Title Page
 1. Introduction and literature...
 2. Research concepts and backg...
 3. The katumani case study
 4. Presentation of research findings,...
 5. Development of farm practice...
 6. Summary of farm practice...
 7. Impact of recommendations
 8. Value of forecast informati...
 9. Conclusions
 List of Figures

Title: Impact of weather analysis on agricultural production and planning decisions for the semiarid areas of Kenya
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00072246/00001
 Material Information
Title: Impact of weather analysis on agricultural production and planning decisions for the semiarid areas of Kenya
Physical Description: 46 leaves : ; 28 cm.
Language: English
Creator: Stewart, J. Ian
Hash, Charles T
Publication Date: 1980?
Subject: Meteorology, Agricultural -- Kenya   ( lcsh )
Agricultural productivity -- Kenya   ( lcsh )
Crops and climate -- Kenya   ( lcsh )
Genre: government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
non-fiction   ( marcgt )
Spatial Coverage: Kenya
Bibliography: Includes bibliographical references (leaves 43-46).
Statement of Responsibility: J. Ian Stewart and Charles T. Hash.
General Note: Cover title.
General Note: Typescript.
Funding: Electronic resources created as part of a prototype UF Institutional Repository and Faculty Papers project by the University of Florida.
 Record Information
Bibliographic ID: UF00072246
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 76877014

Table of Contents
    Title Page
        Title Page
    1. Introduction and literature review
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
    2. Research concepts and background
        Page 6
        Page 7
    3. The katumani case study
        Page 8
        Page 9
    4. Presentation of research findings, and discussion
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
    5. Development of farm practice recommendations
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
    6. Summary of farm practice recommendations
        Page 29
        Page 30
    7. Impact of recommendations
        Page 31
    8. Value of forecast information
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
    9. Conclusions
        Page 39
        Page 40
        Page 41
    List of Figures
        Page 42
        Page 42a
        Page 42b
        Page 42c
        Page 42d
        Page 43
        Page 44
        Page 45
        Page 46
Full Text

Impact of Weather Analysis on Agricultural Production

and Planning Decisions for the Semiarid Areas of Kenya

J. Ian Stewart1 and Charles T. Hash2

1. USDA/USAID Agrometeorologist, Kenya Agriculture Research Institute,

Muguga, Kenya.

2. USAID Assistant Agricultural Development Officer, Nairobi, Kenya.


4... -'*


Record population growth rates require cultivation of food crops by

smallholder farmers in ever drier zones of Kenya's semiarid lands. Rainfall

is limited, variable and unpredictable, but maize, widely known for suscep-

tibility to drought, remains the staple and favorite food crop.

A case study is presented in which "effective rainfall" for Katumani

type maize, grown at Katumani, Machakos District, is evaluated for each of

the 48 seasons in the 24-year record. The newly developed analysis takes

into account rainfall, evaporation, soil depth and water holding capacity,

and growth characteristics of the crop influencing water uptake and yield.

Conclusions are:

1. The analysis evaluates suitability of a given crop for production in

any location where rainfall and evaporation records are available.

2. Dates of onset of the rains at Katumani are sortable into periods

termed "early" (implies expectation of high to medium water adequacy for

maize production), "late" (medium to low expectation), and "too late" to

recommend planting. These expectations determine recommendations for initial

seed and fertilizer rates.

3. Although unpredictable earlier, the rainfall pattern following onset

soon sorts itself into one of three categories of water adequacy for maize

(high, medium or low). Recommendations for thinning to final plant densi-

ties, and for adjusting nitrogen fertilizer rates through sidedressing are

based on the perceived category.

4. Approximately two months before harvest, total season effective rain-

fall can be estimated, and predictions of yield provided to farmers, econo-

mists, and other planners concerned with food supplies .

5. An estimate is presented of the value of meteorological information

to maize production in Machakos and Kitui Districts.


1 -

1. Introduction and Literature Review

Kenya today leads the world in population growth rate. Consequently,

the high rainfall areas which are less than 20% of the total, have for

some time been overcrowded. The result is a fast moving expansion of

smallholder farming into ever drier zones.

The new farming communities and their agricultural advisors have no

backlog of relevant experience to guide them toward the crops and prac-

tices most suited to stabilization and maximization of their production

of food. There is a recognized need for a major research effort to speed

the flow of practical information to the people. Thus, the top priority

in Kenya's current development plan is research and development of the

semiarid lands.

In accord with the above, the USAID/KARI (Kenya Agriculture Research

Institute) project was initiated in October, 1977 under the title

"Research on Cropping Systems for the Marginal Rainfall Areas", with

headquarters at Muguga. It was extended and expanded in October, 1979

under the title "Dryland Cropping Systems Research Project".

The present project is coupled into a larger "program" together with

two similarly oriented FAO/UNDP sponsored projects headquartered at the

Katumani National Dryland Farming Research Station in Machakos District.

The aim of the program is to introduce crop varieties and practices, i.e.,

cropping systems, which will maximize yields and stabilize food production

in areas where mean annual rainfall lies within the range of 500-800 mm.

The pattern of rainfall is bimodal, with the annual amount nearly

evenly divided between two distinct seasons. These are termed the "short

- 2 -

rains" and "long rains", with peaks in November and April respectively.

About 90% of the total falls in the cropping seasons, which in essence

reduces the range of mean annual rainfall useful for crop production

to 450-750 mm, and of mean seasonal rainfall to 225-375 mm. Needless to

say, the portion of gross seasonal rainfall actually utilized by the crop

will vary with circumstances, but will always be less than 100%.

Thus, shortage of water is clearly the limiting factor in cropping

the project areas. But the most vexing problem is not water shortage per

se rather it is the tremendous variability in rainfall from year to

year and season to season. In planning for the coming season there is

currently little or no predictability to the date of onset of the rains,

their amount, distribution or duration.

This paper presents a newly developed methodology for analysing the

rainfall record of a given location, together with other weather, soil,

management, crop, and economic factors, to determine for a specified crop

how it should have performed season by season throughout the period for

which rainfall data are available. This involves separate estimates for

each season of "effective rainfall" for production of the study crop, the

associated crop yield, and the economic returns.

The goals of the analysis are to assess:

1. Suitabilities of crops for cultivation in different areas.

2. Criteria by which farmers and their advisors can interpret current

rainfall events for selecting crops to plant, and for decision

making on levels and types of inputs, and on practices, including

methods and timing of land preparation, times of planting, initial

seeding and fertilization rates, and thinning time adjustments of

plant densities and fertilizer levels.

3 -

3. Criteria for predicting total season effective rainfall well in

advance of maturity and harvest.

The effective rainfall prediction will form the basis for predicting

crop yields, using methods developed by the principal author and colleagues

(Stewart and Hagan, 1969; Stewart, 1972; Stewart and Hagan, 1973; Stewart,

-Hagan and Pruitt, 1974; Stewart, et al, 1975; Stewart, Hagan and Pruitt,

1976; Stewart, 1977; Stewart, et al, 1977; Stewart and Wang'ati, 1978),

and adopted by the Food and Agriculture Organization of the United Nations

(Doorenbos and Kassam, 1979). The yield predictions will be of use not

only to the farmers themselves, but to regional and national planners who

must cope with the awesome responsibilities of assuring adequate food


Considerable research applicable to various questions implied in the

goals above has been carried out in East Africa. Statistical methods of

placing confidence limits on monthly or shorter term rainfall means have

been developed to a high level (Glover and Robinson, 1953a,b; Robinson and

Glover, 1954; Manning, 1950, 1956, 1960). Maps have been published for all

of East Africa showing rainfall amounts at probabilities of 10% and 20%

(East African Meteorology Department, 1961). Walker and Rijks (1967)

computerized Manning's method for rapid evaluation of confidence limits of

20-day period rainfall expectations. However, Woodhead, Waweru and Lawes

(1970) point out the high cost of computation per rain gauge site, and

establish that the variability computed for a single well chosen site is

representative of "quite a large neighborhood". They develop case studies

for areas of 1,000 km2 each around the maize producing centers of Kitale

and Machakos. Their findings hold promise for development of a workable

system for extending guidelines developed in the present study to the

farm level.



Relations between water requirements of crops and evaporative condi-

tions have received considerable attention since publication of Penman's

(1948) classical paper. Studies of evaporation based on his approach have

greatly aided understanding (McCulloch, 1965; Woodhead, 1968), and have

led to publication of evaporation maps for East Africa (Dagg, et al, 1970).

Additionally, a number of studies have included direct measurement of

evaporation using pans, many of which were installed in a network of agro-

meteorological stations more than two decades ago (McCulloch, 1965; Dagg,

1969; Wang'ati, 1972). A recent study for Machakos and Kitui Districts

(Braun, 1977) combines rainfall and Penman Eo values to calculate pro-

babilities that rainfall during the principal growing seasons will be

less than the equivalent of 2/3 Eo and 1/2 E Crop failure is assumed in

the latter case, while 2/3 Eo is taken as an approximation of the water

requirement of a typical crop.

