and shaping the
surface into ridges which are very similar to the roaded catchments used
nowadays in Australia (see Plate 5.25). The hydrological effects of removing
the stones are explained in Section 6.2.1. and the effect of building ridges in
Section 6.2.2.

The collection systems in the Negev show that the hydrologic principles were
fully understood. Stone-lined conduits were used to take the water quickly
down to the farms with minimum loss by infiltration and evaporation, and the
catchments for these drains were long and thin so that rain was collected
even from the low intensity winter rainfall common in this area. The
distribution systems were also simple in construction but sophisticated in
concept, and allowed the water to be transferred and diverted around the farm
quickly and efficiently with minimum labour requirements.

As a result of the successful studies in the Negev desert, trials of run-off

farming were carried out in the Khost plain, in Paktia province of Afghanistan.
The terrain is similar to the Negev; stony barren hills where run-off is collected
in channels and led down to the plain. Level fields, surrounded by an earth
bank, are laid out in shallow steps so the water is passed down-slope from
field to field (Plate 5.26). After the soil profile has been saturated a cereal crop
is grown on a combination of subsequent irrigations if there is rain, and the
stored moisture in dry periods.

Another ancient collecting system called 'meskats' comes from the Sousse
region of Tunisia with a rainfall of about 300 mm. Olives are grown on the
better soils at the bottom of slopes. On the upper slopes, the soils have a
tendancy to surface crusting and yield run-off into long thin catch- ment areas
feeding the water down in channels to the olive groves (Plate 5.27).

These methods, developed for regions with winter rainfall, may not be suitable
for the semi-arid tropics. Summer rainfall is more intense, so in general the
catchment area required will be smaller, and provision for the safe passage of
exceptionally high rates of run-off must be included. Figure 5.11 shows a
system which has been tried on a small scale in two areas in Kenya, in
Baringo by the Baringo Pilot Semi-Arid Area Project (BPSAAP) and in Katui
by the Arid and Semi-Arid Lands Project (ASAL). Although not yet adopted to
any great extent by the local farmers, the national policy of land allocation to
what has traditionally been a mobile pastoral community, will probably lead to
a stronger motivation for regular cultivation on permanent plots, which will not
be possible without some form of water harvesting. Ratios of catchment area
to cultivated land of up to 5:1 are expected to be suitable. For the system to
work well, the cultivated area should be levelled so that there is little or no
lateral slope along each terrace, and only a gentle slope across the terrace to
avoid erosive flows. Stone-packed spillways built into the earth bund allow
surplus water to pass progressively downslope to lower terraces. The danger
of damage from excessive floods should be avoided, not by trying to control
the water once it has reached the cultivated area, but through the provision of
overflow by-passes set into the bank of the collection drain. There is no doubt
that such a system can be made to work on an experimental or demonstration
basis, and the question is whether the pressures of social change will make it
sufficiently attractive to people unaccustomed to a cropping system which
requires this degree of precision and maintenance.

Another form of water spreading in USSR known as 'liman' irrigation is

described by Kovda (1961). This consists of what are described as contour
ramparts across the slope, perhaps an apt description since they are large
structures up to two metres high and four or five metres wide (similar to the
'murundum' water conservation structures used in Parana State in Brazil).
Sometimes drain holes are used to pass the water through to lower levels,
although an open spillway would appear to offer a simpler and more effective
way of controlling the water movement.

The diversion of run-off water onto prepared level terraces is another ancient
and widely used method. This is the basis of growing irrigated paddy in humid
areas where water supplies are assured, and the same principles may be
used to obtain benefit from intermittent water supplies in semi-arid regions.
For example, in the mountainous dry areas of northern Pakistan, in the district
of Gilgit and the Hunza valley, agriculture is entirely dependent upon the
diversion of springs, streams, and melt waters from snow and glaciers, onto
terraced alluvial outwash fans which are accurately called oases, Plate 5.28.
Some amazing construction feats have been used to carry the 'khuls' across
unstable scree slopes, and laboriously cut by hand out of rock faces, Plate
5.29.

