You are on page 1of 11

1998. J. Soil. Sci. Soc. Sri Lanka 10:36-56.

IMPACT ASSESSMENT OF SOIL CONSERVATION MEASURES


IN DRY ZONE RAINFED FARMING

P.B. Dharmasena
Field Crops Research and Development Institute
Maha Illuppallama

ABSTRACT

Declining soil fertility and moisture stress have been identified as the main
physical constraints in the dry zone rain-fed agriculture. Both these constraints
are surfaced with accelerated soil erosion taking place in rain-fed upland farming
areas. The erosion becomes hazardous in these lands due to reasons such as high
soil erodibility, shallow soil depth, occurrence of erosive rains when the land is
bare and vulnerability of already degraded lands to further erosion.

A study carried out at Mahailluppallama with the objective of finding the


effectiveness of various soil conservation practices has led to developing a Soil
Conservation Impact Assessment (SCIA) Model for use in soil conservation
planning. The SCIA model provides opportunities to compare effects of different
mulch types, rates and land management practices on soil erosion, which will lead
to select an effective soil conservation package for a farmer on the basis of his
resource availability and preference.

The model suggests that the conventional type of bunding is not adequate to
reduce the soil loss to a permissible limit. Use of a mulch without any other land
management method also does not protect the soil adequately. Mulching if
combined with conventional bunding is an effective measure to control soil
erosion. Alley cropping with gliricidia planted on small graded ridges and mulch
strips with live leguminous creepers (Mucuna utilis, Centrosema pubescen)
perform well in controlling soil erosion in the dry zone rain-fed upland farming.

Key words: Erosion modelling, strip-mulch, graded alley, bunding, soil life.

1
INTRODUCTION

The traditional chena cultivation in the dry zone of Sri Lanka has been subjected to
various modifications during last few decades as pressure on the land increased with
trends of production needs. As a result fallow period has become shorter and this has led
to make the soil gradually less fertile with respect to physical, chemical and biological
conditions. Experimental evidence from Kurundankulama dry farming project indicated
that soil fertility quickly depleted under continuous cultivation on the uplands, and that
the addition of fertilizers did not result in substantial crop returns without addition of
organic manures such as cattle dung or compost (Abeyratne, 1956).

On the basis of experiments conducted at Mahailluppallama, Somasiri et al (1990)


spelled out following four factors as major constraints in rainfed farming: (a)
deterioration of the surface soil structure seriously affecting soil tilth and moisture
conservation; (b) proliferation of obnoxious weeds which makes land preparation
extremely difficult; (c) depletion of nutrient reserves in the soil; and (d) as the rainy
season is short the need for timely land preparation to establish crops with the on-set of
rains.

As the soil erosion process continues the surface soil layer changes with respect to its
physical composition as well as structural configuration. Both these changes lead to form
a more compacted soil layer which does neither retain much water nor keep favourable
soil tilth for cropping. Most of the nutrients are washed off by erosion, and the remaining
soil becomes infertile. Such environment favours weed growth rather than a crop,
causing the weed problem which is the most apparent constraint to the farmer. Thus, a
land with more weeds and of compacted soil provides a difficult situation for timely
cultivation. Therefore, the four problems diagnosed above by Somasiri et at (1990),
could be easily interpreted as problems arising from soil erosion (Dharmasena, 1992a).

Soil erosion caused by extensive clearing of catchment forest of village tanks, and
farming without conserving soil has been a major issue for last few decades in the dry
zone agriculture of Sri Lanka (Brohier, 1975). The consequences became more visible
due to population increase and subsequent pressure on the arable land. In this process,
not only the upper land becomes less productive or perhaps unproductive (Tennakoon,
1980), but also the tank below is filled with sediment (Dharmasena, 1992b) making the
entire farming system agriculturally inefficient or unproductive.

Due to lack of knowledge on the magnitude and the rate of soil erosion in the dry zone,
the causes have not been properly diagnosed and the problem has not been adequately
addressed. Thus, rainfed upland farming system is becoming non-profitable to farmers,
resulting in resource degradation and causing irreversible hazards to the environment.

OBJECTIVES

The study was planned to focus on two aspects. First was to assess the present erosion
status in the rain-fed upland, and to propose a permissible level of erosion for a given
farming situation. Second was to develop a soil conservation impact assessment model
which can be used as an estimating tool in soil conservation planning to assess the impact
of various conservation measures that are being practised and proposed by the study.