Dagg (1965) recognized that crop water requirements and crop ability

to extract to extract soil water through deepening and proliferation of

roots, both change with growth stage as the season advances. And that the

amount of extractable soil water also depends on soil depth and water

holding capacity per unit depth. The precision required in determining the

water balance could not, however, be easily achieved in monitoring field

soil profiles alone. Research work, therefore, started on the design and

installation of weighing lysimeters (Glover and Foresgate, 1964; Foresgate,

Hosegood and McCulloch, 1965). Such lysimeters have since been used for

determining growth stage related water requirements of sugarcane (Blackie,

1969), tea (Dagg, 1970), maize and beans (Wang'ati, 1972), and bananas

(Nkedi-Kizza, 1973).

Stewart and Mugah (1979) have used a similar lysimeter to determine

water requirements of Katumani Composite B maize, the subject crop for

- 5-

this paper. Weekly values of maximum evapotranspiration (ET ) have been

determined throughout the season. These have been related to Class A pan

evaporation rates determined simultaneously at Muguga, to form transfer-

able ratios of ET (maize)/Epan for use in estimating water requirements

of this important food crop in other locations or in different seasons

in which evaporative conditions differ. With reference to transferability,

it is of interest to note that the ratios found at Muguga are nearly

identical to those determined for two other maize cultivars at Davis,

California (Stewart, 1972; Stewart, et al, 1977).

Rooting habits of crops and consequent patterns of soil water extrac-

tion are an important subject in the present study. Such studies with

Katumani maize have recently been completed in three locations in Kenya,

and by Mr. J.'O. Mugah at Davis, California where comparisons with other

maize cultivars can be drawn (Stewart, 1972; Stewart, et al, 1977).

Similarly, yield responses to different levels of water deficiency

are an important topic of research in the project, and these too have

been completed for Katumani maize. The transferable "water production

functions" produced form the basis for predicting yield in different

effective rainfall settings.

Other research topics pursued in East Africa, and of direct interest

to the present study, include relationships between maize yield, rainfall

and plant densities (Dowker, 1963), and a recent study on effects of

time of thinning (Allan and Laycock, 1979). Additionally, Semb and Robin-

son (1969) quantify the naturally occurring nitrogen levels at the

Katumani station and elsewhere in Kenya, while Semb and Garberg (1969)

present effects of planting dates and nitrogen fertilizer rates on maize

yields. Okalebo (1977) reports more recent findings regarding phosphorus

and nitrogen fertilizer effects on maize yields at several sites. Both


6 -

plant density and nitrogen fertilizer effects on yields of Katumani maize

have been studied under varied water levels in the present project.

Finally, attempts have been made, but with limited success to date,

to forecast the onset of the short rains (Gichuiya, 1968), and the amount

of the long rains at Nairobi through correlation with earlier occurrences

at Iringa, Tanzania (Cocheme and Zazzara, 1969). Future efforts are plan-

ned within the project to improve such forecasts.

2. Research Concepts and Background

A simplified water balance calculation is used to estimate actual

evapotranspiration (ETa) by the crop in different seasons, separately

for each 10-day period from planting to maturity, a total of 120 days.

Rainfall from the date of onset to planting is assumed to be either

stored in the root zone for future extraction, evaporated before planting,

or, if in excess to the soil storage capacity, lost to percolation below

the root zone.

To make the calculations one must combine findings from research on

the particular crop with measurements (or estimates) of weather, soil, and

management taken at the planning site. From research the needs are a. ra-

tios of ET /E for each period, b. a growth stage related time and depth

function of maximum soil water extraction, determined when the crop is

under stress, and normalized by relation to the water holding capacity of

the soil (Stewart, et al, 1974).

From the planning site we need a. the daily rainfall record, b. the

daily Class A pan evaporation record or equivalent, c. a one-time meas-

urement of the soil depth and field capacity, and d. a reasonable basis

for assuming runoff will be prevented, weeds will be controlled, and the

seeding rate will be sufficient to produce a stand which can fully utilize

the rainfall.

* Vt



Meaningful guidance to the farmer requires translation of effective

rainfall into estimates of crop yield, and this in turn requires interpre-

tation in terms of gross and net economic returns. Just as effective rain-

fall depends on factors other than total rainfall, crop yield depends on

more than effective rainfall even when we assume all other growing con-

ditions are satisfied and water is indeed the limiting factor.

To estimate actual yield we must first answer the question "water is

limiting from what initial yield level?" In other words we must know what

yield might have been with adequate water. This depends on the crop type

and variety, but, like the water requirement, it also depends largely on

evaporative conditions during the growing season being considered. For

example, evaporation rates are decidedly higher in the short rains (late

spring and summer) than in the long rains (fall and winter). Hence, maize

water requirements and potential yields when water is adequate are both

greater in the short rains.

Actual crop yield depends on the degree of satisfaction of the crop

water requirement (ETa/ETm), the potential yield, and the way the parti-

cular crop responds to different fractional levels of drought. Requirements

for estimating ETm and ETa/ETm have been covered above. Additional needs

for yield prediction are two functional relations which must be researched

for the crop species in the first instance, and the variety in the second.

The first of these is illustrated in Figure 1 for maize cultivars of

approximately 120 days maturity.

.Figure 1 shows a relation between maximum yield attainable with

climatically adapted maize cultivars and energy (primarily sunshine and

ambient temperature) available for growth. In the figure energy available

is represented by the mean daily rate from planting to maturity of

evaporation from an unscreened Class A pan in green grass surrounds.

- 3 -

Results from ten experiments are shown, drawn from three distinct

climates in the USA, and representing three different maize cultivars,

plus a wide range of weather conditions in six experiments at two loca-

tions in Kenya, all with Katumani maize.

The second crop function to be quantified in research is the "yield

reduction ratio" or YRR defined by Stewart, et al (1975), as the percen-

tage decline in yield below the attainable maximum, relative to the per-

centage by which ETa is less than ETm. The YRR describes a genetic char-

acteristic of the variety, coupled with the degree of crop canopy cover

which is largely a management factor (seeding and thinning practices),

and influenced also by the effects of rainfall frequency on soil surface

evaporation. For Katumani maize, considering the above, the YRR is 1.45.

Thus, for example, if ETa is 20% less than ET actual yield will be

1.45 X 20 = 29% below the attainable maximum.

3. The Katumani Case Study

The study extends over the available 24-year rainfall record, which

is continuous since October, 1956. Altogether it covers 48 growing seasons,

including the short rains from 1956 1979, and the long rains from 1957 -

1980. Since the short rains maize crop may still be in the field when the

long rains begin, the discussion here assumes different plots of land are

planted in the two seasons. Note that this would not have to be the case;

current research by the project agronomist is demonstrating that "relay"

planting of maize into standing maize may actually be advantageous in some

cropping systems.

Records of evaporation from a screened Class A pan are available from

the Katumani station for the past seven years. Actual data were therefore

utilized for the fourteen seasons of record, while 7-year means were used

to estimate evaporation in the other thirty four seasons. The dates span-


9 -

ned by each season were judged from the dates of onset of the rains, con-

sidering also the duration of the initial rainfall eventss, a period for

land preparation and planting, and the 120-day maturity of Katumani maize.

In regard to evaporation rates, at a given location they are strongly

linked to the annual cycle of weather patterns, and unlike rainfall, are

not highly variable from year to year. However, there is some variation to

contend with, and it arises mainly from the rainfall variation, because

rainfall also means reduced sunshine and temperatures. Therefore, the

practical outcome is that in wet seasons reduced evaporation rates lead to

reduced water requirements for the crops, hence to improved ratios of

ETa/ETm. Contrariwise, drought in dry seasons is exacerbated by increased

water requirements.

Soil properties, mainly depth and water holding capacity, and the

ability of Katumani maize to extract soil water from different depths at

different growth stages, have been studied in detail in a series of five

full term experiments carried out at Katumani station. Tremendous natural

variability is found in the experimental field, with depths ranging from

45 255 cm, and total water holding capacities, i.e., extractable water

plus that not available to the crop, ranging from as little as 13% by

volume to as much as 40%. The findings are incorporated into the present

analysis, and can be generalized to most locations in Kenya's semiarid


In the same experiments, and in two concurrent experiments in the

lysimeter field at Muguga, and at Kambi ya Mawe (a hotter, drier sub-

station of Katumani) determinations have been made of a. the potential

yield of Katumani maize with adequate water in different evaporative con-

ditions (see Figure 1), b. the effects on yield of different levels of

drought, i.e. the YRR, and c. the interactions on yield of levels of

water supply, plant population and nitrogen fertilizer. Findings from

- 10 -

this research are built into the effective rainfall analysis, and into

the ensuing recommendations and farm practice guidelines presented here.

4. Presentation of Research Findings, and Discussion

Overview of Rainfall

Mean annual rainfall at Katumani in the past twenty four years has

been 709 mm, with a low of 425 mm in 1976, and a high of 1120 mm in 1963.

Variation in shorter periods is very much greater as shown in table 1.

The table starts with October rather than January to more clearly match

the rainfall pattern with cropping seasons.