Another interesting example of diversion onto terraces comes from the

Manakhah region of the Yemen Arab Republic (Damagnez 1979, Vogel
1985). In an area of annual rainfall of 400-600 mm, level bench terraces are
constructed on steep slopes up the hillsides, with rock risers 2-3 m high and
the terraced fields 2-l0 m wide (Plates 5.30 and 31). Run-off from the
uncultivated and unterraced upland is carried to the terraces in diagonal
ditches reminiscent of those in the Negev. Stone-lined canals carry the water
progressively down through the terrace system, sometimes with the canals
running underneath the terraces. This sophisticated and efficient system is
presently falling into decay (see Plate 4.18) because of labour migration to the
oil producing countries of the Middle East. A similar system in the Yemen
Arab Republic, called 'sayl' irrigation, is the diversion of flood water out of
wadis onto levelled terraces.

The management of grazing land is discussed in Chapter 7, but a water

spreading system from Australia is worth mentioning. In New South Wales,
Australia, run-off occurs from eroded and degraded ridges and is absorbed on
the flat land below the ridge, but produces only inedible scrub vegetation,
sometimes called 'mulga' after the dominant shrub, Acacia aneura. Two
approaches are used to make more efficient use of the rain- fall. One is to
encourage useful grass growth by retaining it on the ridges using short
contour furrows as described in Section 7, (Cunningham 1967 and 1975). The
other approach is to catch the run-off in drains and lead it to where it can be
spread on land for fodder crops and pastures. The collector drain has the
bank on the lower side; the spreader drain has the spoil bank on the upper
side so that the water spills evenly over the downhill edge of the channel
(Figure 5.12). During large run-off flows, water also collects on the upstream
side of the spreader channel, so gaps are provided to allow this water to pass
through into the spreader channel (Quilty 1972).

5.3.3 Inundation Methods

Some examples of naturally occurring short-term flooding were described in

Section 5.3.1, mainly on flood plains and in valley bottoms. This Section
describes systems where flood waters are impounded and retained long
enough to saturate the soil so that a crop can be grown on moisture stored in
the soil.

Simple systems on a small scale are used in the Sudan. On gently sloping
land embankments known as 'teras' collect and hold surface run-off, and after
the water has soaked into the soil a short duration millet is planted and
matures in eighty days (Pacey 1986 p 131). In the semi-arid north-east of
Brazil, the government research organization CPATSA in Petrolina has
studied a similar method, and produced designs for the optimum shape and
size of the bunds. There are also recommended cropping methods to make
maximum use of the stored moisture, with variations in the cropping pattern
depending upon the amount of rainfall and the amount of run-off stored
(EMBRAPA 1981). The inundated areas are known as 'vazantes'.

In India the use of contour bunds to retain run-off has a mixed story of
success and failure. When used mainly as a soil and water conservation
measure in areas of reasonable rainfall (750 mm - 1250 mm) on medium-
deep vertisols, it was found that the disadvantage of waterlogging in the
vicinity of the bund both uphill and downhill exceeded the advantage of
increased cropping from the stored moisture in a dry season (ICRISAT 1975).
But in areas of lower rainfall the method has been more successful, for
example in the Siwana district of western Rajasthan. With a rainfall of 250 mm
or more, low contour bunds only 0.3 m high divide the land into strips which
are progressively inundated down the slope with flooding to a depth of 0.2 m
during the monsoon. When the rain ceases, the stored water infiltrates and a
winter crop is sown.