2
METHODOLOGY

The study was carried out in two rain-fed upland sites in the dry zone of Sri Lanka in
Yoda Ela Farm and Paindikulama in Anuradhapura district during the period from 1988
to 1990. It aimed at understanding following aspects: (a) rainfall - erosivity relationship;
(b) erosion status in different land uses such as chena, scrub and matured forest, (c) effect
of different rates and types of mulch on soil loss; (d) comparison of different soil and
crop management practices; and (e) development of the Soil Conservation Impact
Assessment Model.

Rainfall-erosivity relationship
The erosivity was calculated from recording rain gauge charts by computing EI30 values
(Wishmeier & Smith, 1958) and related to monthly values of rainfall measured at
Mahailluppallama during the study period.

Erosion status in different land uses


Runoff field plots of 22 m long and 4 m wide were constructed with collecting tanks at
Paindikulama study site. Locations were selected to measure runoff and soil loss from
forest, scrub and chena lands, using two plots for each land use. Runoff was measured,
water samples collected for each storm separately, and the soil content eroded was
determined.

Mulching experiments
Runoff plots of 10 m x 2 m were constructed with a single collecting tank. Plots were
cropped for both yala and maha seasons and a mulch was applied at different rates in
each season. Four types of mulches were tested on their performance in soil erosion
control by using these runoff plots. The randomized complete block design was adopted
with three replicates.
Runoff samples were obtained from collecting tanks for each storm separately. The
magnitude of soil loss was calculated by obtaining oven dry weight of the sediment
sample, which was used to estimate the total amount by considering the volume of water
in the tank.

Comparison of different soil and crop management practices


The experiment was carried out in 100 m x 40 m field plots bounded by earth bunds. As
the expected runoff volume is high, the collecting tank system was designed with multi-
shot divisors which split the flow into equal parts and pass one part, as a sample, into a
second collecting tank and so on. Runoff and soil loss were measured by sampling water
from each collecting tank and total soil loss was calculated accordingly.

Following practices were tested: (a) Plough farming - Land was ploughed (not in a
specific direction) and cultivated; (b) Bund farming - Land was conserved by
constructing a series of earth bunds across the slope with a slight gradient of about 0.4 -
0.5 percent. These bunds were 60 cm high, and 120 cm wide at the base and laid at 15 m
distance apart. A shallow surface drain was constructed at upper side of the bund usually
10 - 15 cm deep and 30 - 40 cm wide; (c) Strip mulch farming - A mulch strip was

3
maintained at 5 m intervals and 0.75 m wide rows across the slope. Centrosema
pubescen was selected as the live mulch at the beginning, but this was replaced with
Mucuna utilis due to poor growth of Centrosema. Weed and crop residues were also
heaped on the same strip in order to form a runoff filter reinforced by the live mulch. At
the beginning of the season excess of this trash materials was spread over the cropped
land as a mulch; and (d) Graded alley cropping - Gliricidia sepium was used as the
hedgerow and planted with a slight gradient (hence it is named `graded alley') to facilitate
the diversion of excess runoff into a protected waterway.

Gliricidia was planted at 4 m spacing with 0.5 m within row distance. Small soil ridge
was formed along the hedgerow with a small drain on the upper side. In addition a grass
strip of 2 m was established in place of each fifth hedgerow at 20 m interval where a line
of fruit crops was grown.

Development of Soil Conservation Impact Assessment (SCIA) model


The impact of various conservation practices was assessed by conducting series of
experiments mentioned early. In practice, farmers can adopt more than one of these
measures simultaneously and select practices according to their preference. Thus, in
erosion control planning, it is important to develop a model which could assess the
impact of a package of practices that a farmer selects, or the requirement of an erosion
control package for a certain farming situation. The Soil Conservation Impact
Assessment (SCIA) Model was developed to fill this planning gap on the basis of the
results obtained in the present study.

Results of above mentioned experiments were compared using relationship of erosion


parameters developed by Wishmeier and Smith (1978) known as Universal Soil Loss
Equation (USLE). The equation was slightly modified (Dharmasena, 1994) to serve the
purpose as follows: A = RK (LS)max Ec., where, A = Soil loss (mt/ha), R = Erosivity
(mt.m/ha), K = Erodibility (LS)max = Maximum slope - length factor one can achieve by
manipulating land slope and length in a given piece of land and Ec = soil and crop
management factor termed erosion coefficient. The model was formulated by using data
gathered in the study as illustrated in Fig. 1.