TABLE 1 Katumani Monthly Rainfall Means, and Recorded Extremes


Mean 29 160 93 55 42 84 145 68 9 6 4 14

Low 0 27 12 0 0 0 20 11 0 0 0 0

High 183 489 219 202 177 169 346 151 35 36 20 134

Table 1 illustrates the following points:

a. The relative variability of rainfall in any one month is much greater

than for the annual total. It is only in November and December, and

again in April and May that any rain at all seems guaranteed, and this

can be an insignificant amount. All other months have experienced zero

rainfall at least once in the twenty four years. On the other hand, all

months except June, July and August have experienced heavy rains, rang-

ing from two to six times the normal. September should also be grouped

with the dry months of June-August, because the 134 mm high shown occur-

red only once, with the next highest figure being 42 mm.

b. The most constant factor in the Katumani rainfall pattern is the near

total dryness in the cool cloudy months of June-September. Obviously,

crops in the field in this period will be totally reliant on stored soil

- 11 -

water. This very much applies to long rains maize which matures at the

earliest around June 20, and latest about August 25. This constitutes a

strong argument for planting maize as early as the rains allow.

c. The other dry period occurs in January-February, which, being summer,

is the hottest time of the year, and during which the highest ETm rates

occur. These months form the latter half of the short rains growing sea-

son, and generally speaking, as they go, so goes the season. January and

February can be good, moderate, or poor rainfall months, with maize

yield expectations in accordance.

d. The fairly reliable wet periods are November to mid-December, and late

March through early May. However, as noted, any of these periods can

be very dry on occasion.

Overview of Evaporative Conditions

Evaporative conditions are much more uniform from year to year than

is rainfall, and also are quite uniform over rather large areas of similar

terrain. A relatively modest amount of carefully taken data, when combined

with research findings for a specified crop variety, can suffice as a basis

for estimating crop water requirements and potential crop yields.

However, such estimates must be linked to specific growing seasons,

designated by planting dates and maturity times, because evaporative condi-

tions change rapidly from one part of the year to the next. Table 2 shows

monthly means and extremes of evaporation rates experienced in the past

seven years at Katumani. These are from a Class A pan, screened with a 25 mm

square mesh, sited in a hectare of grass which dries in July-September, and

in some years January to early March.

I- I

12 -

TABLE 2 Katumani Monthly Evaporation Means, and 7-Year Extremes, mm/day


Mean 6.3 5.0 4.8 5.8 6.4 6.4 5.2 4.0 3.4 3.1 3.7 5.1

Low 5.8 4.2 4.0 4.3 5.2 4.5 4.5 3.5 3.0 2.7 3.2 4.3

High 7.1 5.9 5.6 7.4 7.6 8.2 6.1 4.8 4.3 3.9 4.5 5.5

Taken all together, the evaporation rates at Katumani, and throughout

arable Kenya, are quite moderate by temperate zone standards. The principal

reasons for this are the uniform 12-hour days, and the cloudiness inherent

in tropical rainfed cropping situations. The practical implication is that

sunshine hours will be the ultimate limiting factor for yields of many

crops, certainly including maize, beans, grain sorghum, and others which

thrive in warm, sunny climes.

From table 2 we may deduce that short rains evaporation rates should

average from 5.5 to 5.9 mm/day over the 120-day maize season, depending on

planting date, while long rains season means will decline from 5.1 for a

February 20 planting to 3.6 for an April 20 planting. To relate to Figure 1

these rates must be divided by .93, i.e., unscreened pan = screened pan/.93.

Thus potential yields in the short rains are normally in the range of 7.8

to 8.4 t/ha, while comparable figures in the long rains decline from 7.2 to

4.8 t/ha with later onset of the rains. This presents a second compelling

reason for planting as early as the rains allow.

The Onset of the Rains

One of the more important developments which has emerged from the

effective rainfall analysis is the quantitative determination of which rain-

fall events should be accepted by the farmer as the "onset" of the rains

for planting purposes. Additionally, the analysis indicates the earliest

and latest dates acceptable (riskwise) for the onset. Rain preceding the

I., .

13 -

earliest acceptable date carries with it too great a risk of a long dry

spell ta fo.U.Qwr,,while onset too, late..means toQ. short- a.season. far maize


For reasons to be more fully developed, the above considerations are

most important in the long rains. Note that the onset events and dates in-

dicated by the analysis may or may not coincide with the official Meteoro-

logical Department announcement of the arrival over Kenya of the Inter-

Tropical Convergence Zone. In many instances, long rains maize planting

should precede this announcement. Table 3 shows, for both seasons, how mean

expectations for rainfall and evaporation change with the date of onset,

and how these changes affect both maize water requirement and potential

yield, and additionally affect both rainfall and yield actually expected.

In the short rains, dates of onset have ranged from October 20 to

November 23, a span of 35 days in which expected gross rainfall in the maize

season decreases at a rate exceeding 4 mm/day. However, once begun, the pat-

tern of the short rains is frequent falls through at least the first 50 days,

which in heavy rainfall years results in rapid filling of the soil storage

capacity, with runoff or deep percolation of the excess. In the end then,

expectations for effective rainfall decrease at the much lower rate of about

1.5 mm/day.

Simultaneously, evaporation rates (average for the maize season ahead)

are rising, hence the crop water requirement also rises. Thus the expected

water adequacy (ETa/ET ) decreases with.delayed onset, as does expected yield.

In the long"rains, dates of onset have ranged from February 10 to April

15, a span of 65 days in which expected gross rainfall decreases at a rate

exceeding 3 mm/day, and effective rainfall decreases at a similar rate. But,

due,.to falling evaporation rates, the maize water requirement is also de-

creasing. However, the rate of decrease is about 2 mm/day, which does not

Table 3. Expectations for rainfall and evaporative conditions during the Katumani maize growing season
as related to the date of onset of the rains. Consequent estimates of crop water requirements
and effective rainfall, plus yield potentials and actual yield expectations assuming best
management conditions.

Expectations Based on 24-Year Record

(ETA, mn







Research Based Estimates

+ r-


Crop Water
(ETM, mm)



Yield of




Crop Water



Actual Yield
of Maize



* Standard Class A Evaporation Pan, Unscreened (Screened = .93 Unscreened)



Date of
of Rains


Feb 10
Feb 20
Mar 1
Mar 11
Mar 21
Apr 1
Apr 11



- 15 -

fully compensate for the negative effects of delayed onset. These effects

are a rapid decrease in potential maize yield, and a still more rapid de-

crease in the actual yield expectation.

The main lessons to be learned from the information in Table 3are:

a. The earlier the onset of the rains, the better the chance of a good

rainfall season.

b. It follows therefore that as quickly as the rains permit, the land

should be worked and maize planted.

c. Maize should not be planted in the short rains if the onset is after

November 23, nor in the long rains if onset is after April 3. The reason

may be seen in the ETa/ETm ratios in the column titled "crop water

adequacy". A satisfactory maize crop is not likely to be produced with

effective rainfall much below 50% of the water requirement.

Rainfall Events signalling Onset and Planting Time

(Short Rains)

The events) which should be accepted as the onset of the rains will

begin at the earliest on October 20. It will last from 1-11 days, and total

a minimum of 30 mm before it stops. Working the land within this period may

be feasible, but planting should await the 30 mm point. Typically, it will

rain for 4 or 5 days with no letup longer than 1 day in that period, and the

total amount will be about 50 mm. Following the onset ("date of onset'is the

first day of the period, whereas "onset" refers to the entire period) there

will be a "dry" period when planting should immediately take place. If heavy

rains resume quickly, field operations can be delayed for weeks. To clarify

the situation, out of 24 years, there were 3 in which the dry period lasted

only 2 days. It lasted 3 days in 5 years, 4 days in 6 years, and 5 days in

2 years. Thus the dry period lasted only 2-5 days in 16 of 24 years.

There is no way to view the onset and judge therefrom the length of

16 -

dry period expected. Therefore, planting must be accomplished just as quickly

as possible. Undoubtedly this is one reason farmers sometimes plow and plant

prior to the date of onset. However, this procedure has other drawbacks.

Small rains may germinate the seeds early and them leave them to wither;

pests may destroy the seed while it awaits the rains, perhaps for weeks; and

perhaps most damaging, a major flush of weeds generally accompanies the

germinating maize. If these are not expeditiously destroyed, they will dras-

tically reduce yields. When planting follows onset, the initial weed flush

is destroyed in seedbed preparation.

(Long Rains)

The rainfall event acceptable as the onset will begin on or after Febru-

ary 10, and will total at least 40 mm before the dry period when planting

should take place. The onset will last 1-8 days, then be followed by 2-28

days of essential dryness. Typically, onset lasts 5 days and totals 68 mm,

then is followed by a 10-day dry period. As in the short rains, planting

immediately after onset is very important for maximizing yield.

Quantification of Effective Rainfall Categories

The most interesting discovery in the course of developing the analysis,

and one which appears to have far reaching implications for scientifically

guiding farm practices and levels of inputs, as well as for predicting yields

far in advance of harvest is as follows:

Despite our pre-season inability to predict when and in what amounts the

rain will come, the picture clarifies itself rapidly once onset occurs. Each

season's rainfall proceeds to sort itself into one of three distinct patterns,

and remains there. This applies in both the short and long rains, but earlier

in the latter, say 20-40 days from the date of onset, while 50 days are re-

quired in the short rains. Thus, maize farmers can be given timely field

guidance on such important matters as adjustment of nitrogen fertilizer rates,

- 17 -

and suitable plant populations to thin to, in accordance with the pattern

of early season rainfall.