In semi-arid areas of India, large-scale inundation schemes have been used

for hundred of years, some with sophisticated forms of water control and land
management. In Bihar and Uttar Pradesh there are thousands of 'ahars'
covering a total of more than 800 000 ha. Low earth bunds are built to retain
run-off during the monsoon and when used on very gentle slopes, sometimes
as low as 0.01 percent, water stored to a depth of one metre will throw back
several kilometres, so large areas of land are covered. Most ahars flood less
than 500 ha but others are as much as 4000 ha (Kolarkar et al. 1980 and
1983). The volume of water stored is less important than the area of land
which is submerged. The soil must have sufficient depth and moisture-holding
capacity to store enough moisture to carry a five month crop which is grown in
'rabi', the dry winter season, after the end of the summer monsoon season,
'kharif'.

The retaining bank is usually not more than 3 m high and may extend several
kilometres on the contour, with some examples up to 10 km. There is usually
a wide waste-weir to serve as emergency spillway, usually stone pitched with
a crest level one metre below the top of the bund to give one metre freeboard
for wave action. The emergency spillway is required partly because of the
danger of a heavy storm occurring when the tank is nearly full, and also
because the ahars are usually built in a series down the slope and the failure
of any one would imperil those lower down the slope.

It is also usual to provide metal sluice-gates set in concrete structures to allow

for quick emptying when necessary at the end of the rains. The water
released may be collected in the next ahar down the slope. Sometimes pipes
are laid through the bund if some of the discharged water is to be used for
irrigation.

The main crop is winter wheat planted as soon as the flooded land has dried
out sufficiently, and then grown out on the stored moisture. Sometimes a
subsidiary crop is taken during the summer monsoon while the land is
flooded, using a variety of rice known as 'floating rice' because it can grow up
through the standing water. This technique of taking one crop of floating rice
followed by a second crop after the water has receded is also practised in
Thailand.

A secondary advantage of this system is that the soils in these semi-arid

areas often have a tendency to salinity, and this is controlled by the regular
leaching by the inundation water. The infiltration of considerable amounts of
water may also have the effect of raising the general level of the water table
near to the ahars, with improvement of the supply in shallow wells.

Another form of inunadation farming is used in the Jaisalmer district of

Rajasthan (Kolarkar et al. 1983). Here the topography is more broken, and the
method is to build an earth bund across the valley plain to catch and store
run-off and silt from the surrounding barren hills. These are called
submergence tanks or 'khadins' (Figure 5.13). In the district of Jaisalmer there
are more than 500 such tanks with an inundated area of more than 12 000 ha.
A catchment ratio of at least 15:1 is usual and the tanks are designed to fill
from rainfall of 75-100 mm, in an area where the annual rainfall is 165 mm
with great variability. The main construction features are the same as the
ahars, that is a wide emergency spillway and sluices to release the water. The
depth of water stored varies from 0.5 to 1.25 m and it usually disappears from
seepage and evaporation by early November, when the winter crop is sown,
either winter wheat or the grain legume Cicer arietinum, (Bengal gram or
chickpea). Similar systems of inundatiion farming Pakistan are known as
'sailabas' and 'kuskabas'.

Secondary benefits of most inundation systems are the leaching of salinity by

the stored water and the improvement to the yield of wells downstream.

5.3.4 Flood Diversion

The diversion of run-off water was discussed in Section 5.3.2. This Section is
concerned with diversion and spreading of floods and spate flows.