Input data
A computer programme was designed using GWBASIC for the SCIA model. The input
data required to run the programme are soil erodibility, land slope (%), slope length (m),
total effective soil depth (m), minimum soil depth required for the crop (m), mean
monthly rainfall (mm), the type of farming practice and rates and types of mulch.

Model output
The SCIA model would provide soil loss tolerance level, maximum soil life, predicted
soil loss rate and soil life with a given conservation package. Soil life is defined as the
remaining useful productive life of a soil at existing rates of erosion or at rates that have
been calculated from proposed activities (Stocking and Pain, 1983).

Sensitivity analysis
Sensitivity of soil loss and soil life to land slope, slope length, soil depth, erodibility and
mulch rate was studied. Sensitivity values were calculated by using the equation:

4
(∆ output)/(output)
Sensitivity value = ------------------------------------
(∆ parameter)/(parameter)
where, _ output is the change in the output value and _ parameter is the increment
made on the input parameter.

In the calculation a ten percent increment of the input parameter was used to find the
change in the output value. The exercise was carried out to the most practical range of
the input parameters, and sensitivity variation for a particular input parameter was
determined.

RESULTS AND DISCUSSION

Rainfall-erosivity relationship
The relationship found between monthly rainfall (R) and erosivity (EV) is:
EV = 1.14 R - 3.4, (r2 = 0.84**), where R is in cm and EV is in mt.m/ha.

Such relationships have been considered valid in many countries for very large areas with
similar rainfall pattern in erosivity mapping (Morgan, 1986).

Erosion from different land uses


The experiment was carried out during the period from November, 1988 to May 1990 to
determine the magnitude of soil loss under forest, scrub and chena lands. Modified
USLE was used to calculate monthly Ec. values for these land uses (Table 1).

Table 1. Monthly Ec ranges observed for chena, scrub and forest lands.
Land use Ec range
Forest (selectively logged) 0.000 - 0.001
Scrub (chena abandoned for more than 8 years) 0.000 - 0.006
Chena (cultivating at least once in two years) 0.001 - 0.435

During the period of observations, 37 storms were measured. Total rainfall of these
storms was 1138 mm, and the total erosivity was 105 mt m/ha. Soil loss measurements
showed that from cultivated chena 12.5 mt/ha of soil were lost on 5 % land slope, and 23
mt/ha of soil on 10 % land slope. The study period covered two maha and two yala
seasons, but first maha season was incomplete as measurements were started late in
November. Since all rainfalls were not considered in the study, reported values can not be
used to calculate annual soil loss from these land uses. Soil erosion from forest lands was
found negligible, and the fallow land covered with regrowth of natural vegetation has
kept the erosion to a minimum.

The erosion coefficient value of chena lands was as high as 0.44 in October, but as low as
what scrub land produced in the months of low rainfalls. This was most probably due to
a). the weed cover which protects the soil from erosion, b). the low erosivity and c). the
negligible runoff in the off season months. The erosion coefficient of forest land very
seldom reached 0.001. However, such values can not be expected from agricultural lands

5
even under good level of management. The highest erosion coefficient (Ec.) value
observed from scrub land was 0.006 in September in the absence of any human activity,
and this may be due to poor cover of the natural vegetation which was affected by the
long term dry period. With these reasons, it could be concluded that the permissible
erosion coefficient value should be as low as 0.006, and also it may not be practicable to
attempt lower values than that of natural erosion. On the other hand, scrub vegetation the
natural regrowth formed in chena lands after keeping undisturbed for a considerable
period of time (at least 6 - 8 years), has a good potential to protect the land from erosion.
The ground cover becomes poor in September after experiencing a long dry period from
June to August. Few rains occurring in September are convectional and bearing a high
erosive power. Even under such adverse conditions, erosion coefficient of these scrub
lands did not exceed 0.006. Thus, the permissible soil loss rate was decided to be at the
erosion coefficient (Ec.) of 0.006.

Mulching experiments
Results of mulching experiments are summarized in Table 2. Ec. values were calculated
by using the modified USLE. During the period of study, varieties of rainstorms were
experienced. Intensities as high as 110 mm/hr were among them. The daily rainfall was
sometimes as high as 140 mm. The rainfalls lower than 15 mm had not produced any
runoff from these plots and were disregarded in analysis. Few storms could not be
measured because the daily rainfall exceeded the capacity of collecting tank or due to
tank leaks; hence these rainfalls were also omitted in the analysis. Rainfalls occurred
during off seasons or before placing the mulch were also excluded. About 40 storms
which contained a total of 1577 mm of rainfalls were considered in the analysis.