Additionally, total effective rainfall for the season can be closely

estimated two months or more before maize harvest. The estimates can then

be used to make yield predictions for the benefit of farmers, agricultural

extension advisors, and planners at all levels, whether economic, social,

or political.

The 24-year rainfall records for the short and long rains respectively,

are presented in the next two sections. Each record is analysed to show how

to quantitatively interpret early season rainfall for the purposes above.

Short Rains; the Record and its Interpretation

Figure 2 shows when planting should have occurred in all the seasons

since 1956, save 1970 when onset was in mid-December, far too late for

planting maize. If one had planted then, gross rain would have totalled

only 160 mm, and effective rain only 140 mm. These are the smallest amounts

in the record.

Figure 2 also shows the total effective rain in each season, and sorts

them into three categories, designated as "A", "B", or "C" type seasons.

"A" type seasons are those in which the water adequacy was high, ranging

from 0.68 1.0. Good crops of maize would be expected in these seasons,

which, if given optimal management other than water, should range between

4.4 6.7 t/ha of grain at 15.5% moisture content. In "B" seasons water

adequacy ranged from .50 .61, with possible yields of 2.3 3.4 t/ha. In

"C" seasons water adequacy was .26 .48, with possible yields of 0 2.0 t/ha.

Finally, Figure 2 contains three arrows showing the trends in maize

water requirements (upward), gross rainfall expectation (downward), and

effective rainfall expectation (downward, but less rapidly). As may ba noted,

the correlation between effective rainfall and the starting date of the

- 18 -

season is not at a high level, but it does exist, hence dictates early

planting. The rising trend of crop water requirement is another argument

because it means the same rainfall will be less adequate as the season


Figure 3 shows the three distinct patterns of cumulative effective

rainfall for A, B, and C type seasons, from planting to maturity. The

effective rainfall present at the time of planting is the amount which fell

during onset, less 20 mm assumed to have evaporated in seedbed preparation.

Also shown in Figure 3 are the time patterns of cumulative water re-

quirements for the three types of seasons. The greatest requirement is for

the lowest rainfall season (type C), due to higher evaporative rates.

Each of the season types occurred in eight of the twenty four years,

i.e., 1/3 of the time. The framing lines above the patterns represent the

extreme high value at any given maize growth stage during the eight years

in question. Similarly, the line below the C type pattern represents the

extreme low values for the seven years in which maize planting would have

been recommended (1970 onset too late for maize). The low values for A and

B type seasons are shown only for the last 20 days of the season where the

three patterns become totally distinct.

Figure 3 shows effective rain in the first 50 days of the short rains

is similar in all years, the lowest averaging 2 1/2 mm/day, and the highest

4 mm/day over that period. However, the gross rainfall figures 50 days from

date of onset show differences sufficiently great to categorize most years

properly as A, B or C, and thus to provide guidelines for farmers to use

in thinning and fertilization practices.

Details of the guidelines are developed later, but pertinent comments

are these:

a. Five of the eight A type seasons would have been correctly categorized,

- 19 -

while the other three would have been treated as B seasons. Research

plot expectations in the eight years would total 45.5 tons of maize grain

from one hectare of land. In farming practice, a good manager following

the guidelines closely should have realized an estimated 33.3 t/ha.

b. Of the eight B seasons, the system would have judged one as A, five as B,

and two as C. Following those guidelines, a good farmer should have har-

vested a total of 18.4 t/ha. For comparison, research expectations would

be 22.1 t/ha.

c. Of the eight C seasons, five would have been correctly judged, and the

other three treated as B type. Total yield in research plots would be

estimated at 8.3 t/ha, but farm expectations would be small in the better

of these years, and nothing in the poorest seasons.

d. Summarizing, the recommendations would have led to planting of maize in

twenty three of the twenty four short rains seasons. Sixteen moderate to

good crops should have been produced totalling 51.7 t/ha. Additionally,

six C type seasons should have produced very small crops, with the other

two being complete failures.

Long Rains; the Record and its Interpretation

Considering the criteria for onset earlier defined, Figure 4 shows that

the first opportunity for planting maize in the long rains may be anytime

from mid-February to late April. However, planting is not recommended if

the date of onset is after April 3. The reason may be seen in Figure 4 which

shows data points for four seasons at the April 21 planting date. Of these,

three were C type seasons, while the fourth was very low in the B range.

Certain crops other than maize would present better prospects in these


The figure shows that of the twenty four seasons, twelve rate as A,

six as B, and six as C. Adequacy of water in category A is .77 1.0, in B

20 -

it is .48 .74, and in C it is .21 .43. Research plot yields should be in

the range of 4.0 7.2 t/ha in category A, 1.7 3.4 t/ha in category B, and

0 0.9 t/ha in category C.

As in the short rains, the expectations for gross and effective rain-

fall drop swiftly as the date of onset gets later. However,,in this instance

the actual correlation is much greater than in the short rains. The reason

is seen in Table 1 which shows the rains are never significant after May,

and can at times end early in May.

Unlike the short rains, the water requirement of maize falls with later

onset in the long rains. This explains why expected water adequacy declines

as slowly as it does with later onset.

Figure 5 shows the cumulative season effective rainfall patterns for

categories A, B, and C during the fifteen years in which date of onset was

"early", i.e., from February 10 to March 15. Unlike the early short rains

patterns, C type seasons in the long rains separate themselves from the bet-

ter seasons very quickly, and the overlap between A and B type seasons is

very little indeed. This then offers a perfect situation for close guidance

of fertilization and thinning practices at the farm level, in accordance

with early season rainfall.

Figure 6 deals with the remaining nine years of the long rains record,

in which the date of onset was "late", from March 16 to April 15. Four of

these seasons were C type, Four B type, and one A type. In Figure 6 the lat-

ter season is lumped into the B category, and in fact forms the line which

represents the upper extreme of that category.

Early season rain patterns are different when onset is late than when

it is early. Figure 6 shows the late onset rains come with a rush, similar

to the short rains pattern. However, unlike the short rains, there is again

an early and complete separation of B type patterns from C type most help-

ful for determining farmer guidelines.

'** .i


21 -

As before, the long rains recommendations are detailed later, but some

pertinent comments are as follows:

a. Categorization of the season is more accurate in the long rains than the

short, and can be accomplished earlier, at 30 40 days after date of

onset, as opposed to 50 days in the short rains.

b. Of the twelve A type seasons, eight would have been judged correctly,

and the other four as B type. Total farm yield should have been 44 t/ha.

c. All six B seasons would have been judged correctly, but the one with date

of onset after April 3 would not have been planted. Yield in the other

five seasons should have totalled 12.1 t/ha farmed.

d. All six C seasons would have been correctly judged, but the three with

onset after April 3 would not have been planted. Of the three planted,

two should have produced very small crops, and the third total failure.

e. Summarizing, the recommendations would have led to planting in twenty of

the twenty four long rains seasons. Seventeen moderate to good crops

should have been harvested, totalling 56.1 t/ha, with two more very

small crops as in "d" above.

5. Development of Farm Practice Recommendations

Recommendations developed in this study concern a. the earliest and

latest acceptable dates of onset of the rains, b. criteria by which to

judge an acceptable onset for maize planting, c. the rapidity with which

maize planting should proceed after onset, d. planting geometry, i.e., row

and plant spacing, e. the kinds and amounts of fertilizer to apply at

planting, f. when and how to categorize the type of season being experienced,

g. how to thin the stand in accordance *ith season category, h. at what

levels to sidedress nitrogen fertilizer in accordance with season category,

i. tied-ridging of soil to maximize rainfall capture, and j. weeding prac-

tices to minimize competition for water and nutrients.

- 22 -

Items a, b, and c above have received sufficient discussion in the

previous sections.

Planting geometry recommendations are based on needs for later adjust-

ment in accordance with season categorization. Desired final stands are

60,000, 40,000, and 20,000 plants per hectare respectively, for A, B, and

C type seasons. A principle followed in setting these population levels is

that the least effective rainfall expected in the given type of season

should be sufficient to enable each plant left in the field to attain more

than 60%, but less than 100% of its potential maximum weight (total dry

matter). Smaller plants may fail to maintain a normal "harvest index" or

ratio of grain dry matter to total.

Selection of nitrogen and phosphorus as the recommended fertilizer

elements is supported by reports in the literature, research by the prin-

cipal author in Kenya, field observations, and by current experiment

station recommendations. However, the latter are fixed at levels roughly

akin to those developed herein for type B seasons, thus are not subject to

adjustment in accordance with actual rainfall amounts. A purpose of this

study is to define procedures for making such adjustments, so as to avoid

non-productive expenditures in poor rainfall seasons, and to take proper

advantage of the opportunity for greater returns in high rainfall seasons.

Based on a lower requirement by the crop for phosphorus than for

nitrogen, and on preliminary experiments and observations by the author,

the recommendation herein is a fixed 20 kg/ha P205, applied at planting

time in all maize planting seasons. It is recognized that the need may be

greater than this in certain phosphorus fixing soils in the semiarid areas.