Diversion of flood water from its channel usually involves some form of
structure, a barrage or weir to divert the water. For small schemes, simple
diversions may be constructed each year using stones and boulders, perhaps
in wire netting, or poles and brushwood where these are available. It may be
sensible to expend only limited effort on building such struc- tures where it is
probable that they will be periodically destroyed by flash floods, and an
example from Yemen was quoted in Section 4.1.4. The floods are also likely
to damage the conveyance channel unless it is pro- vided with safety devices
to spill water picked up in large floods. To avoid these problems a method has
been widely used in India for centuries, the inundation canals, or 'pynes'. A
canal is excavated in the river bank so that the level of the canal bed is
considerably higher than the bed level of the river. During low flows the canal
is dry, but when the flood level rises to that of the diversion canal it starts to
flow. There is no need for any structure in the stream or river channel. This
system is widely used in the Sind province of Pakistan to collect water from
the Indus river and its tributaries during the flood period from April or May to
September. The grade of the distribution canals is slightly flatter than that of
the river bed to increase the command of the diverted flood water. This also
means that there is heavy deposition of sediment parti- cularly at the
beginning of the canal, and regular clearing is essential. Some control is also
required of the diverted flood water and this is usually done by a regulator
several kilometres from the river. Subsequent control and distribution of the
water is the same as for any surface irrigation scheme, except that the flow is
ephemeral and used for heavy inundation so less precise control is required.

The other possibility is to divert the flood water out of its channel by raising the
water level through some form of weir or barrage. These may be temporary
and expendable in simple schemes, or permanent structures. The problems to
be overcome by all flood diversion schemes are:

· damage to the diversion works during flash floods;

· control and spreading of the diverted flood water;

· deposition of sediment carried within the flood water.

In some cases the last one may be considered beneficial, as in the flood
irrigation in the valley of the Nile, and flood irrigation in China from silt-laden
rivers, where this is known as 'warping' and considered to be part of the
process of maintaining fertility. The mica-laden melt waters from the glaciers
of the Himalaya are used to build up the physical volume of soil by spreading
the water onto gravel beds. On the other hand, too much deposition of
sediment can lead to undesirable change of soil texture, or to reduced yield of
grassland.

Another example of large-scale diversion of flood waters comes from the

Yemen Arab Republic where the valleys discharge into the Tihama desert
which bounds the Red Sea, using elaborate diversion structures as shown in
Plate 5.32 (Damagnez 1979). Another variation of large-scale use of flood
water is illustrated in Plate 5.33, where structures are used to control the flood
water after it has spilt naturally from the River Niger. Plate 5.34 shows an
example of using spate flood water to inundate an olive orchard in Sfax,
Tunisia, in an area with an annual rainfall of 150 mm.

The diversion of flood water can also be practised on a smaller scale as

illustrated in Figure 5.14 showing an interesting project in Tigre, Ethiopia.
Water is first diverted from the river and then distributed using a simple
system of structures constructed from local stone, shown in Plates 5.35 and
5.36. This is a development project sponsored by the UK charity War on
Want, and is continuing. A similar scheme on a pilot project in Turkana,
Kenya, is shown in Figure 5.15. This pilot project was successful for a short
time but was soon abandoned, first because of siltation at the diversion weir
and later because the river changed its course. A basic problem with the
introduction of schemes like this is that permanent cultivation on settled sites
is not part of the traditional farming system, so there is little interest, and little
motivation for the regular maintenance which is required. However there are
some indigenous water spreading schemes, and examples from the Sahelian
belt are reported by Pacey (1986), stretching from Southern Mauritania and
Niger to Sudan and Somalia. Other examples of indigenous soil and water
conservation methods in Africa south of the Sahara are given in a study
commissioned by IFAD (1986).

Examples of the diversion of flood water to improve grazing are discussed in

Chapter 7.

5.4 SURFACE DRAINAGE

The main problem in semi-arid climates is shortage of water, but there are still
occasions when at particular times or on special soil conditions the problem is
too much water, and some form of drainage is required.

A soil where some form of water control can be important is the vertisols,
heavy clay soils, usually dark, also called cracking black clay soils, cotton
soils, and self-mulching clays. They are also known collo- quially as 'one-day
soils' because they are difficult to cultivate when dry and hard, and also
unworkable when saturated and sticky, but they do have a short window while
drying out when the moisture is just right for cultiva- tion. The physical
properties causing this are a high clay content, with the clay mineral being
montmorillonite or illite, which, because of their 2:1 lattice structure, expand
when wet, and shrink and crack when dry. A characteristic surface feature is
the development of an uneven surface with raised mounds and shallow
depressions caused by repeated cycles of wetting and drying and known in
Australia as 'gilgai'.