Table 2. Ec values for different types and rates of mulch

Mulch rate (t/ha)


Season* Mulch Rainfall
(mm) 0 1 2 4 8

Yala 1988 Crop residue 354 0.32 - 0.23 0.11 0.08


Yala 1989 Weed residue 156 0.45 - 0.20 0.16 -
Maha 1989/90 Rice straw 562 0.29 - 0.12 0.06 0.04
Yala 1990 Gliricidia 210 0.48 0.41 0.28 - 0.15
Maha 1990/91 Gliricidia 295 0.38 0.23 0.08 0.05 -
* Yala - March to August, Maha - September to February.

Results indicated that effectiveness varies remarkably with type and rate of mulch.
During 90/91 maha season gliricidia mulch was placed just before the rains, therefore, its
response was so high. Apart from that, rice straw was found more effective than other
mulch types most probably due to its slow rate of decomposition.

A mulch rate of 4 t/ha could greatly reduce the soil loss, but none of the materials could
achieve the permissible Ec. value of 0.006. This indicates the necessity of combining
other erosion control practices along with the application of mulch.

6
Effectiveness of soil and crop management practices
An assessment was made to determine the effectiveness of 4 different soil and crop
management practices at Yoda Ela area in plots of 100 m x 40 m by measuring soil loss
and runoff for five seasons (3 yala seasons and 2 maha seasons). This included nearly 50
storms during the period from April, 1988 to January, 1991. The summary of the soil
loss with Ec. values for each season is given in Table 3.

Table 3. Soil loss of different soil and crop management practices.


Rainfall Soil loss (t/ha) Ec.
Season mm No. EI30 P B S G P B S G
Yala 1988 377 11 40 1.0 0.5 0.9 1.1 0.05 0.03 0.06 0.08
Yala 1989 185 07 15 0.8 0.5 0.1 0.0 0.10 0.06 0.01 0.01
Maha 1989/90 682 16 60 12.4 8.3 2.0 1.3 0.43 0.26 0.09 0.06
Yala 1990 194 04 19 1.1 0.8 0.3 0.2 0.12 0.08 0.04 0.03
Maha 1990/91 369 10 29 3.5 1.0 0.2 0.4 0.26 0.07 0.02 0.04
Average Ec. Values 0.19 0.10 0.04 0.04
P - Plough farming B - Bund farming
S - Strip mulch farming G - Graded alley cropping

The results showed that the average Ec. value was reduced from 0.19 to 0.10 due to
bunding instead of ploughing. The value could be reduced further down to 0.04 when
practices such as strip mulch and graded alleys were adopted. The reported Ec values
vary with the season though the soil and crop management practices are same because the
vegetation formed during seasons was different. Thus, it is logical to postulate that the
effectiveness of the soil and crop management or Ec varies with the seasonal influence on
the vegetation which determines the kinetic energy of the overland flow. Thus, an
attempt was made to find the nature of relationship between the Ec values of ploughed
land and the other management practices observed during five seasons for use in the
modelling process. The regression analysis shows that there is a significant linear
relationship of Ec values between these practices. The relationships are:
Ec (B) = 0.54 . Ec (P) - 0.006, (r2 = 0.82**)
Ec (G) = 0.15 . Ec (P) + 0.001, (r2 = 0.89**)
Ec (S) = 0.18 . Ec (P) - 0.003, (r2 = 0.62**)

where, Ec (B) is the erosion coefficient of bunded land, Ec (P) is erosion


coefficient of ploughed land, Ec (G) is erosion coefficient for graded alley
cropping and Ec (S) is erosion coefficient for strip mulch farming.

The relationships show the relative effectiveness of bunding, strip mulch farming and
graded alley cropping with respect to plough farming on soil erosion process. They also
indicate that the Ec value would reach the maximum allowable level (0.006) in a certain
farming practice, if an additional measure could support it. Nevertheless, the above
relationships could be used to compute the resultant Ec value, when a management
method is combined with other erosion control measures. The findings were used in the
development of Soil Conservation Impact Assessment (SCIA) model.