However, this rate appears adequate for Katumani type soils in light of

present information.

Nitrogen fertilizer recommendations are based on the principle of

matching probably available soil nitrogen to a conservative estimate of

- 23 -

expected effective rainfall. This procedure begins with the assumption

that something of the order of 30 kg/ha of crop available nitrogen is pre-

sent in each season, deriving from the combination of natural mineralization,

which accompanies the rains, and whatever residual fertilizer nitrogen may

be left over from the previous crop. This is sufficient nitrogen to enable

production of about 1.5 t/ha of maize grain.

Effective rainfall requirements for a yield of 1.5 t/ha are estimated

to 230 mm in the short rains, and 170-180 mm in the long rains. These

amounts correspond to C type seasons, which range up to 260 mm in the short

rains, and 180 mm in the long rains.

The recommendation of 30 kg/ha of nitrogen (N) in B type seasons assumes

70 % efficiency of utilization, sufficient to increase yield by 1.1 t/ha,

from the base of 1.5 on up to 2.6 t/ha. At this level, estimated effective

rainfall needs are 275 mm in the short rains, and 215-230 mm in long rains.

Overall ranges of effective rainfall in short rains are 260-350 mm, and in

long rains, 180-280 mm.

In A type seasons the soil is wetted more frequently and thoroughly,

and the efficiency of N utilization is assumed increased to 85%. Consequent

yield expectations are increased to 5.1 and 4.2 t/ha in short and long rains

respectively, with N applications of 80 and 60 kg/ha. Effective rainfall

required for these yields approximates 385 mm in short rains, and 300 mm in

long rains. Overall ranges of effective rainfall in A seasons are 350-460 mm

and 280-440 mm respectively in short and long rains.

The more general considerations which follow have been used to divide

fertilizer amounts between planting time applications, and those sidedressed

at thinning:

a. Phosphorus is to be applied at planting, in light of the general finding

it is of great importance to early root development.

I. .%

24 -

b. In A type seasons when higher yields are the goal, some applied N is

required before the season can be categorized, Thus, when onset is early,

and the chances of an A season good, N is recommended at planting. The

amount is that required by a B season, so, should that develop, no side-

dressing will be required. If a C season develops, the fertilizer is not

necessarily lost experiments show carryover is at a high level in the

semiarid areas.

c. When onset is late, and a B season the best reasonable hope, N should be

withheld until the season is categorized.

Planting time seed and fertilizer rates, procedures for categorizing

the season, and then for adjusting plant populations andNrates, are discus-

with reference to Charts 1, 2, and 3.

Chart 1 deals with the short rains. The uppermost boxes contain ranges

of dates of onset termed "early" and "late" implying a shift in rainfall

expectations, quantified in terms of probabilities in Table 5. The shift

calls for different planting time seed and N rates, and for different cri-

teria to categorize the season. Details of seeding rates and configurations

are found in Chart 3.

The second row of boxes in Chart 1 is preceded by the words "50-day

total rain from onset", followed by figures in the boxes which sort the

season into its proper category. It is at this time the evaluation can be

made. Later is too late for thinning and sidedressing. Earlier is desirable,

and is possible in seasons of very high or very low early rainfall. For exam-

ple, if onset is early, and rainfall at 40 days exceeds 238 mm, it is an A

type season, as shown in the third row of boxes. Following through, the

fourth row of boxes shows the final plant population should be 60,000/ha,

and the fifth box shows N should be sidedressed at 50 kg/ha. Just below the

boxes, planting time fertilizer rates are shown, equalling, in this example,

.Chart 1. Using the date of onset of the short rains to adjust seeding and fertilization rates at planting,
and using the actual rainfall total at thinning time 40 days later to adjust final plant
population and rate of nitrogen side-dressed.



Period Which Includes Date
of Onset of the Rains

50-Day Total Rain
from Onset I

Type of Season E

Plant Population
after Thinning L

Nitrogen to Apply
as Side-Dressing L

Fertilizer Applied
at Planting Time

Work Sequence at Thinnin

- 30 kg/ha N + 20 kg/ha P205 + - 20 kg/ha P205 -

ig Time: a. Weed and loosen soil. b. Thin plants. c. Side-dress N. d. Ridge
soil along plant rows. e. Tie ridges with soil blockages every 3m.


- 26 -

30 kg/ha N plus 20 kg/ha P205. The recommended sidedressing for an A season

will increase N to 80 kg/ha.

Chart 2 is for long rains evaluation and guidance, and once again the

date of onset is the initial point of reference. The separation of seasons

into their respective categories is both earlier and clearer in the long

rains, and may be judged in the period between 30 and 40 days after date of

onset. Regarding the latter as day 1, simply divide the total (not effec-

tive) rainfall since onset by the number of days, and compare the resulting

mean daily rainfall to the values in the second row of boxes.

For example, suppose at 38 days total rainfall is 114 mm. Mean daily

rainfall then is 114/38 = 3.0 mm/day. If date of onset were March 4 (early)

the season would fall in category B. However, if date of onset were March

20 (late) the season would fall in category C. Following this evaluation,

thinning and sidedressing should be done in accordance with specifications

in the fourth and fifth rows of boxes.

Details of seeding rates and patterns are shown in Chart 3, as are

patterns of plants as they should appear after thinning. The upper part of

the chart shows the seeding pattern for both the short and long rains when

onset is early. Rows are 75 cm apart with seed holes every 33 cm, and 2

seeds in each hole. The result with perfect germination would be 80,000


When onset is late, seeding should be as shown in Chart 3 for 60,000/

ha. Row and hole spacings remain the same, but as one proceeds down the

row, first 2 seeds are dropped, then 1, then 2, 1, 2, 1, etc. After the

season is categorized and thinning is to be done, the final plant patterns

desired for A, B, and C type seaso'is are as shown in the chart.

,Chart 2. Using the date of onset of the long rains to adjust seeding and fertilization rates at planting,
and using the actual rainfall total at thinning time 35 days later to adjust final plant
population and rate of nitrogen side-dressed.

Period Which Includes
Date of Onset of the

R, mm/day =
Total Rainfall/ R
Days Since Onset

Type of Season "A

Population of
Plants After 60,OC

Nitrogen to
Apply as 30 k!

Fertilizer Applied ,
at Planting Time

Work Sequence at Thinning Time:


- 30 kg/ha N + 20 kg/ha P205 -

a. Weed and loosen soil. b. Thin plants.
soil along plant rows. e. Tie the ridges


[Planting Not

Apr 4-Apr 15
-- iip Ib

S- 20 kg/ha P205 -

c. Side-dress nitrogen. d. Ridge
with soil blockages every 3 m.

- 28 -

Chart 3


As Planted
for Early Onset

As Planted
for Late Onset

Type "A" Seasons
Final Stand

Type "B" Seasons
Final Stand

33 cm
(2) (2) (2) (2) (2) (2) (2)
80,000/ha 75 cm

(2) (2) (2) (2) (2) (2) (2)

(1) (2) (1) (2) (1) (2) (1)

(2) (1) (2) (1) (2) (1) (2)

(1) (1) (1) (1) (1) (1) (1)

(1) (1) (1) (1) (1) (1) (1)

Type "C" Seasons
Final Stand

(1) (0) (1) (0) -


(0) (1) (0) (1) -

(1) (0)

- (1)

(0) (1) (0)

- 29 -

6. Summary of Farm Practice Recommendations

Short Rains Early Onset, October 20 to November 3

a. Land preparation, fertilization and planting

Plow early, before October 20 if possible, or in a lull between first

rains before determining they constitute onset (30 mm). At the latest,

plow immediately after onset.

Immediately following onset (and plowing), spread fertilizer evenly on

the surface at the rate of 20 kg/ha P205 and 30 kg/ha N.

Harrow to incorporate fertilizer and prepare seedbed for planting.

Plant 2 seeds in each hole as in Chart 3.

At 3 to 4 weeks after planting thin the stand to 60,000/ha as in Chart 3.

b. Categorizing the season, and final adjustment of population and N.

At 50 days, i.e. date of onset + 49 more, check total rainfall. If 238

mm or more, it is an A type season; if less it is B type.

If an A season, leave stand at 60,000/ha, and sidedress with 50 kg/ha N.

If a B season, thin stand to 40,000/ha (see Chart 3). No N is added.

Destroy weeds, loosen soil and pull it into ridges along the plant rows,

covering sidedressed N in the process. Tie the ridges together with

soil blockages to prevent rains running off.

Short Rains Late Onset, November 4 to 23

a. Land Preparation, fertilization and planting

* Plow early (as above for early onset).

* Immediately after onset and plowing, broadcast fertilizer at the rate

of 20 kg/ha P205'

Harrow to incorporate fertilizer and prepare seedbed for planting.

Plant 60,000 seeds/ha as shown in Chart 3.

At 3 to 4 weeks after planting, thin the stand to 40,000/ha as in Chart 3.

- 30 -

b. Categorizing the season, and final adjustment of population and N.

At 50 days check total rainfall. If it is 206 mm or more, the season

is type B; if 205 mm or less, it is C type.