The occurrence of vertisols is mapped in Figure 5.16 from Dudal (1965). The
two main kinds of formation are the large uniform areas derived in situ from
extensive volcanic flows of basalt, or smaller local- ized alluvial deposits in
valley bottoms as in Ethiopia, Zimbabwe, and Zambia. In both situations,
vertisols have considerable economic signifi- cance because they are a land
resource with under-utilized potential owing to the difficulty of handling them.

One technical solution is the installation of subsurface drainage, and this was
shown to be effective more than twenty years ago in Zimbabwe (Robertson
1964). Pipes of concrete, fired clay, and bamboo were all effective but
uneconomic because of the cost of materials and transport. Today, the most
appropriate method would be the installation of plastic tube drains by the
trenchless method. The techology is available, but a high return would be
required to justify the installation costs. Mole drains would be cheaper but the
extremes of wet and dry conditions would lead to a short life.

Surface drainage is the more practical alternative and has been thoroughly
developed and tested at ICRISAT in India, using the broad bed and furrow
method described in Section 5.2.1 (Krantz et al. 1978, 1981, and Swindale
and Miranda 1984). In India, surface drainage can lead to the opportunity for
early dry-sowing just before the start of the monsoon where the onset is fairly
predictable (Krantz 1981). This a1lows double cropping instead of the
traditional practice of the vertisols lying fallow during the monsoon, and only
being cropped in winter on the stored moisture. In Ethiopia the occurrence of
vertisols is in smaller areas in valley bottoms, which remain either uncropped
altogether or only cropped towards the end of the rainy season. Improved
ease of cultivation, through surface drainage by raised bed and furrow or by
shallow open drains, could lead both to improved production from the areas
now cultivated, and also to a large increase in the area used for crop
production.

Vertisols offer the most opportunity for the use of surface drainage, but there
are other situations where surface drainage may be useful. The BBF system
has also been developed on alfisols in north eastern Brazil (Lal 1985) and a
variation on the same theme called ridge-and-furrow has been used on quite
light soils in Zimbabwe and Kenya. In the last two cases, shallow soils
become waterlogged because of a limited moisture storage in the profile, and
a simple surface drainage system can lead to better opportunities for
cropping, or an extended season for grazing. An interesting but specialized
need for surface drainage comes from Ethiopia in the growing of teff which is
the main food grain

The seeds are minute and broadcast onto a very fine tilth, and can easily be
washed away by any surface run-off. The traditional solution is small open
drains on a very steep gradient to intercept any surface flow. The need to
prevent surface run-off is considered by the farmers to override other
considerations like soil erosion or conserving moisture, and there results the
illogical situation, shown in Plate 5.37, where the traditional open drains have
been cut through more recent systems of channel terraces.

Another example of surface drainage to solve a particular problem comes

from Portugal where the situation is temporary waterlogging at the time of
preparing or planting winter cereals. The method is called 'pnudivales' which
are small channels constructed after seed-bed prepara- tion and before
seeding. They are made with three passes of a tractor. First with a
mouldboard plough to create a furrow, followed by two passes with a mounted
ridger to produce a shallow open drain between 3 and 4 m wide and 0.25 m
deep at the centre. This allows subsequent drilling, cultivating, and harvesting
in straight lines ignoring the open channels. These are set out on a gentle
gradient of 0.6 to 0.8 percent (Sims 1983).

5.5 OTHER SOURCES OF WATER: SNOW, DEW AND

MIST
5.5.1 Snow Management

In high latitudes, some or all of the precipitation may fall as snow, particularly
in the northern hemisphere, and in semi-arid regions there may be the
possibility of increasing the effectiveness of the precipitation by management
of the snowfalls.

Trapping snow during the winter to increase available moisture in the short
summer is a regular practice on the semi-arid steppes of Eastern USSR in
Kazakhstan. Some of the first snow is caught by the stubble of the preceding
crop, or, in the case of summer fallow, by low lines of mustard planted for the
purpose. As soon as 50 or 100 mm of snow has accumulated it is pushed into
ridges by heavy-duty tractors with a V-shaped bulldozer on the front and an
inward sloping half grader blade on either side, so that the snow is pushed
into two windrows or snow ridges one on either side of the tractor. A typical
design is 6 m between centres of the ridges with 3 m of untouched snow
between. The ridges are up to 1 m high and 3 m wide, and serve as wind
breaks to catch more snow between the ridges. In a year of very heavy snow
the total snow pack between the ridges can be up to 600 mm but this is more
than desirable since the melt water causes erosion problems; 400 mm is
normal and more desirable.
The conservation bench terraces, described in Section 5.2.3, have the main
purpose to hold on the level terraces the run-off from catchment areas on the
uphill side. In the two most northern states of the USA where the method was
tested, it was found that they also increased the amount of trapped snow,
partly as a result of the terraces acting as a physical barrier, and also because
of the stubble of the previous summer crop grown on the terraces (Haas and
Willis 1968 on North Dakota, and Black 1968 on Montana). Two other
methods of snow trapping have also been practised in Montana. Neff (1980)
reports that contour furrows caught snow equivalent to an additional 22 mm of
rain, and also reduced the run-off from winter rain and the snow melt in the
spring. In eastern Montana, Saulmon (1973) used a standard snow fence
(1.25 m high with vertical wooden slats) to trap snow on small catchments
which provide surface run-off for small stock watering ponds. The additional
depth of snow gave an increase of 100 mm of surface run-off.

5.5.2 Dew and Mist

The extensive literature on dew as a source of water contains many

qualitative opinions about its importance or lack of it but few quantita- tive
scientific studies. The most detailed study in semi-arid areas is that in the
Negev by Evanari et al. (1982). They conclude that the annual total of dew
may be significant in comparison with the annual rainfall. An example is a
range of 25-35 mm from dew at Avdat where the mean rainfall is 150 mm. The
dew formation is considerably more reliable than the rainfall, so that it is
possible, as in 1962/63, for the annual dew formation at 28.4 mm to exceed
the rainfall at 25.6 mm. This is very much greater than the amounts of dew
formation precisely measured in England by Monteith (1957) who found that
the average dewfall was from 2-5 mm. However, the desert dew is made up
by a large number of very small amounts and not of any consequence as a
water supply which can be collected, nor for any higher forms of desert plants,
although it is used to some extent by the lower forms like lichens and
xerophytic algae.

The Negev studies also showed that the piles of stones sometimes previously
assumed to be dew collectors for the production of grape vines do not acquire
enough dew on the surface of the stones for it to drip off, and that the mounds
are in fact for the purpose of increasing surface run-off. There are also reports
from China of a long history of growing melons in soil beds covered with a
layer of gravel 100-150 mm thick and known as 'gravel fields for melon'.
Condensation and dew formation dripping down to the soil is supposed to be
the purpose of the gravel but its effect may perhaps be to maintain an even
soil temperature or to reduce evaporation.
There are a number of examples of artificial collectors, such as plastic sheets
being used to collect enough dew to keep alive forest seedlings in the desert
in Israel (Gindel 1965), and another use of plastic sheets to produce
emergency supplies of drinking water in Australia, but these do not have any
agricultural significance.

There are also reports of aerial wells which supply drinking water in many dry
areas, some with piped supplies, but the studies in the Negev would suggest
that the process is more likely to have been the interception and condensation
of fog or mist, or direct run-off, than dew formation (UNEP 1983). The story of
dew formation and mist condensation is interest- ing, and full of
contradictions, but the conclusion has to be that this is not a significant source
for water for agriculture.