7
The SCIA model

The SCIA model was developed theoretically on the basis of modified USLE and
practically on the basis of the observations made of experiments described above. A
sensitivity analysis was carried out for input parameters of SCIA model. Table 4 shows
the results of the sensitivity analysis.

The sensitivity of land slope parameter increases with the increase in land slope for both
soil loss and soil life. The slope sensitivity value exceeds 1.0 at 3 percent land slope for
soil loss and at 4 percent land slope for soil life. Slope length is less sensitive to both soil
loss and soil life. The average sensitivity values are 0.5 for soil loss and 0.4 for soil life.

The sensitivity of soil life to soil depth is generally high. It shows an exponential
variation and the value is higher than 1.0 for any soil depth studied. The sensitivity
decreases exponentially up to the soil depth of 0.6 m and then reaches a constant level.
Soil depth is not sensitive to soil loss results in the model. The sensitivity of erodibility
is about 1.0 for both soil loss and soil life.

Table 4. Sensitivity ranges for input parameters.


Input Sensitivity value for
Parameter Range Soil depth Soil life
Land slope 1-8% 0.4 - 1.5 0.5 - 1.5
Slope length 20 - 300 0.5 0.3 - 0.6
Soil depth m 0.0 3.6 - 1.3
Erodibility 0.3 - 1.0 1.0 0.9
Mulch rate m
Weed 0.1 - 0.8 0.8 - 1.5 0.7 - 1.8
Rice straw 0.5 - 4.0 2.0 - 0.7 2.8 - 0.8
Gliricidia t/ha 0.9 - 1.6 1.0 - 1.8

A similar pattern is found for both soil loss and soil life outputs with different mulch
rates. High sensitivity values are shown for rice straw mulch at the rates below 1 t/ha, but
decreases rapidly up to 2 t/ha. Then the sensitivity is maintained at 1.2 for the range of
mulch rates from 2 to 3 t/ha and shows a further decline. Weed and gliricidia mulches
are less sensitive in lower and higher mulch rates, and most sensitive to the range 1 - 3
t/ha.

Model estimation for Mahailluppallama rain-fed upland farming situation


The programme was run using available data and under most prevalent conditions at
Mahailluppallama as an example. Following preliminary data were used.
Annual rainfall = 1386 (mm)
Monthly rainfall (mm)
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
84 54 84 172 86 17 34 38 80 246 254 237

Erodibility = 0.5, maximum land slope = 8 %, maximum slope length = 500 m, total
effective soil depth = 0.5 m and minimum soil depth required for crop growth = 0.3 m.

8
The soil loss tolerance level calculated by the model was 3.0 mt/ha/year and the
maximum soil life was estimated to 1000 years. These results indicate that any attempt to
reduce the soil loss lower than, or to increase the soil life to a value higher than the above
values would not be advantageous.

The results of computer simulation confirmed that the application of mulch alone would
not be adequate to reduce the field rate of erosion to the soil loss tolerance level. It is
evident that an application of one ton/ha of rice straw would have a similar effect as the
application of 2.7 t/ha of weed residue mulch or 3.3 t/ha of gliricidia prunings. Soil life
would be increased to 50 years, if the land is mulched either with rice straw at 1 t/ha or
with weed residue at 2.5 t/ha or gliricidia mulch at 3.5 t/ha.
Results of the computer simulation for land management practices are given in Table 5.
Table 5. Effect of land management practices on soil loss and soil life
Farming method Soil loss (t/ha) Soil life (years)
Traditional chena 11.1 41
Plough-farming 9.2 50
Bund farming 5.6 83
Strip mulch farming 2.7 177
Graded alley cropping 2.3 205

Data in Table 5 indicate that the soil life would be double even if the land is
conventionally bunded. A four time increase in soil life could be achieved by managing
the land with strip mulch method. The graded alley farming was found to be superior to
all practices as it can make an increase of soil life five times compared to traditional way
of farming.

A model simulation exercise was carried out to find the combined effect of mulch with
different land management practices on soil loss and soil life. The results are given in
Table 6. The simulation data reveal that the combination of land management practices
with mulching can reduce the soil loss to an appreciable level. If the permissible soil loss
rate is assumed as 3 mt/ha/year, then the required rate of mulch can be calculated by
using above information. The minimum requirement of mulch under different land
management practices is given in Table 7.
The results indicate that the conventional bunding must be combined with agronomic
practices such as mulching to attain the tolerable limit of erosion under general
conditions prevailing in the study area. It is not practicable to use gliricidia as a mulch in
conventionally bunded lands. With strip mulch or graded alley a small amount of rice
straw or weed residue would protect the soil to the required level. A mulch of gliricidia
of a very low rate (1 - 1.5 t/ha) is sufficient in these farming methods to protect the soil
adequately.

9
Table 6. Combined effect of land management with mulching (computer
simulation results)
Rate Bunded Strip mulched Graded alley
Mulch (t/ha) Soil loss Soil Soil Soil Soil Soil life
(t/ha) life loss life loss (years)
(years) (t/ha) (years) (t/ha)
Rice straw 1 4.3 108 1.3 376 1.5 346
2 2.9 163 0.9 618 1.1 494
3 2.1 226 0.6 961 0.8 648
4 1.7 296 0.5 1440 0.7 797

Weed 1 7.3 63 2.3 206 2.3 212


2 5.9 79 1.9 262 1.9 259
3 3.7 127 1.1 455 1.3 399
4 2.6 183 0.8 721 1.0 546

Gliricidia 1 10.7 43 3.5 135 3.2 146


2 7.1 64 2.3 211 2.2 216
3 4.8 97 1.5 333 1.6 314
4 3.7 127 1.1 455 1.3 399

Table 7. Required mulch rates under different land management methods.


Management Minimum mulch rate (t/ha)
practice Rice straw Weed Gliricidia
Bunding 2.0 3.5 <4
Strip mulch <1 <1 1.5
Graded alley <1 <1 1.0

CONCLUSIONS

The following conclusions were derived from this study.


1. Monthly erosivity (EV in mt.m/ha) can be estimated by using monthly rainfall (R in
cm) as EV = 1.14 R - 3.4 for the central dry zone of Sri Lanka.
2. The permissible soil loss rate is 3 mt/ha/year for rainfed upland soils.
3. Soil erosion can be reduced to one third when weed or rice straw mulch is applied at
the rate of 2 t/ha or crop residue (maize) or gliricidia mulch at the rate of 4 t/ha.
However, mulching alone is not adequate to reduce soil loss to the permissible rate.
4. The SCIA model suggests that the conventional bunding must be combined with
agronomic practices such as mulching to achieve the required level of soil
protection.
5. It is not practicable to use gliricidia as a mulch in conventionally bunded lands.
6. A small amount of rice straw or weed residue can protect the soil adequately when
strip mulch or graded alley farming is practised. A mulch of gliricidia of a very low
rate (1 - 1.5 t/ha) is adequate in these farming methods to conserve the soil
adequately.

10
REFERENCES

Abeyratne, E.L.F., 1956. Dry land farming in Ceylon. Trop. Agric., 112:191-229.
Brohier, L., 1975. Food and the People, Lake House Investments Ltd. p 39 - 43.
Dharmasena, P.B., 1992a. Soil erosion control measures for rainfed farming in the
dry zone of Sri Lanka. Ph.D Thesis (Unpublished), University of Peradeniya,
Peradeniya.
Dharmasena, P.B. 1992b. Magnitude of sedimentation in village tanks. Trop.
Agric., 148:97-110.
Dharmasena, P.B., 1994. Conservation farming practices for small reservoir
watersheds: a case study from Sri Lanka. Agro-forestry Systems, 28:203-212.
Morgan, R.P.C., 1986. Soil erosion and conservation. Longman Scientific and
Technical. John Wiley and Sons. New York.
Somasiri, S., J. Handawala, W.L. Weerakoon, P.B. Dharmasena, S.N.
Jayawardena., 1990. Rainfed Upland Farming for the Dry Zone. Agro-
technical Information Bulletin, Dept. of Agriculture.
Stocking, M.A. and Adam Pain., 1983. Soil life and the minimum soil depth for
productive yields: developing a new concept. University of East Auglia.
Discussion paper No. 150.
Tennakoon, M.U.A., 1980. Desertification in the dry zone of Sri Lanka. Central
Bank of Ceylon Monograph.
Wishmeier, W.H., and D.D. Smith., 1958. Rainfall energy and its relationship to
soil loss. Transactions of American Geophysical Union. 39(2):285-291.
Wishmeier, W.H., and D.D. Smith., 1978. Predicting rainfall erosion losses. USDA
Agric. Res. Serv. Handbook 537.

11

You might also like