If a B season, leave stand at 40,000/ha, and sidedress 30 kg/ha N.

If a C season, thin stand to 20,000/ha as per Chart 3. No N is applied.

Destroy weeds, loosen, ridge and tie soil as above.

Long Rains Early Onset, February 10 to March 15

a. Land preparation, fertilization and planting

Plow immediately after onset (40 mm).

Broadcast fertilizer at 20 kg/ha P205 and 30 kg/ha N.

Harrow to incorporate fertilizer and prepare seedbed.

Plant 80,000 seeds/ha as in Chart 3.

b. Categorizing the season, and final adjustment of population and N.

At 30-40 days after date of onset, determine mean daily rainfall by

dividing total rain by number of days. If 3.2 mm/day or more, the season

is type A; if in the range 2.2-3.1 mm/day, it is type B; if 2.1 mm/day

or less, it is type C.

If an A season, thin stand to 60,000/ha, and sidedress 30 kg/ha N.

If a B season, thin stand to 40,000/ha. No N is added.

If a C season, thin stand to 20,000/ha. No N is applied.

Destroy weeds, loosen, ridge and tie soil as above.

Long Rains Late Onset, March 16 to April 3

a. Land preparation, fertilization and planting

* Plow prior to, during, or immediately following onset (40 mm).

*-Broadcast fertilizer at 20 kg/ha P205.

* Harrow to incorporate fertilizer and prepare seedbed.

* Plant 60,000 seeds/ha as in Chart 3.

b. Categorizing the season, and final adjustment of population and N.


At 30-40 days after date of onset, determine mean daily rainfall. If

3.6 mm/day or more, the season is type B, if 3.5 im/day or less, it

is a type C season.

If a B season, thin stand to 40,000/ha, and sidedress 30 kg/ha N.

If a C season, thin stand to 20,000/ha. No N is applied.

Destroy weeds, loosen, ridge and tie soil as above.

Both Seasons Very Late Onset

If onset occurs after November 23 in the short rains, or April 3 in

the long rains, maize planting is not recommended.

7. Impact of Recommendations

It is estimated that if the above recommendations were followed, maize

yields as set forth in Table 4 might reasonably be expected across those

areas of Kenya's Eastern Province where Katumani maize is normally grown.

TABLE 4 Maize Yield Expectations Following Recommendations Based on

Date of Onset and Season Categorization tons/hectare

Season Category Judged

Actual Type from Early Rains

of Season A B C

(Short Rains)

A: Good 5.1 2.6 1.5

B: Fair 2.75 2.6 1.5

C: Poor 0 .25 .35

(Long Rains)

A: Good 4.2 2.6 1.5

B: Fair 2.75 2.6 1.5

C: Poor 0 .25 .35

%. .'

- 32 -

A preponderance of farmers in the area produce maize using very low

levels of inputs and sparse plant populations. These practices allow them

to produce a modest yield (approximately 0.85 t/ha) in all but the poorest

seasons, and expose them to only very low levels of financial risk. Consi-

dering the perceived uncertainties about rainfall, the weak financial posi-

tion of smallholders, and the consequences of maize crop failure to farm

families, such a "minimum cash input" policy is probably wise.

The existence now of a means to forecast seasonal rainfall utilizing

information on the onset and intensity of a given rainy season, provides

an opportunity for maize growers to take appropriate action as recommended

herein. It may, however, be preferable for some growers, given certain

information, to continue with the minimum cash input alternative, or, for

others perhaps to opt for even higher levels of inputs and potential re-

turns than these recommendations would produce.

8. Value of Forecast Information

Based on the historical record, good, fair, and poor seasons occur

with equal probability in the short rains, while for the long rains, half

the seasons have been good ones, with the remaining half being equally

divided between fair and poor seasons. Working through the twenty four years

of record, the probabilities for good, fair and poor seasons were revised

using Baye's Theorem (Raffia, 1970), for the short and long rains, given

precipitation 50 and 35 days respectively from date of onset. The results

of these revisions are presented in Tables 5 and 6.

Current prices were applied to the incremental inputs and outputs

resulting from adoption of the recommendations for the various season types

ir Table 7 to obtain the net value of incremental production. Costs were

estimated, based on those for Machakos District, published by the Ministry

of Agriculture (1980), but adjusted to 1981 levels.

TABLE 5 Short Rains:

Probabilities of Occurrence of Season Categories First Prior to Onset, and then in

Accordance with Early or Late Date of Onset. Finally, Probabilities that Seasons

Fitting the Rainfall Criteria Shown, will Actually End in the Category Indicated

Season Prior to At Early Onset Early Onset + 50 Days At Late Onset Late Onset + 50 Days

Category Onset Oct 20 Nov 3 R>238 mm R4237 mm Nov 4 23 R)206 mm R4205 mm

(13 Years) (6 Years) (7 Years) (11 Years) (4 Years) (7 Years)
A: Good .333 .462 .833 .143 .182 .5 0
B: Fair .333 .308 .167 .429 .364 .5 .286

C: Poor .333 .231 0 .429 .454 0 .714

TABLE 6 Long Rains: Probabilities (As in Table 5 above)

Season Prior to At Early Onset Early Onset + 35 Days At Late Onset Late Onset + 35 Days

Category Onset Feb 10 Mar 15 Rf3.2 3.1)R-2.2 R42.1 Mar 16 Apr 15 IR3.6 (mm/day) R3.5

(14 Years) (8 Yrs) (5 Years) (1 Yr) (10 Years) (5 Years) (5 Years)
A: Good .5 .786 1.0 .6 0 .1 .2 0

B: Fair .25 .143 0 .4 0 .4 .8 0

C: Poor .25 .071 0 0 1.0 .5 0 1.0

ds, Costs and Value of Production by Season Type and Alternative Action Machakos District

(Yields in metric tons per hectare; Values in Kenya Shillings)
Good Season Fair Season Poor Season
Action A Action B Action C Action A Action B Action C Action A Action B Action C

Short Rains Early Onset
Incremental Yield 4.25 1.75 .65 1.90 1.75 .65 0 .25 .35
Incremental Value 4486.30 1847.25 686.10 2005.55 1847.25 686.10 0 263.90 369.50
Preharvest Costs 1194.65 701.00 703.65 1194.65 701.00 703.65 1194.65 701.00 703.65
Harvest & Marketing 765.70 319.05 118.55 346.40 319.05 118.55 0 49.65 69.50
Net Incremental Value 2525.95 827.20 -136.10 464.50 827.20 -136.10. -1194.65 -486.75 -403.65

Short Rains Late Onset
Incremental Yield 4.25 1.75 .65 1.75 1.75 .65 0 .25 .35
Incremental Value 4486.30 1847.25 686.10 1847.25 1847.25 686.10 0 263.90 369.50
Preharvest Costs 1194.65 654.00 262.00 1194.65 654.00 262.00 1194.65 654.00 262.00
Harvest & Marketing 765.70 319.05 118.55 319.05 319.05 118.55 0 49.65 69.50
Net Incremental Value 2525.95 874.20 305.55 333.55 874.20 305.55 -1194.65 -439.75 38.00

Long Rains Early Onset
Incremental Yield 3.35 1.75 .65 1.90 1.75 .65 0 .25 .35
Incremental Value 3536.25 1847.25 686.10 2005.55 1847.25 686.10 0 263.90 369.50
Preharvest Costs 993.00 701.00 703.65 993.00 701.00 703.65 993.00 701.00 703.65
Harvest & Marketing 610.75 319.05 118.55 346.40 319.05 118.55 0 49.65 69.50
Net Incremental Value 1932.50 827.20 -136.10 666.15 827.20 -136.10 -993.00 -486.75 -403.65

Long Rains Late Onset
Incremental Yield 3.35 1.75 .65 1.75 1.75 .65 0 .25 .35
Incremental Value 3536.25 1847.25 686.10 1847.25 1847.25 686.10 0 263.90 369.50
Preharvest Costs 993.00 654.00 262.00 993.00 654.00 262.00 993.00 654.00 262.00
Harvest & Marketing 610.75 319.05 118.55 319.05 319.05 118.55 0 49.65 69.50
Net Incremental Value 1932.50 874.20 305.55 535.20 874.20 305.55 -993.00 -439.75 38.00

TABLE 7 Incremental Yiel

- 35 -

With reference to Table 7, incremental yields are valued at 1056.56

Kenya Shillings per metric ton (2204.62 lb). The approximate exchange rate

is 8.25 shillings/dollar. Preharvest costs assume transportation of inputs

20 km at 20 Kenya cents per 100 kg per km. Interest on preharvest incremen-

tal working capital is charged at 12% per annum for six months. Harvest and

marketing costs assume picking costs KSh 3.30/100 kg grain; shelling at KSh

0.75/100 kg grain; gunny bags at KSh 9.00 each; all grain transported 20 km

at KSh 0.20/100 kg per km; and interest on all costs at 12% per annum for

one month.

Weighting the per hectare incremental value of production net of incre-

mental costs preharvest and harvesting with the revised probabilities

for the various season types, allows one to place a value on incremental

maize production per hectare, given the forecast information which leads to

a certain action being taken. Estimates of these values are contained in

Table 8.

Table 8 presents the estimated impact of the information on per hectare

income from maize growing, under two assumptions. The first evaluation is

from the view point of a risk neutral decision maker who wishes to maximize

returns, given all the input-output possibilities presented here. The second

evaluation assumes the recommendations discussed earlier are followed.

Table 9 assesses the impact on incremental maize production per hec-

tare of following a risk neutral maximizing decision rule, and of following

the recommendations. These values are of interest to policy makers in that

Kenya has been forced to import maize to feed her people for the past two

years. If the practices shown in this paper were adopted in the semiarid

areas of Eastern Province, additional maize production (shown in Table 10)

well in excess of present levels of importsI/ is a clear possibility. As

1/ 400,000-450,000 tons estimated in 1980; a similar amount anticipated 1981.

- 36 -

TABLE 8 Comparison of per Hectare Returns (KSh) Under Pure Maximizing Risk

Neutral Decision Rule versus Recommended Actions


Risk Neutral Rule

Action Return

Short Rains-Early Onset
*Total R > 238 mm
*Total R < 237 mm

Season Average
Short Rains-Late Onset
*Total R / 206 mm
*Total R < 205 mm

Season Average

Short Rains Average

Long Rains-Early Onset
**Mean R > 3.2 mm/day
**3.1>/ R/ 2.2 mm/day
**Mean R < 2.1 mm/day

Season Average

Long Rains-Late Onset
**Mean R > 3.6 mm/day
**Mean R < 3.5 mm/day

Season Average

Long Rains Average

Annual Average






- 403.65







Action Return






- 403.65






of Season












* Measured at date of onset

** Computed at approximately

+ 49 days.

35 days after date of onset.

- 37 -

TABLE 9 Comparison of Incremental Maize Production Under Pure Maximizing

Risk Neutral Decision Rule versus Recommended Actions (t/ha)

Risk Neutral Rule Recommended

Average Average
Season Incremental Incremental
Action Yield Action Yield
Short Rains Early Onset
*Total R > 238 mm A 3.858 A 3.858
*Total R < 237 mm B 1.107 B 1.107

Season Average 2.377 2.377

Short Rains Late Onset
*Total R >/ 206 mm A 3.000 B 1.750
*Total R < 205 mm C 0.436 C 0.436

Season Average 1.368 0.914

Short Rains Average 1.915 1.706

Long Rains Early Onset
**Mean R > 3.2 mm/day A 3.350 A 3.350
**3.1 ,R,2.2 mm/day A 2.710 B 1.750
**Mean R\< 2.1 mm/day C 0.350 C 0.350

Season Average 2.907 2.564

Long Rains Late Onset
**Mean R > 3.6 mm/day B 1.750 B 1.750
**Mean R 3.5 mm/day C 0.350 C 0.350

Season Average 0.910 0.910

Long Rains Average 2.075 1.875

Annual Average 3.990 3.581

Measured at date of onset + 49 days.
** Computed at approximately 35 days after date of onset.

- 38 -

levels of production in the area could be estimated two to three months in

advance of harvest, public decision makers could have considerable time to

handle the shortages and gluts that, instead of chronic shortage, would

characterize the area.

TABLE 10 Possible Increased Maize Production from Semiarid Areas of

Eastern Province

Additional Maize Production, tons

Season Hectares Normally Risk Neutral Recommended

Planted to Maize Maximizing Rule Action

Short Rains 210,000 402,150 358,260

Long Rains 180,000 373,500 337,500

Average Annual Increase 775,650 695,760

The values per hectare from Table 8 under the risk neutral maximizing

decision rule can be compared to the values generated using the probabil-

ties of occurrence of the several types of seasons before knowledge about

the onset of the rains is considered. The difference between these two

values is the value per season per hectare of the meteorological forecast

information. Table 11 presents the latter values.

TABLE 11 Estimated Value per Hectare of Meteorological Forecast

Season Value, KShs

Short Rains 325.03

Long Rains 597.44

Annual Value 922.47 KSh/ha

- 39 -

The above valuations are based on the assumption that the decision

maker has a neutral attitude toward risk taking. It is a popular belief

among workers in development that African peasant farmers are averse to

taking risk. If the smallholder maize growers in Eastern Province are also

risk averse, the value to them of reliable rainfall prediction is greater

than to risk neutral decision makers. Conklin, et al (1977), develop this

point more fully.

Although the payoff to adoption of the recommendations, or the risk

neutral maximizing rule, is substantial, it is doubtful that maize growers

in the project area will adopt them without some protection from the risk

of financial losses which result when poor seasons occur. An insurance

scheme for adopters is a real possibility. Investigations of appropriate

insurance premiums, and the expansion of the scheme to consider various

intercropping alternatives would appear to be very much in order.

9. Conclusions

Despite our inability to predict the notoriously variable rains in

Kenya's semiarid areas, a newly developed analysis of "effective rainfall"

for maize production does permit their after the fact interpretation in

time to give practical advice on farm practices and levels of inputs. The

analysis of the available weather records primarily rainfall and evapor-

ation when coupled with suitable research findings concerning the crops

and soils of interest, accomplishes the following:

a. Permits evaluation of the suitability of a given crop for production

at the planning site.

b. Defines the earliest and latest acceptable dates of onset of the

rains for growing the study crop.

c. Quantifies the initial rainfall which should be accepted by the far-

mer as the signal to plant his crop.

- 40 -

d. Reveals that date of onset of the rains is correlated with total sea-

son rainfall expectation, and with intensities of early season rains. Hence,

pinpoints ranges of dates properly termed early, late, and too late as

regards planting, and quantifies early season rainfall amounts which indi-

cate either a good, fair, or poor season is in store.

Application of the above information occurs at three stages:

a. The date of onset of the rains triggers recommendations to farmers on

date of planting, seeding rates, and initial fertilizer applications.

b. Rainfall totals 50 days after date of onset in the short rains, and

30-40 days into the long rains permit categorization of season type, and

determine farmer recommendations on thinning to desired stands, and on

adjustment of nitrogen fertilizer levels through sidedressing.

c. At 75 days into the season, or approximately two months before harvest,

total season effective rainfall can be estimated, and predictions of maize

yield provided to farmers, economists, and other planners concerned with

food supplies.

The estimated average per hectare increase in maize production on an

annual basis (two seasons), expected from following the recommendations

presented, is 3.58 t/ha, for which the average incremental net return is

currently estimated at 1,790 Kenya Shillings, or approximately $217/ha.

Considering the area normally planted to maize in Eastern Province, the

possible annual incremental maize production is 696,000 metric tons. This

is more than sufficient to cover the present need Kenya has to import maize

to feed her people.

l; lA.

41 -

10 Acknowledgements

The study was carried out within the Dryland Cropping Systems Research

Project, jointly funded by the United States Agency for International Devel-

opment and the Government of Kenya, and headquartered in the Kenya Agricul-

ture Research Institute at Muguga. Experiments were conducted with Katumani

maize at Muguga, and at the Katumani National Dryland Farming Research Sta-

tion, where the Director, Mr. A. M. Marimi and his staff have kindly afforded

us research facilities, and the meteorological data utilized. Officials of

the Ministry of Agriculture have likewise been most gracious and helpful in

providing needed data on maize acreages and prices, and prices of inputs

and services for producing, harvesting and marketing the crop. We are very

grateful to all of the above.

The principal author further wishes to render thanks for the helpful

addition of research findings of former colleagues R. J. Hanks and J. P.

Riley at Utah State University, and R. E. Danielson and W. T. Franklin at

Colorado State University, and great appreciation to R. M. Hagan and W. 0.

Pruitt at the University of California, Davis, with whom experiments cited

were carried out.

- 42 -

Figure Legends

Figure 1 -

Figure 2 -

Figure 3 -

Figure 4 -

Figure 5 -

Figure 6 -

Maximum Yields of Climatically Adapted Maize Cultivars Grown

in Irrigated Research Plots Using Best Known Management

Techniques Related to Energy Availability for Growth,

Represented by Seasonal Average Daily Rate of Evaporation.

Katumani Short Rains: Characteristics over 24-Year Period

from 1956 1979.

Katumani Short Rains: Three Distinct Patterns of Effective

Rainfall for Maize Production, Representing Good, Fair, and

Poor Seasons.

Katumani Long Rains: Characteristics over 24-Year Period

from 1957 1980.

Katumani Long Rains, Early Onset (February 10 March 15):

Three Distinct Patterns of Effective Rainfall for Maize

Production, Representing Good, Fair, and Poor Seasons.

Katumani Long Rains, Late Onset (March 16 April 15): Two

Distinct Patterns of Effective Rainfall for Maize Production,

Representing Fair and Poor Seasons.

- 42a -

12- /
* Experimentol Sites 8 Yeaors
I Kotumoni, Kenya 1977
2 0 2- 1978
E 3- 1978
4 4 1979
S8 5 Mugugo, Kenya 1978 0l"
6 Mugugo, Kenya 1979
S7 .- eC r,""1 C!. 1974
6 8-Logon., Utoh 1975 >@
9 Davis, Calif. 1974 Y M -.740 + .444Ep
c 10-Dovis, Calif. 197 r2 .89

4 7
- .7
S2 -

_- /-I I I I I I I I
I 2 3 4 5 6 7 8 9

Figure 1 Maximum yields of climatically adapted maize cultivars

grown in irrigated research plots using best known

management techniques related to energy availability

for growth, represented by seasonal average daily rate

of evaporation.


S 500

2 (456)


cO 1322)





















A "A"
0 "B"
-A "C"






II 21


30 50 80 100 120

Figure 2 Katumani short rains: Characteristics

over 24-year period from 1956 1979.

Figure 3 Katumani short rains: Three distinct patterns

of effective rainfall for maize production,

representing good, fair, and poor seasons.

~I-IP)iCI -r_

'I 'I *

42c -



U; (415) OAo
400 > 'r 4

o o\. <(332)
3 0000 -'q

o# /0V -

0 (216)
o 0o,
u (158)


W 21 I II 21 I II 21

Figure 4 Katumani long rains: Characteristics over 24-year

period from 1957 1980.







(394, ETm)















0 30 50 80 100 120

Fig 5 Katumani long rains, early onset (Feb 10 Mar 15):

Three distinct patterns of effective rainfall for

maize production, representing good, fair, & poor




(330, ET,)



20mm Soil Evoporotion
Los Prior to Panting


30 50 80 100 120

Fig 6 Katumani long rains, late onset (Mar 16 Apr 15):

Two distinct patterns of effective rainfall for

maize production, representing fair & poor seasons.

- 43 -


Allan, A. Y., and D. Laycock. 1979. Maize yields in Western Kenya: Effects

of plant numbers per hill and stage at thinning. E. Afr. agric. For. J.


Blackie, J. R. 1969. Lysimeter study of the water use of sugar at Arusha

Chini, Tanzania. Record of Research Annual Report, 1968. East

African Agriculture & Forestry Research Organization, Muguga, Kenya.

Braun, H. M. H. 1977. The reliability of the rainy seasons in Machakos

and Kitui Districts. Miscellaneous Paper M12. Kenya Soil Survey,

National Agricultural Laboratories, Ministry of Agriculture, Nairobi.

Cocheme, J. and G. Zazzara. 1969. A contribution to the forecasting of

the early.rains at Nairobi. E. Afr. agric. For. J. 34(3):312.

Conklin, F. S., A. E. Baquet, and A. N. Halter. 1977. A Bayesian simula-

tion approach for estimating value of information: An application to

frost forecasting. Technical Bulletin. Agricultural Experiment Station,

Oregon State University, Corvallis. 64p.

Dagg, M. 1965. A rational approach to the selection of crops for areas of

marginal rainfall in East Africa. E. Afr. agric. For.J. 30:296-300.

Dagg, M. 1969. Factors affecting the use of evaporation pan data in East

Africa. E. Afr. agric. For. J.35:203-210.

Dagg, M. 1970. A study of the water use of tea in East Africa using a

hydraulic lysimeter. Agr. Meteorol. 7:303-320.

Dagg, M., T. Woodhead, and D. A. Rijks. 1970. Evaporation in East Africa.

Bull. I. A. S. H. 15:61-67.

Doorenbos, J. and A. H. Kassam. 1979. Yield response to water. FAO Irri-

gation & Drainage Paper 33. FAO, United Nations, Rome. 193p.

Dowker, B. D. 1963. Rainfall reliability and maize yields in Machakos

District. E. Afr. agric. For. J. 28:134-138.

- 44 -

East African Meteorological Department. 1961. 10% and 20% Probability

maps of annual rainfall of East Africa. E. A. M. D., Nairobi.

Forsgate, J. A., P. H. Hosegood, and J. S. G. McCulloch. 1965. Design and

installation of semi-enclosed hydraulic lysimeters. Agr. Meteorol.


Glover, J. and J. A. Forsgate. 1964. Transpiration from short grass.

Quart. J. R. Met. Soc. 90:320-324.

Glover, J. and P. Robinson. 1953a. A simple method for assessing the

reliability of rainfall. J. Agric. Sci. 43:275.

Glover, J. and P. Robinson. 1953b. A simple method of calculating the

reliability of rainfall. E. Afr. agric. For. J. 19:11-13.

Gichuiya, S. N. 1968. The onset of the short rains. E. A. Meteorological

Department, Nairobi. Presented at Fourth Specialist Committee on

Applied Meteorology. 7p.

Manning, H. L. 1950. Confidence limits of expected monthly rainfall.

J. Agric. Sci. 40:169-176.

Manning, H. L. 1956. The statistical assessment of rainfall probability

and its application in Uganda agriculture. Proc. Roy. Soc., B.


Manning, H. L. 1960. Cotton crop potential and rainfall expectation in

Uganda. Tropical Meteorology in Africa. Ed. Bargman, D. J.

Nairobi: Munitalp Foundation.

McCulloch, J. S. G. 1965. Tables for the rapid computation of the Penman

estimate of evaporation. E. Afr. agric. For. J. 30:286-295.

Nkedi-Kizza, P. 1973. A study of the consumptive use of water by the

nakyeteigo banana (Musa acuminata). University of Makere, Kampala,

Uganda. Thesis submitted for MSc degree. 217p.


45 -

Okalabo, J. R. 1977. Maize response to three high analysis phosphate

fertilizers in some soils of East Africa. E. Afr. agric. For. J.


Penman, H. L. 1948. Natural evaporation from open water, bare soil and

grass. Proc. R. Soc. (London). A.193:120-145.

Raiffa, Howard. 1970. Decision Analysis. Addison Wesley, Reading, Mass.

Republic of Kenya. 1980. Yields, costs, prices 1980. Min. of Agric.,

Land & Farm Management Division, Nairobi.

Semb, G. and P. K. Garberg. 1969. Some effects of planting date and

nitrogen fertilizer in maize. E. Afr. agric. For. J. 34:371-379.

Semb, G. and J. B. D. Robinson. 1969. The natural nitrogen flush in

different arable soils and climates in East Africa. E. Afr. agric.

For. J. 34:350-370.

Stewart, J. Ian and Robert M. Hagan. 1969. Predicting effects of water

shortage on crop yield. J. of the Irrigation and Drainage Div., ASCE.

95(IR2):91-104. Proc. Paper 6443.

Stewart, J. Ian. 1972. Prediction of water production functions and

associated irrigation programs to minimize crop yield and profit

losses due to limited water. University of California, Davis.

PhD Dissertation. 200p.

Stewart, J. Ian and Robert M. Hagan. 1973. Functions to predict effects

of crop water deficits. J. of the Irrigation and Drainage Div., ASCE.

99(IR4):421-439. Proc. Paper 10229.

Stewart, J. Ian, Robert M. Hagan and William 0. Pruitt. 1974. Functions

to predict optimal irrigation programs. J. of the Irrigation and

Drainage Div., ASCE. 100(IE2):179-199. Proc. Paper 10626.

Stewart, J. Ian,Ram Deo Misra, William 0. Pruitt and Robert M. Hagan. 1975.

Irrigating corn and grain sorghum with a deficient water supply.

Transactions of the ASAE. 18(2):270-280.

- 46 -

Stewart, J. Ian, Robert M. Hagan and William 0. Pruitt. 1976. Water

production functions and predicted irrigation programs for principal

crops as required for water resources planning and increased water use

efficiency. Completion Report 14-06-0-7329. USDI/USBR Engineering and

Research Center, Denver, Colorado. 90p.

Stewart, J. Ian. 1977. Conservation irrigation of field crops: A drought

year strategy. California Agriculture. 31(4):6-9.

Stewart, J. I., R. J. Hanks, R. E. Danielson, E. B. Jackson, W. 0. Pruitt,

W. T. Franklin, J. P. Riley and R. M. Hagan. 1977. Optimizing crop

production through control of water and salinity levels in the soil.

Technical Completion Report to USDI/OWRT. Utah Water Research Lab.

Report PRWG151-1. Logan, Utah. 206p.

Stewart, J. Ian and Fred J. Wang'ati. 1978 Research on crop water use and

drought responses in East Africa. In Proc. Intnl. Workshop on the

Agroclimatological Research Needs of the Semi-Arid Tropics, 1980.

ICRISAT, Hyderabad, India. 230p.

Stewart, J. Ian and Joseph 0. Mugah. 1979. Lysimeter water balance for

bare and cropped soil. Proc. of the Third Annual General Meeting of

the Soil Sci. Soc. of East Africa. Muguga, Kenya. 20p.

Walker, J. T. and D. A. Rijks. 1967. A computer programme for the calcu-

lation of confidence limits of expected rainfall. Expl. Agric. 3:337-341.

Wang'ati, F. J. 1972. Lysimeter study of water use of maize and beans in

East Africa. E. Afr. agric. For. J. 38(2):141-156.

Woodhead, T. 1968. Studies of potential evaporation in Kenya. Ministry of

Natural Resources, Water Development Department, Nairobi. 69p + maps.

Woodhead, T., E. S. Waweru and E. F. Lawes. 1970. Expected rainfall and

Kenya agriculture confidence limits for large areas at minimum cost.

Expl. Agric. 6:87-97.

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs