Effect of Urbanisation On Nbi Climate

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/265612737
Effects of Urbanization on Climate of Nairobi City
Article · December 2010
CITATIONS READS
4 3,203
3 authors, including:
Victor Ongoma Nzioka John Muthama

Mohammed VI Polytechnic University University of Nairobi
127 PUBLICATIONS 2,265 CITATIONS 141 PUBLICATIONS 792 CITATIONS
SEE PROFILE SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Journal of Environmental and Agricultural Sciences editor.jeas@outlook.com View project
Agroecology View project
All content following this page was uploaded by Victor Ongoma on 28 September 2014.
The user has requested enhancement of the downloaded file.

ISSN 1995-9834
Journal of Meteorology
and Related Sciences
A Journal of Kenya Meteorological Society
December 2010 - Volume 4

Journal of Meteorological and Related Sciences
Chief Editor Technical Editor
Dr Raphael Okoola Misheck M. Kainga
University of Nairobi Kenya Meteorological Society
Associate Editors
Dr. C. Mutai R. Ojala P. Ambenje,

Kenya Meteorological Department Kenya Meteorological Department Kenya Meteorological Department
W. Nyakwada Dr. S. Marigi Dr. G. Ouma
Kenya Meteorological Department Kenya Meteorological Department University of Nairobi
Prof. F. Mutua Prof. J. Ng’ang’a Dr. F. Opijah
University of Nairobi University of Nairobi University of Nairobi
Dr. J. Ininda S. Mwangi G. Kongoti
University of Nairobi Kenya Meteorological Department Kenya Meteorological Department
Dr. J. N. Mutemi Prof. N. J. Muthama S. King’uyu
University of Nairobi University of Nairobi Kenya Meteorological Department
Dr. A. Opere Dr. C. Oludhe, Prof. S. Oteng’i
University of Nairobi University of Nairobi Masinde Muliro University
V.N. Sakwa Dr. P. Camberlin
Kenya Meteorological Department University of Bourgogne
Prof. G. Karani Prof. L. Ogallo
School of Applied Sciences IGAD Climate Prediction & Applications Centre
University of Wales, Institute of Cardiff
The Journal of the Kenya Meteorological Society will accept original research papers for publication
following the recommendations of a review panel. All areas of meteorology, and related Sciences will be
considered.
Copyright © Kenya Meteorological Society (KMS), 2010. Permission to use figures, tables, and brief
excerpts from this journal in scientific and educational works is hereby granted provided that the source is
acknowledged. Authorization to photocopy items bearing individual KMS copyright notices at the bottom
of their first page is granted by KMS. Republication, systematic reproduction, and other uses of any mate-
rial in this journal require written permission from the KMS.
The JOURNAL OF KENYA METEOROLOGICAL SOCIETY (ISSN 1995-9834) is published by the
Kenya Meteorological Society at Dagoretti Corner, Ngong Road, P.O. Box 41959 00100 GPO Nairobi, Kenya.
Email: kmsmails@yahoo.com, Website: www.kenyametsociety.org, Tel: +254 20 2147770, or +254 20
3867880 ext 2230 .The price of single issues is Kenya Shillings 750. Subscriptions outside Kenya carry an
additional postage charge determined at the time of ordering.
Information for Authors

1. Introduction promptly inform the chief editor if it is submitted
This document provides essential infor- for publication elsewhere before its disposition
mation required by authors to submit manu- by the journal. This information can be put in the
scripts to Kenya Meteorological Society (KMS) cover letter that accompanies the submitted
journal. It is intended to serve as a quick refer- manuscript.
ence for frequently needed information. Each manuscript must be accompanied
Authors are encouraged to use this docu- by a statement transferring copyright from the
ment for more complete and more up-to-date in- authors (or other holder of the copyright) to the
formation on KMS style and the publication proc- Kenya Meteorological Society (KMS).
ess. The appropriate form for the transfer of
2. Author disclosure and transfer of copyright the copyright to KMS can be obtained from the
When a manuscript is submitted, the au- Editor's office or from the Society. The written,
thor must inform the chief editor of that publica- signed transfer of the copyright is required in
tion if it has been previously published in any lan- order for KMS to have valid rights to continue its
guage or if it is under consideration for publica- wide dissemination of research results and other
tion by another journal. The author must also scientific information. …….Continued on page 55
DECEMBER 2010 J.Meteorol. Rel. Sci., 4 3 –20 (2010) 3
Spatial and Temporal Rainfall characteristics Over Seychelles
V.J. Amelie
Seychelles Meteorological Services.
J.M. Ininda and R.E Okoola
Department of Meteorology, University of Nairobi, Kenya
CORRESPONDING AUTHOR
J. M. Ininda Department of Meteorology,
University of Nairobi
P. O. Box 30197
00100 NAIROBI
KENYA
Tel: 254 20 3878340
Fax: 254 20 3878343
Email : jininda@uonbi.ac.ke
(Manuscript received 16 September 2009, in final form 26 June 2010)
ABSTRACT
While Seychelles lies close to the Equator, it experiences only one
rainy season. The rainfall characteristics over this region is evident from the
Satellite observations which show that during most part of the year the active
clouds are concentrated to the eastern sector of the equatorial Indian Ocean
and Seychelles comes under active weather only during southern hemisphere
summer. The main objective of this study was to therefore investigate the spa-
tial and temporal characteristics of rainfall over Seychelles during the rainy
season. The specific objectives included demarcating Seychelles into rainfall
homogeneous zones, identify the rainfall season, determine the onset and ces-
sation of the rainy season. The data used in the study were the daily and
monthly rainfall over Seychelles for the period 1972 to 2006 and NCEP re-
analysis data. The methods used to analyze these data were Principal Compo-
nent Analysis (PCA), time series analysis, pentad and cumulative curves.
The results from PCA analyses showed that Seychelles may be divided
into four rainfall homogeneous zones. The study also shown that the rainfall
season occur between November and March. January was observed to be the
peak rainfall month, while July has the lowest amount of rainfall. The onset
and cessation of the rainy season occur in the month of November and March
respectively. Analysis of the low level flow showed the Near Equatorial
Trough (NET) is the main system influencing rainfall over the Seychelles re-
gion. During the time of maximum rainfall in January, the Inter-Tropical Con-
vergent Zone (ITCZ) lies far to the south of the country
1. Introduction (Figure 1). It comprises of 115 islands with a

Seychelles is a small country in the In- population of about 85,000 most of who live
dian Ocean, located approximately 04°S and on the main island, Mahe, which covers
55°E roughly 150 square km and is 27 km long and
seven km wide.
4 Journal of Meteorology and Related Sciences Volume 4
Intro.
Figure 1(a): Position of Seychelles in the Indian Ocean. The red arrow depicts the approximate loca-
tion of Seychelles (4°41’S, 55°E)
Figure 1(b): An enlarge map of Seychelles showing the main Island (Mahe) and the nearby Islands.
Just like most developing countries, the planning. While most of the regions located
socio-economy of Seychelles is largely depend- near the equator have two rainfall peaks, Sey-
ent on rainfall. For instant, rainfall is the main chelles has only one rainfall season which nor-
source of fresh water for the for domestic and mally starts around November and ends in
industrial use in the country. Therefore in order March. Moreover the rainfall over this country
for the country to have sustainable develop- is observed to have large spatial variability
ment, the nature of rainfall variability has to be that has been attributed to the complex topog-
understood and factored into the economic raphy over the Island.
DECEMBER 2010 AMELIE ET AL 5
The main objective of this study was to circulation is the major factor that keeps most
therefore investigate the spatial and temporal active weather activities to the far east of Sey-
characteristics of rainfall over Seychelles dur- chelles, or if there are other weather elements
ing the rainy season. The study included de- which contribute to make the weather to be-
marcating Seychelles into rainfall homogene- have the way it does over the country. A num-
ous zones, identify the rainfall season, deter- ber of studies have been conducted in the past
mine the onset and cessation of the rainy sea- on the tropical weather in this particular re-
son. gion, specifically on the influence of the I.T.C.Z
and its interrelationship with the rainfall pat-
1.1 Literature Review
tern (Okoola, 1998, 1999, Hasternrath, 2007).
Many of the previous studies show that the
Over Seychelles extreme rainfall events
rainfall characteristics over Seychelles is influ-
have been observed when the rainfall season
enced by various systems which include the
coincide with high frequency of Tropical cy-
Near Equatorial Trough (NET), the Southeast
clones occurrence. At times the track taken by
monsoon, tropical cyclones, easterly waves,
the cyclone brings a lot of active weather asso-
the subtropical anticyclones and mesoscale
ciated with its feeder clouds.
circulations. Sadler et. al. (1987) have shown
This observation of limited weather develop-
that the northeast monsoon at the gradient
ment over Seychelles can be argued in relation
level recurve and become westerlies near their
to the Walker Circulation, where one of its as-
southern limit. This recurving creates a trough
cending arm lies around 70-100° East associ-
(the Near Equatorial Trough) that lies over
ated with active weather during the most part
Seychelles during the period December, Janu-
of the year, and a descending arm around the
ary and February (DJF). example of this is
region of Seychelles where relatively dry
given in Figure 2(a).
weather is more dominant. It is necessary to
Figure 2(a): Mean surface level streamline analysis over the Indian Ocean for January (Sadler et
al., 1987). Seychelles enclosed by the red circle lies under influence of Near Equatorial Trough
(the red line representing axis of NET over Seychelles). The blue line is representing the I.T.C.Z
On the other hand during June, July and Au- troughs in the vicinity of the Equator, the
gust (JJA) season the trough is located further I.T.C.Z and the NET. These two troughs have
north of Seychelles with greatly reduced rain- been highlighted in Figure 3 above. Net nor-
fall at this time (Figure 2b) mally occurs in the winter hemisphere and is
Asnani (2005) noted that there exist closer to the Equator than the I.T.C.Z.
Figure 2b: Mean surface level streamline analysis over the Indian Ocean for July (Sadler et al.,
1987). Red line shows axis of the Trough (NET) to the North of Seychelles.
Okoola (1998) using Outgoing Long- 2.1 Data source and type
wave Radiation (OLR) from satellite observa- The main data set used in this work
tions has shown that throughout the year the was the daily and monthly rainfall from 1972
most active weather is concentrated far to the to 2006 for the stations of Seychelles. The
east of Seychelles beyond longitude 70° East. wind data was the reanalysis data available
Some of the finding in his study are shown in from the National Centers for Environmental
Figures 3(a) and 3(b) which indicate the mean Prediction-National Center for Atmospheric
stages of transition of low OLR values across Research (NCEP-NCAR).
the region. Sadler et al., (1987) have also
2.2 Methodology
shown that there is a well organized equato-
The methods used include statistical,
rial trough which may be associated with the
graphical and analysis of wind pattern. The
occasional active systems from the Far East
statistical methods included, the time series
that move over the country in a form of east-
analysis and principal component analysis.
erly waves and influence the weather.
2.2.1 Time Series Analysis
2.0 Data and Methodology
In this study, time series analysis was
This section is devoted to the descrip-
carried out to investigate the seasonal rainfall
tion of the data and the methods that were
variations. The components of the time series
used in this study.
Figure 3: Example of mean stages of transition of low OLR values across Equatorial regions during
period (a) 31 Jan – 4 Feb. and (b) 31 May – 4 Jun. The shaded parts indicate regions of low OLR associ-
ated with high cloud concentration. Modified from Okoola, (1998). Mahe depicted in the red circle.
that were examined included the trend and Fourier analysis method which was used to
cyclic variations describe periodic characteris- detect cyclical variations of rainfall in this
tics. A detailed description of these compo- study. Spectral analysis has been used by
nents is provided by Kendall et al., (1983). many authors to examine cyclic variations.
The normalized power spectrum ¦(w) takes
2.2.1.1 Trend analysis ∞
1
Many methods have been used to de- ∑ e − iωr ρ ( r )
scribe trend in climatological data. These inform¦ (w) = 2π r =−∞
clude: graphical, polynomial, and statistical

methods. In this study the trend analysis is (1)
carried out using graphical method. -p£w£p
The graphical method involves plotting the
rainfall data against time. This method pro- Where ω = 2π f is the angular frequency and
vides a quick visual observation of the pres-
ence of trend in a given time series. ¦ is the frequency and i = −1 .
2.2.1.2 Spectral Analysis In order to obtain consistent estimates of ¦

Spectral analysis is a technique for ex- (w), smoothing functions λ (r) are used. The
smoothed spectral density function, f (ω ) ,
amining the hidden periodicities (cycles or
oscillations) of any time series at certain fre-
′
may be expressed as f (ω ) = λ ( r ) f (ω ) .
quencies. Other methods include the fast
Where λ (r ) is the smoothing weights or lag Kenya and East Africa by Ogallo (1989),
windows. Examples of lag windows that are Oludhe, (1987), Basalirwa (1991), Ininda
used to smoothen the power spectrum are: (1995), and Okoola (1996) among many oth-
Truncated Periodogram, Bartlett, Daniel, Tukey ers. Detailed discussions on this method are
Humming, Tukey Hunning, Parzen and Bark- presented by Cattell, (1966) and Herman,
lett–Priestley windows. The most commonly (1967).
used windows are the Parzen and Tukey win- Two methods were used to deter-
dows (Jenkins and watts, 1968). The window mine the significant principal components,
employed in this study is the parsen window. namely the Kaiser’s Criterion and the Scree
This type of window is chosen because it is method.. The Kaiser’s Criterion assumes that
non-negative over the whole range of fre- all principal components with eigenvalues
quency and therefore avoids leakage of the greater or equal to one are significant (Nie et
power spectrum. al., 1970)
The dominant periodicity is identified by the
In the Scree Method logarithm of the
peaks in the spectral density function. The sig-
eigenvalues is plotted against the eigen
nificant peaks at 95% confidence level were
number, at the point where the line becomes
f CL ,95% almost horizontal marks the cut off of the
those above the threshold spectrum,
given by significant components. The details of the
f ′(ω ) χν2,95% methods can be found in Okoola (1996).
f CL ,95% = PCA assumes that the records are orthogonal
ν (independent). Since meteorological pa-
(2) rameters like temperature are influenced by
Where f (ω ) is the mean of the smoothed pe-
′ synoptic or regional factors, some similari-
ties should be expected between tempera-
χν2,95% tures of the neighbouring stations. Such
riodogram for the spectral estimates, is
similarities are not taken into account in PCA
the 95% point of the χ -distribution with ν
2
solutions, making some of the derived pat-
degree of freedom. The details of determining terns physically unrealistic (Richman, 1981).
the significant spectral peaks are described by The problem is sometimes reduced through
Minja (1984). the rotation of the principal components
(Child, 1990; Okoola, 1996) and other verifi-
2.2.2 The Principal Component Analysis cation techniques.
The Principal Component Analysis Under the rotations, the frames of
(PCA) helps to reduce data by grouping the reference of the principal components are
variables having high relationship together. In turned about the origin until some alterna-
this study, PCA analysis was utilized to find the tive position has been reached. Such changes
inter relationship between rainfall stations have been noted to reduce some ambiguities
over Seychelles and group together the highly associated with the unrotated component
correlated in order to create homogeneous solutions (Richman, 1986; Okoola, 1996 and
zones. many others). The rotation does not affect
In the PCA analysis, the orthogonal functions the total variance explained by the eigenval-
are defined as exact mathematical linear trans- ues. The two common methods of are the
formation of the original data is given by `orthogonal` and the `oblique` rotations. In
m
the orthogonal rotation the reference axes
Z i = ∑ ai Ci
i =1
are maintained at 90º; while in oblique case
(3) the components are partially correlated. The
types of the orthogonal rotations include the
Where Z i is variable i in the standardized form, varimax, quantimax, and equimax methods.
Ci represents the principal component I, ai is Kaiser’s varimax method (Kaiser, 1958) is
the standardized multiple regression coeffi- more popular and is used in this study.
cient of the variable i on component i. It has been the used by many authors in-
This method has widely been used in determin- cluding Ogallo ( 1980; 1988;), Barring,
ing regional homogeneous rainfall zones over (1987), Oludhe, (1987) and Okoola, (1996).
2.2.3 Determination of onset and cessation and the loadings are positive. Factor 1 represents
of seasonal rains based on pentad the dominant mechanism. For factor 2 and 3,
and cumulative curves some of the loadings are positive while others are
Five-day (pentad) rainfall totals negative. These factors therefore represent the
was used in this study to determine the on- mechanism that would have opposite effect at
set/withdrawal of the rainy seasons. This some locations. These three Eigenvectors ac-
unit of time (pentad) has routinely been counted for 79 percent of the variance.
used in India to determine the dates of on- The three significant were rotated. The
set/withdrawal of monsoon rains there “clusters” or homogeneous regions were deline-
(Asnani, 2005). Also Alusa and Mushi ated based on loadings with magnitude equal or
(1974) have used pentad rainfall to decide greater than 0.4 on a given
on the normal dates of onset/withdrawal of
seasonal rains over East Africa. Asnani rotated eigenvector. Four regions were obtained
(2005) has pointed out that the five-day as shown in figure 5a. Regions 1 and 2 cover the
period has come to be regarded as a useful larger proportion of the country. Region 1is lo-
unit of time in dealing with meteorological cated to the western side of the mountain ranges
phenomena in the tropics. Cumulative pen- while region 2 is on the eastern side. The rainfall
tad rainfall was plotted against time and over the two regions is modified by the topogra-
the pentad of onset/withdrawal is taken as phy depending on the direction of the wind flow.
the time when there is a sudden increase/ Region 3 is a low coastal land being influenced by
decrease in rainfall values (Ogallo et al. land and sea breeze. Region 4 is a high ground
1994). located to the northern tip of the country.
Inter-station correlation is shown in Table
2.2.4 Method of wind analysis 5. From this Table, it can be seen that the stations
Wind vectors were plotted to deter- within the same rainfall homogenous are highly
mine the low level flow pattern. The main correlated.
aim was to locate the mean position of the
Near Equatorial Trough and the Inter- 3.2 Temporal variation of rainfall over
Tropical Convergent Zone the systems that Seychelles
are associated with rainfall. Analysis of the mean monthly rainfall
( shown in figure 6) indicate that there is only one
3.0 Results and Discussion rainfall season over Seychelles. It also shows that
The results obtained from the study Seychelles receive significant rainfall throughout
are presented in various subsections below. the year, the lowest mean monthly being ob-
served around June / July.
3.1 Results from Principal Component In order to determine the main rainfall
Analysis season, a ‘wet’ month, the rainfall climatology was
This section presents the results considered. Similar approach has been used by
from PCA that were used to group the spa- Okoola and Ambenje (2003) in their study for the
tial patterns of the seasonal rains Congo Basin which is one of the global region
(November to March) over Mahe into ho- with abundant rainfall throughout the year just
mogeneous spatial zones. Figure 4 gives a like Seychelles. They adopted a threshold of
typical example of the results which were 100mm. This threshold has also been suggested
derived from principal component analysis. by Jackson and Weinand, (1994).
The characteristics are displayed in the However, many stations over Seychelles
form of a scree plot for the seasonal records experience more rainfall than the Congo Basin,
during the period 1972 to 2006. The factors therefore the threshold value to define a wet
and their eigenvalues are shown in Table1. month over the country should be higher than the
The examination of the scree plot (figure 4) 100mm. Based on the analysis of the mean
indicate that the first three Eigenvectors monthly rainfall ( Figure 6) a threshold of 200mm
were significant. Table 2 shows the Factor of rainfall was defined for a wet month over Sey-
loading for each station from un-rotated chelles. The main rainfall season over Seychelles
PCA. It can be seen from the table that Fac- is November to March.
tor 1 has the highest loading at all stations
Scree Plot
12
10
8
Eigenvalue
0
0 5 10 15 20
Number of Factors
Figure 4: Showing the plot of Eigenvalues v/s Factor number
Figure 5(a): Mahe divided into four homogeneous regions. Regions 1, 2, 3 and 4 are represented by
La Gogue, Le Niol, Belombre and Anse Forbans respectively.
3.3 Onset and cessation of seasonal rains

based on cumulative curves
Examples of cumulative pentad rain-
fall are shown in Figures 7, 8 and 9. Regions 1
and 3 indicate onset (cessation) pentad as 59
(17), which fall in October (May), while re-
gions 2 and 4 indicate the onset (cessation)
pentad as 65 (13), which fall in November
(March). Thus the regions to the west receive
rainfall earlier than those to the east. These
results suggest that the rainfall generating
system over Seychelles advance from the
west to east and recede westward so that the
zones located to the west have a longer rain-
fall season. However, as indicated from the
analysis of the monthly rainfall, the main
rainfall season over most of Seychelles is
from November to March.
3.4 Results from trend and spectral

analysis
Examples of graphical plots from
trend analysis are shown in Figure 10. The
seasonal rainfall from all the four region indi-
cated increasing trends suggesting enhanced
rainfall in the recent past.
Spectral analysis results indicated three ma-
jor quasi-periodic fluctuations in the sea-
sonal rainfall data. The peaks were centred
on 2.5–3 years, 3.5–5 years, and 7–10 years.
Figure 11 shows some of the results from the
spectral analysis of seasonal rainfall. Previ-
ous studies obtained similar results and have
associated the physical significance of these
Figure 5(b): Mahe (Seychelles) a mountainous peaks to some global scale phenomena such
Island with few low lying areas towards coast- as the Quasi-Biennial Oscillation (QBO).El-
lines. Niño Southern Oscillation (ENSO).
600.00
Anse Forbans
500.00 Le Noil
Bon espoir
400.00
Rainfall (mm)
La Gogu
300.00
200.00
100.00
0.00
JUL AUG SEP OCT NOV DEC JAN FEB MAR APR MAY JUN
Figure 6: Monthly mean rainfall for the four representative stations over Seychelles.
(the threshold for a wet month is 200mm).
Accumulative frequency for year 1

(Le Noil)
1600
1400
1200
1000
Rainfall (mm)
800 r/f accu
600
400
200
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73
Pentad
Figure 7: Red line (sudden rise) at pentad 65 and Green line (sudden falls) at pentad 13
Accumulative frequency for year 4

(Lagogue)
2000
1800
1600
1400
1200
Rainfall (mm)
1000 r/f accu
800
600
400
200
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73
Pentad
Figure 8: Pentad 65 shows sudden rise in rainfall while pentad 13 depicts sudden fall.
Table 1: The Factors and Eigenvalues 3.5 Results from Low Level Wind
Analysis.
Factor Eigenvalues Percentage vari-
Wind vector analysis based on
ance Explained
NCEP re-analysis data for January which
1 10.336 64.6 is the peak rainfall month over Sey-
2 1.295 8.1 chelles is shown as figure 12. The analy-
3 1.035 sis revealed that the wind has a cyclonic
6.5
curvature close to the Equator, which
4 0.806 5.0 represents the Near Equatorial Trough
5 0.557 3.5 (NET), and convergent between the
6 0.487 3.0 Northwesterly and Southeasterly flow
7 0.359 occurring around 12 degree south dur-
ing the rainy season. The research has
2.2 further shown that the wind pattern is
8 0.295 1.8 well defined during the wet season
9 0.226 (Westerlies at lower levels) and during
1.4
the dry season the wind is generally
10 0.177 1.1 easterly except for a shallow depth of
11 0.139 0.9 Westerlies. The study has also shown
12 0.106 0.7 that during the rainy season, which has
13 0.085 0.5 been defined starting from November to
14 0.052 March, NET is the main system to influ-
0.3
ence Seychelles’s weather but not the
15 0.025 0.2 I.T.C.Z, since during
16 0.019 0.1
Table 2: Loading of the Three Dominant Components on Various Stations
RAINFALL STATIONS FACTOR

1 2 3
COMPONENT LOADINGS
ANSE ROYALE 0.744 0.154 0.373
BON ESPOIR 0.800 0.318 0.003
LA MISERE 0.878 0.225 -0.033
ST LOUIS 0.895 -0.211 0.080
ROCHON 0.868 -0.124 -0.077
HERMITAGE 0.835 -0.141 -0.076
BELOMBRE 0.331 -0.695 0.557
ANSE FORBANS 0.905 0.133 0.129
TEA FACTORY 0.903 0.018 0.041
QUATRE BORNES 0.780 0.211 0.167
LE NOIL 0.903 -0.182 -0.167
GRAND ANSE 0.769 -0.268 0.057
CASCADE 0.935 -0.021 -0.005
AIRPORT 0.883 0.178 -0.276
ANSE BOILEAU 0.724 0.312 -0.043
LA GOGUE 0.426 -0.510 -0.641
Frequency accumulation for year 9

(Belombre)
4000
3500
3000
2500
Rainfall (mm)
2000 r/f accu
1500
1000
500
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73
Pentad
Figure 9: Pentad 59 represents sudden rise and pentad 17 portraits sudden falls in rainfall.
Table 3: Loading of each rainfall station on the Rotated Principal Component.
STATION Loading on the Three rotated PCA

RPCA1 RPCA2 RPCA3
ANSE ROYALE 0.771 0.343 -0.069
BON ESPOIR 0.850 0.002 0.140
LA MISERE 0.871 0.065 0.247
ST LOUIS 0.696 0.450 0.406
ROCHON 0.694 0.278 0.465

HERMITAGE 0.657 0.281 0.460
BELOMBRE 0.038 0.945 0.090
ANSE FORBANS 0.872 0.243 0.187
TEA FACTORY 0.805 0.266 0.314
QUATRE BORNES 0.803 0.178 0.068
LE NOIL 0.686 0.270 0.576
GRAND ANSE 0.557 0.440 0.404
CASCADE 0.810 0.272 0.382
AIRPORT 0.822 -0.060 0.456
ANSE BOILEAU 0.774 -0.046 0.147
LA GOGUE 0.056 0.056 0.920
Table 4(a): Communality for various stations Table 5: The communality for each station in
in region 1. the two regions with more than two stations.
STATION COMPONENT LOAD- The station with the highest communality is
bolded.
A.ROYALE 0.782 Region Station Communality
A.ROYAL 0.782
E
Region 1 B.ESPOIR 0.854
L.MISER 0.861
E
B.ESPOIR 0.854 A.FORBA 0.922
NS
L.MISERE 0.861 TEA.FAC 0.899
TORY
A.FORBANS 0.922 Q.BORNE 0.849
S
TEA.FACTORY 0.899 GRAND 0.737
ANSE
Q.BORNES 0.849 A.BOILE 0.786
AU
GRAND ANSE 0.737
Region 2 ROCHON 0.851
A.BOILEAU 0.786 ST LOUIS 0.890
HERMIT- 0.897
AGE
Table 4 (b): Communality for various sta- LE NIOL 0.906
tions in region 2.
STATION COMPONENT C A S - 0.899

LOADINGS CADE
AIRPORT 0.876
ROCHON 0.851
the peak rainfall in January I.T.C.Z lies far
south of Seychelles. Throughout the whole
rainy season, Seychelles remains within a
ST LOUIS 0.890 broad trough which at times becomes active/
organizes over Seychelles itself. The result is
similar to that obtained by Sadler et al,
HERMITAGE 0.897 (1987).
4. Conclusion
LE NIOL 0.906 The results from the study showed
that Seychelles can be demarcated into four
rainfall homogeneous regions. The main fea-
ture associated with the spatial rainfall distri-
CASCADE 0.899 bution being topography. The study also re-
vealed that although Seychelles lie near the
equator, it does not have two rainfall peaks is
AIRPORT 0.876 the characteristics of the equatorial zones,
but it has one peak rainfall season which oc-
cur from December to February. The peak
Anse Forbans (Region 1)
1600.0
y = 6.4757x + 970.33
1400.0
1200.0
.
1000.0
Rainfall (mm)
800.0
600.0
400.0
200.0
0.0
Fig 10 a
Le-Niol (Region 2)
3000.0
y = 14.655x + 1607.5
2500.0
.
2000.0
Rainfall (mm)
1500.0
1000.0
500.0
0.0
1975 /76
1977 /78
1979 /80
1981 /82
1983 /84
1985 /86
1987 /88
1989 /90
1991 /92
1993 /94
1995 /96
1997 /98
1999 /00
2001 /02
2003 /04
2005/06
Fig 10 b
Belombre (Region 3)
3000.0
y = 9.7973x + 1442.1
2500.0
.
2000.0
1500.0
Rainfall (mm)
1000.0
500.0
0.0
1975 /76
1977 /78
1979 /80
1981 /82
1983 /84
1985 /86
1987 /88
1989 /90
1991 /92
1993 /94
1995 /96
1997 /98
1999 /00
2001 /02
2003 /04
2005/06
Fig 10 c
La Gogue (Rgion 4)
2500.0
y = 8.2792x + 1179.3
2000.0
.
1500.0
Rainfall (mm)
1000.0
500.0
0.0
1975 /76
1977 /78
1979 /80
1981 /82
1983 /84
1985 /86
1987 /88
1989 /90
1991 /92
1993 /94
1995 /96
1997 /98
1999 /00
2001 /02
2003 /04
2005/06
Fig 10 d: Figure 10: The rainfall time series for the four representative stations. All the regions indi-
cate increasing trend in rainfall.
Le-Niol
4.0
3.5
.
3.0
2.5
Spectral Density
2.0
1.5
1.0
0.5
0.0
2.0
2.1
2.3
2.5
2.7
3.0
3.3
3.8
4.3
5.0
6.0
7.5
10.0
15.0
Period (Years)
Fig 11 (a)
Belombre
3.5
2.5
.
2
Spectral Density
1.5
0.5
0
2
2.143
2.308
2.5
2.727
3.333
3.75
4.286
7.5
10
15
Period (Years)
Fig 11 (b)
Figure 11: Examples of spectral density over Seychelles.

Figure 12: Mean gradient level streamline analysis over the Indian Ocean during January based on
NCEP reanalysis data (1971 to 2000). NET representing by the red line is near Seychelles while the
second NET representing by the blue line is the I.T.C.Z
rainfall month is January while minimum Basalirwa C.P.K. 1991: Rain gauge Network
rainfall is observed in July. The dates of onset Design for Uganda. Ph.D Thesis
of the rainy season over Seychelles range University of Nairobi
from pentad 59 (October) in some regions to Brier, G.W. 1977: Quasi-biennial Oscillation
pentad 65 (November) in others. It was noted and feedback processes in the atmos
that the regions with delayed onset have phere-Ocean-Earth systems. 5th conf.
early cessation. The dates of cessation range On probability and statistics in atmos
between pentad 13 (March) and pentad 16 pheric.
(May). Thus it is evident that the rainfall over Cadet, D., 1978: The importance of moving
Seychelles is associated with the NET rather tropical disturbances over the Indian
than the ITCZ. Ocean during the summer Monsoon.
Indian J. Met. Hydrol. Geophys., 29,
References 138-145.
Alusa, A.L. and M.T. Mushi, 1974: A study of Cattel, R.B., 1966: The Scree test for the num
the onset, duration and cessation of ber of factors. Multivar. Behav. Res., 1,
the rains in East Africa. Preprints, In 245-276.
ter. Trop. Met. Meeting, Amer. Met. Child, D., 1990: Essentials of Factor Analysis.
Soc., 3-146. 2nd Ed. Cassell Educational Ltd.
Asnani, G.C., 2005: Tropical Meteorology, Despande, D.V. 1967: Sunspot cycle and reg
Revised Edition; Published by Prof. ional rainfall. Weather, 23: 78-79.
G.C. Asnani, Pune, India. Edbon, R.A. 1975: The quasi-biennal Oscilla
tion and its association with strato
spheric circulation patterns. Meteorol. Okoola, R,E.A ,1996: Space-Time Character

Mag, 104: 282-297. istics of The ITCZ over Equatorial East
Hasternrath, S, 2007: Circulation mecha- ern Africa During Anomalous Rainfall
nism of climate anomalies in East Africa and Years. Ph.D Thesis University of Nai robi.
equatorial Indian Ocean. Dynamics of A Okoola, R. E., 1998: Spatial evolutions of the
tmosphere and Oceans, 43 , 25-35 active convective patterns across the
equatorial eastern Africa region during
Herman, H.H. 1976: Modern Factor Analysis,
Northern Hemisphere Spring Season
University of Chicago Press.,487pp using OLR records. Meteor. Atmos.
Kaiser, H.F., 1958: The Varimax cri Physics, vol 66, No.1-2, 51-63
terion for analytical factor analysis. Okoola, R.E., 1999: A diagnostic study of the
Psych., 23, 187-200. Eastern African monsoon circulation
Kendall, M, Stuart, A. and Ord, J.K, 1983: during the northern hemisphere spring
The advanced theory of statistics. Vol. 3. season. Int J Climatol 19: 143- 168.
Design and analysis and time series. Oludhe, C., 1987: statistical characteristics of
Charles Griffith ltd; London, pp 422- wind power over Kenya. Msc. Thesis
657. University of Nairobi.
Lawrence, E.N. 1971: Recent weather ex Jackson, I.J., and H. Winand, 1994: Towards
tremes and sunspot cycles. Weather, classification of tropical rainfall sta t i o n s .
26: 129-131. J. Climatol., 14, 263-286.
Minja, W.E.S. 1984: A comparative investiga Rasmusson, J.L. and Carpenter, R.D. 1983:
Variation in the tropical sea surface
tion of weather anomalies over Africa
temperatures and sea wind fields asso
during the 1972 Drought and 1977-
ciated with the Southern Oscillation/El
1978 wet periods. MSc Thesis Univer Nino. Mon. Weather Rev. 110: 354-384.
sity of Nairobi. Richman, M.B., 1981: Obliquely rotated prin
Nie, N.H., D.H., Bent, and C.H. Hull, 1970: Sta cipal components: An improved
tistical package for social sciences meteorological map typing technique. J
(SPSS) compiled by Glasgow Univer .Appl. Meteorol., 20, 1145-1159
sity, McGraw-Hill, UK. Sadler, J. C., M. A. Lander, A. M. Hori and L. K.
Ogallo, L.J. 1980: Time series analysis of rainf Oda, 1987: Tropical marine climate
all in East Africa. Ph.D. Thesis, Univer Atlas vol I: Indian Ocean and Atlantic
sity of Nairobi. Ocean Department of Met. University of
Ogallo, L.J. 1983. ‘Quasi-periodic patterns in Hawaii Report No. UHMET 87-01
East African rainfall records’, Kenya J. Barring, L., 1987: Spatial patterns of daily rain-
Sci. Technol., A3, 43–54. fall in central Kenya. Application of -
principal component analysis, common
Ogallo, L.J. 1988: Relationships between sea
Factor analysis and spatial correlation, J
sonal rainfall in East Africa and the
climatol.7, 7-289.
Southern Oscillation. J. Climatol., 8, 31 Basalirwa C.P.K. 1991: Rain gauge Network
- 43. Design for Uganda. Ph.D Thesis Univer-
Ogallo, L.J. 1989: The Spatial and temporal sity of Nairobi
patterns of the East African Seasonal Alusa A.L, and Mushi M (1974) A study of the
rainfall derived from principal compo onset, duration and cessation of the
nent analysis. J. Climatol., 9, 145-167 rains in East Africa. Preprints, Interna-
Ogallo, L.J., R.E. Okoola and D.N. Wanjohi, tional Tropical Meteorology Meeting,
1994: Characteristics of quasi-biennial Nairobi, 21 Jan-7Feb 1974. AMS, Bos-
oscillation over Kenya and their predict ton, 133-140.
ability potential for the seasonal raifall. Richman, M.B., 1981: Obliquely rotated prin
Mausam, 45 No. 1, 57 – 62. cipal components: An improved
meteorological map typing technique.
Okoola and Ambenje, 2003: Transition from J .Appl. Meteorol., 20, 1145-1159
Ininda, J.M., 1995: Numerical stimulation of the
the Southern to the Northern Hemi
influence of sea surface temperatures
sphere summer of zones of active con (SST) anomalies on East African sea-
sonal rainfall, Ph.D. Thesis, University of
vection over the Congo Basin. Meteorol Nairobi.
Atmos Phys 84, 255-265.
DECEMBER 2010 J.Meteorol. Rel. Sci., 4 21–33 (2010) 21
Analyses of Annual Droughts in Kenya Using an Objective Annual Rainfall Drought Index
Francis M. Mutua,
Abdel A.F. Zaki

UNESCO-Cairo Office, Cairo, Egypt.
Francis M. Mutua,
Department of Meteorology,
P. O. Box 30197
00100 NAIROBI
KENYA
Tel: 254 20 3878340
Fax: 254 20 3878343
Email : fmmutua@gmail.com
(Manuscript received 27 July 2010, in final form 01 November 2010)
ABSTRACT
Droughts are naturally imbedded in the climate system and therefore
inevitable and can be essential regulators of the climate-driven environ-
ments. However, droughts in Africa are some of the most common and often
misinterpreted natural hazards. When left improperly managed for an ex-
tended period of time, they cause unparalled suffering to the poor communi-
ties, mostly by catalyzing other community and nation-level vulnerabilities.
However, droughts do not have to be disastrous; they can even have potential
benefits when properly managed.
As a hazard, inadequate rainfall is usually the primary driver in the
formation and development of droughts, although temperature, wind and soil
moisture can also be significant causative factors. Ironically, there is no one
single and universally accepted definition of drought, an attribute which of-
ten leads to the frequent and unfortunately incorrect diagnosis of drought,
particularly within the developing countries, leading to ineffective manage-
ment of droughts in these countries.
Most commonly, droughts are studied through the use of drought in-
dices which are often used to assess the degree of drought severity. There are
a few drought related indices which exist in the literature today. However,
most of them have pertinent weaknesses which are either related to their
structure or to their limited application in the data scarce countries such as is
the case in most of the developing countries.
In this study, an annual rainfall drought index is developed using an
objective antecedent precipitation concept, based on monthly rainfall data.
The study utilizes monthly rainfall data from 26 meteorological stations in
Kenya. The temporal as well as the spatial distribution of the index is also
analyzed.
Further, the study compares the adequacy of the two commonly used
3-parameter extreme-analysis distributions, namely, the GEV and the normal
probability distributions to investigate the temporal as well as the spatial
characteristics of the index over Kenya. The Akaike Information Criteria and
the L-moment ratio tests are used to assess the goodness-of-fit of
these distributions. Both tests show that the GEV distribution fits the logarith-
mically transformed annual drought indices better that the normal distribu-
tion. The patterns for the distribution of the drought estimates for different
return periods show that drought intensities are strongest in the wet regions
of the country while the weakest intensities occur in the dry and semi-arid
regions of the country. The index performs well in objectively identifying the
drought severity and distribution in Kenya. The index can also be applied in
other regions or countries over which snowmelt and/or ice-melt do not sig-
nificantly contribute to the water balance.
Keywords
Aridity, Drought definitions, drought types, drought severity, drought indices,
Akaike Information Criterion, L-Moments
1. General Introduction and Literat the riverine environments. The natural state of
ure Review these environments fluctuates slowly with the
Aridity is an indication of the de- local and remotely connected fluctuations in the
gree of the near-permanent dryness condi- climate systems. The fluctuations in the climate
tions of the climate over a given location. system are usually caused by a combination of a
Aridity is most often indexed by the ratio of wide variety of both natural and anthropogenic
a region's mean annual potential evapora- factors. These fluctuations occasionally cause
tion to its mean annual precipitation rainfall to fall below the tolerable range and if
(McIntosh 1972). On the other hand, such condition persists for an extended period of
drought is a recurrent natural climatic time, then a drought hazard may ensue. Although
event, which stems from the lack or inade- many erroneously consider drought as a rare and
quate availability of freshwater over an ex- random event, it is actually a “normal”, recurrent
tended period of time (e.g. a season or even feature of climate. It occurs in virtually all climatic
over several years). It occurs in almost all zones (Wilhite and Glantz 1985, Wilhite 1993,
the geographical zones, whenever the Wilhite 2005), although its characteristics vary
freshwater resources fall below the nor- significantly from one region to another. Drought
mally expected levels (Wilhite 1993). How- hazards affect many countries in the world, but
ever, the characteristics of drought vary have severer impacts in Africa due to its unique
significantly from one region to another. socio-economic characteristics.
Droughts are therefore temporary anoma- Because droughts affect so many eco-
lies and as such they differ from aridity, nomic, ecosystem and social sectors, scores of
which is a near-permanent feature of cli- definitions for drought have been developed by a
mate. As such, droughts pose unprece- variety of disciplines. In addition, because
dented challenges to humankind because drought occurs with varying frequencies in nearly
they introduce disturbing and often unfa- all regions of the globe; in all types of economic
vorable dryness perturbations onto the systems and in developed and developing coun-
normality of socio-economic and environ- tries alike, the approaches taken to define it also
mental developments all over the world. reflect regional as well as ideological differences.
Impacts also differ spatially and temporarily, de-
1.1 Definitions of drought pending on the societal context of the drought. A
Generally, the main cause of drought, as a universal definition of drought is therefore an
hazard, is the persistence of unfavorable unrealistic expectation. However, definitions of
weather conditions which lead to scarcity drought can be categorized broadly as either con-
of fresh water resources, high tempera- ceptual or operational (Wilhite and Glantz 1985).
tures and strong winds - phenomena which Conceptual definitions are of the “dictionary”
are intricately imbedded in the climate type, generally defining the boundaries of the
variability. Droughts are therefore an inte- concept of drought and thus are generic descrip-
gral part of both the terrestrial as well as tions of the phenomenon, for example, “a long
DECEMBER 2010 MUTUA AND ZAKI 23
period with no rain, especially during a moisture conditions respond to precipitation

planting season”. anomalies on a relatively short scale. The state of
Operational definitions attempt to groundwater, river flow and reservoir storage
identify the onset, severity, continuation reflect the longer-term precipitation anomalies.
and termination of drought episodes. Defi- These anomalies allow different drought types to
nitions of this type are often used in an be defined conceptually and to be described in
“operational” mode. These definitions can terms of various drought indices. The distinction
also be used to analyze drought frequency, between these types are, however, rather arbi-
severity and duration for a given historical trary as different types of drought may happen
period. simultaneously.
An operational definition of agricul- The drought type terminology is effec-
tural drought might be the one that com- tively an academic exercise. One thing, however,
pares daily or monthly precipitation to the is worth stressing. Different drought "types" are
corresponding evapotranspiration rates to effectively different stages of the same natural
determine the rate of soil moisture deple- and recurring process. The deficiency of rainfall
tion and then express these relationships in (local or remotely related) effectively triggers a
terms of drought effects on the plant behav- drought. The longer and the more spatially exten-
ior at various stages of development sive this deficiency is - the more likely that, other
(World Meteorological Organization impacts (types) of the droughts will occur as the
(WMO) Press Release No 887). The effects result.
of these meteorological conditions on plant Thus, Meteorological droughts attempt to
growth would be re-evaluated continuously explain the primary causes, while agricultural and
by agricultural specialists as the growing hydrological droughts attempt to explain the sec-
season progresses. ondary impacts of the meteorological droughts.
Many disciplinary perspectives of The economical, social and environmental
drought exist. Each discipline incorporates droughts, although, not droughts in the strict
different physical, biological and/or socio- sense, (they are actually consequences of the sec-
economic factors in its definition of ondary drought impacts), are attempts to explain
drought. Because of these numerous and the tertiary impacts of meteorological droughts.
diverse disciplinary views, considerable This is illustrated on the figure below (Figure 1).
confusion often exists over exactly what
constitutes a drought (Glantz and Katz
1977). Research has shown that lack of a
precise and objective definition in specific
situations has been an obstacle to under-
standing drought, which has led to an inde-
cision or inaction on the part of managers,
policy makers and others (Wilhite et al
1986). It must be accepted that the impor-
tance of drought lies in its impacts. Thus, a
definition must reflect the regional impact
or application specifics in order to be useful
in the operational mode by operational
managers. A comprehensive review of the
enigma and definitions of drought is given
in WMO (1975) and Glantz and Katz
(1977).
1.2 Drought Types

A deficit of precipitation has differ-
ent impacts on the different components of
hydrological cycle (river flow, groundwa-
ter) as well as the components of biosphere Figure 1: Concept of the drought Definitions
(ecosystems, humans). For example, soil (Wilhite and Glantz, 1985)
Similarly, several indices have been devel- atmospheric teleconnections. These teleconnec-
oped to measure how seriously freshwater tions include the large-scale systems such as the
availability (rainfall, river flows, snowfall, El-Niño/Southern Oscillation (ENSO), Quasi-
lake levels, groundwater supplies, etc) for biennial Oscillation (QBO), and the intra-
the given ecosystem demand, during a given seasonal waves, amongst many others. The sea
period of time, deviates from historically surface temperatures over the West Indian
established norms. Although none of the Ocean, ITCZ and atmospheric pressure gradients
commonly used indices is inherently supe- over the African and Asian continents drive the
rior to the rest in all circumstances, some monsoons and the associated Low-level Somali
indices are better suited than others for cer- Jet.
tain uses in specific regions (Wilhite and The intensification and relaxation of
Glantz, 1985). these systems determines the moisture injection
into the country and the conditions for the per-
1.3 Objectives of the Study formance of either good or bad rainfall seasons
Studies of the characteristics of in the country. The figure below (Figure 2)
drought in Kenya are often inhibited by the shows the normal oscillation of the ITCZ in Af-
unavailability of drought indices which are rica.
relevant and applicable or sustainable. In It can be observed from the above figure
this study, a practical annual drought index that the local north-south oscillation of the ITCZ
that can be used to study the temporal and is most pronounced over the east African re-
spatial drought characteristics in Kenya is gions. Consequently, the ITCZ criss-crosses this
developed. The index is based on rainfall region much faster than it does elsewhere over
only. the globe. This situation introduces a high vari-
ability and unreliability in the full dominance
1.4 Drought Issues over Kenya and establishment of clear convergence over
Insufficient rainfall is the main these regions. This is therefore one of the main
driver of droughts over Kenya. In this sec- reasons why these regions, which include
tion, the rainfall characteristics over Kenya Kenya, are characterised by high seasonal rain-
are therefore examined in order to appreci- fall variability and reliability and consequently,
ate the drought issues over Kenya. a high persistence of droughts.The other major
The principle synoptic systems con- regional drivers of climate over Kenya are the
trolling rainfall in the East African region proximity to large water bodies, particularly
are the: Arabian high, ITCZ, Mascarene High, Lake Victoria, Indian Ocean and the Congo Ba-
St. Helena High and other oceanic and sin.
Figure 2: The Extreme positions of the ITCZ over the globe

(Source:http://en.wikipedia.org/wiki/ITCZ, Mutua, F.M. and Balint, Z. 2009)
Local features such as topography (which is (Mutua 2000). High spatial and tempo-
highly variable over the country) and other ral variability of the rains over the
smaller East African lakes interact with the country is a major contributor to the
regional and synoptic features to weaken problems of droughts in the country
or strengthen the synoptic climate drivers (International Strategy for Disaster Re-
in the country. Notwithstanding, the pres- duction (ISDR), 2003).
ence of African orography intensifies the • Despite its otherwise favorable loca-
cross-equatorial flow over the continent. tion astride the Equator, Kenya ex-
The cross-equatorial flow occurs even in periences a wide range of climates. In
the absence of African orography, though the arid and semi-arid regions which
this flow is much weaker even when the are mainly in Northern parts of Kenya,
Indian monsoon rainfall is high. However, rainfall is predominantly low and epi-
the location of the meridional jet near the sodic. Rainfall is fairly well distributed
equator in the Somali region is linked to the in the more humid central and west-
Indian monsoon rainfall rather than to the ern districtsof the country. The high-
land–sea contrast over Somalia. Also, the est magnitudes of annual daily rain-
presence of African orography, and not the fall for a given return period are found
strength of the Indian monsoon, controls in the arid and semi-arid regions
the vertical extent of the equatorial merid- which receive almost all of the year’s
ional wind (Chakraborty et al 2008). rainfall over an episodic period of a
However, the complex interactions few days. This creates very unfavor-
of the magnitudes and frequencies of the able conditions for rain-fed agricul-
influences of the synoptic, regional and lo- ture to be practiced reliably, thus
cal climate drivers differ from one part of complicating issues of drought man-
the country to the other (Okoola, 1998 and agement (Gommes and Petrassi,
Ogallo, 1988). In general: 1994).
• Most parts of the country have • The variability of the annual rainfall is
varying degrees of amplitude of bi- low in the central and western
modal rainfall patterns, with the parts of the country. High vari-
longer and rainier season occurring ability of annual rainfall is found in
during the March to May period. the eastern, north-eastern and
Most parts of the country also re- north-western parts of Kenya.
ceive rainfall during the October to Low rain fall variability signifies
December short-rainfall season. high annual rain fall reliability,
During this season, the country re- high annual rainfall predictabil-
ceives rainfall which has a fairly low ity-potential and low incidence o f
spatial and temporal variability country.
6 6 6
Latitude in degrees (North is positive)
5 5 5
4 4 4
3 3 3
2 2 2
1 1 1
0 0 0
-1 -1 -1
-2 -2 -2
-3 -3 -3
-4 Annual Rainfall (mm) -4 Coefficient of Variation -4

of the Annual Rainfall SKEWNESS
-5 -5 -5
33 34 35 36 37 38 39 40 41 42 33 34 35 36 37 38 39 40 41 42 33 34 35 36 37 38 39 40 41 42
Longitude (Degrees East) Longitude (Degrees East) Longitude (Degrees East)
Figure 3: Spatial Distributions of the Mean, Coefficient of Variation and the Skewness Coefficient
of Annual Rainfall in Kenya (Mutua,2000)
• drought. The opposite applies 6

SUDAN
for the case of high rainfall 5
E T H IO P IA
Longitude in degrees (North is positive)

variability (Mutua 2000). The M ANDERA
4
distribution of the skew- LO DW AR
M OYALE
ness coefficient of t h e annual 3

M A R S A B IT
rainfall is similar to that of the 2 U G AN D A W A J IR
co effi cient of variation. This K IT A L E KENYA

1
implies that annual rainfall ELDORET
KAK AM EGA
M ERU
amounts are not symmetrically 0
L ak e
K IS U M UN Y A H U R U R U
NAKURU
NYERI G A R IS S A
K E R IC H O
distributed around the
V icto ria
EM BU
K ISII NAROK
-1 J K IA
W IL SO N
mean val ues and that fre- DAG ORETTI
-2
quent dryness and oc casional M A K IN D U LAM U
episodes of heavy rainfall is a -3 T A N Z A N IA VOI

M A L IN D I
strong phenomenon in the dry -4 M O M BASA In d ian

O c ea n
parts of the country.

L o catio n (o n a m a p of K en ya ) of th e rainfall stations th at w er e used in the stud y
-5
33 34 35 36 37 38 39 40 41 42
Figure 3 shows the distribution of the mean, Figure 4: Map of(Degrees

Longitude Kenya East)
showing the
the coefficient of variation and the skewness location of the rainfall sta-
coefficient of the annual rainfall over the coun- tions which were used in
try. the study
2. Data and methodology for the study trial and error in such a way that for all the
2.1 Data used selected stations, at least one month in any
The study utilizes monthly rainfall data given year has rainfall that falls below the
for 26 (synoptic) meteorological stations in chosen threshold.
Kenya. This data was obtained from the Kenya The annual drought index is then
Meteorological Department with support defined as a power function of the absolute
through the Drought and Low-Flow Analyses sum of the normalized rainfall deficits
Component of the FRIEND†-Nile Project. Due to which fall below the threshold for every
the varying dates of commencement of opera- month of the year. The annual value of the
tion of these stations, Embu, Kericho and Kitale power factor is given by the largest relative
had 27, 29, 24 years of data respectively with run-length of the deficit-months in a given
the rest (23 stations) having data for more than year, while the multiplier coefficient in the
30 years by the year 2002. Figure 4 below gives power function is given by the fraction of
a map of Kenya showing the location of these the deficit-months in the year. Mathemati-
rainfall stations. cally, the annual drought index Di for year i,
is expressed as:
2.2 Methodology
An annual drought index in this study is Bi
 12 
developed on the basis of the understanding Di = Ai ∑ Z i , j  i = 1,2,....., N
that, socio-economic factors notwithstanding,  j =1 
the aggregated annual rainfall deficit is crucial
………… (1)
in defining the level of intensity of the annual
Where , N represents the data length in
drought. The deficits require the determination
years, the subscript j denotes the month of
of a monthly threshold for establishing the
level of the monthly rainfall deficit. It is also i, j Z
year i, and is the magnitude of the
recognized that while the total number of rain- rainfall deficit in month j of year i, which is
fall deficit-months is important in determining defined as:
the annual drought intensity, it is the maximum
run-length of the deficit-months that truly am-  0 X j > Pi ,55j %

=  X i , j − Pj
55%
plifies the annual drought. Z i, j
In this study a monthly threshold given  otherwise
 σj
by the 55-percentile of the annual monthly
rainfall was adopted. This threshold was identi-
Where, 3. Results and Discussion

X i, j Figure 5 below shows the distribu-
= the rainfall in month j of year i in the tion of the annual drought index in compari-
given station. son to the distribution of the annual rainfall
Pi ,55j % in two randomly selected rainfall stations
= the annual monthly 55-percentile of the
rainfall which has an annual monthly from the study cases.
standard deviation sj for the given Figure 5 illustrates that the drought
month j in year i. indices clearly capture the high intensities of
Ai = the fraction of the deficit-months in year the droughts of 1983 and 2000 which devas-
i, and tated almost all the parts in the country.
Bi = the largest relative run-length of the defi- Moreover, it is worthy to indicate that Ai and
cit-months in year i Bi factors succeeded to reflect the effect of
Note that all the terms on the right hand side of the drought period, so that for the same an-
Equation (1) are in themselves crude measures nual rainfall amount, Di differs according to
of drought severity in a given year. Of the three the total number of the dry months. For ex-
terms, B is strongest, while A is also a key term ample in the central parts of Kenya, although
in determining the drought severity in a given the annual rainfall for years 1969, 1976,
year. It is for this reason that the term B is 1996 and 2000 is about 700 mm, there is a
made the power exponent of the Z. Several significant difference in Di values ranging
other combinations of these terms were tried from 0.7 to 2.2 as shown in Figure 6. The
and tested. However, it is the combination same applies on the North-western part of
given by Equation (1) that yielded favorable Kenya, where for low annual rainfall amount
results. (about 80 mm), Di differs from 0.54 t0 1.29
Large values of the drought index D indicate reflecting the impact of the long continuous
intensified annual drought conditions while drought period especially in the severest
drought year of 2000 where Di=1.29
D a g o r e tti C o r n e r in N a ir o b i - C e n tr a l p a r t s o f K e n y a
1800 3 .0 0
A n n u a l R a in f a ll ( m m )
1600 D r o u g h t In d ic e s
2 .5 0
Annual Drought Indices

1400
1200 2 .0 0
Annual rainfall
1000
1 .5 0
800
600 1 .0 0
400
0 .5 0
200
0 0 .0 0
1955
1957
1959
1961
1963
1973
1975
1977
1979
1981
1983
1985
1987
1991
1993
1995
1997
1999
2001
1965
1967
1969
1971
1989
Y e a rs
L o d w a r in N o r t h - w e s te r n K e n y a
700 1.40
600 1.20
A n n u a l R a in f a ll (m m )
Annual Drought Indices
D r o u g h t In d ic e s
500 1.00
Annual rainfall
400 0.80
300 0.60
200 0.40
100 0.20
0 0.00
1956
1960
1962
1964
1966
1968
1970
1972
1950
1952
1954
1958
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
Ye a rs
Figure 5: Distribution of the Annual Drought Index in Comparison to the Distribution of the
Annual Rainfall in two selected locations in Kenya.
6 6 6
5 5 5
4 4 4
3 3 3
2 2 2
1 1 1
0 0 0
-1 -1 -1
-2 -2 -2
-3 -3 -3
The Mean of the Annual The Coefficient of Variation
Drought Indices of the Annual Drought Indices Eigenvector - Weghted
-4 -4 -4 Sum of the Seven
Significant Factors
-5 -5 -5
33 34 35 36 37 38 39 40 41 42 33 34 35 36 37 38 39 40 41 42 33 34 35 36 37 38 39 40 41 42
Figure 6: Distribution of the Mean, Coefficient of Variation and the Eigenvector-Weighted Sum
of the Significant Factors of the annual drought indices over Kenya.
3.1 Spatial Analysis of the Annual distribution location, scale and shape pa-
Drought Indices rameters respectively.
The spatial distribution of the mean The method of probability weighted mo-
annual drought indices and the spatial variabil- ments (WMO, 1989 and Hosking et al., 1985)
ity of the indices are shown on the Figure 6. was utilized to estimate the corresponding
It is noteworthy that higher mean annual val- parameter values. For the GEV distribution,
ues of the drought index are found in the areas the probability weighted moments (PWM)
parameter estimates µˆ , α and κ
of higher mean annual rainfall and vice versa. ˆ ˆ in
This tends to contradict the common percep-
Equation (3) are given as follows:
tion that droughts are commonest in the dry
arid and semi-arid areas. The drought index
therefore seems to differentiate dryness from
drought which can occur not only in the wet kˆ = 7.8590 * C + 2.9554 * C 2
areas where it has more serious implications, 2 * Mˆ 1,1,0 - Mˆ 1,0,0
but also in the dry areas of the country. where C =
3 * Mˆ 1,2,0 - Mˆ 1,0,0
Temporal Analyses of the Annual Drought
Indices. kˆ * ( Mˆ 1,1,0 - Mˆ 1,0,0, )
αˆ =
Simple autocorrelation analysis of the
annual drought indices showed the existence of [
Γ(1 + kˆ) * (1 - 2- k )
ˆ
]
very weak serial persistence in all the rainfall and µˆ = Mˆ 1,0,0 + αˆ [ Γ(1 + kˆ) - 1]/ kˆ
stations. The samples of the annual drought
indices were therefore assumed to represent
In the above expressions,
statistically independent annual extreme N
1
events.
The logarithmically transformed values
Mˆ 1, j , 0 =
N
∑ (F
i =1
i
pp
) j yi ... j = 0,1,2,..... and Fi P
of the D values were used to evaluate the ade- On the other hand, the three parameter log-
quacy of the GEV and the Normal distributions normal probability density function f(y) is
in modeling these indices. The GEV distribution given by:
for the log-transformed variate y can be ex-
−
[ln( D − D 0 ) − µ y ]2
pressed in terms of its cumulative distribution
1 2σ 2
function F(y) as: f ( y) = e y
 (y − µ)
1/k
( D − D 0 )σ y 2π
F ( y ) = exp{ − 1 − k  }
 α  ………….. (4)
…………….. (3) Where Do is the lower/upper bound and my
In Equation (3), -¥<y<¥, and m, a, and k are the
and sy are the mean and standard deviation of

the logarithms of (D-Do) respectively. Hereafter
this distribution is abbreviated as Log-
normal(3). The moment estimates of the Log-
n
1 n
 k
AIC[log− GEV]) = −n ln(α ) − ∑ ln(yi ) + ( − 1)∑ ln1 - ( ln
normal(3) were used in this study and are i =1 k i =1  α
given by the following set of expressions: … (6)
)
) σD On the other hand, the AIC for the Log-
D0 = µ D −
h normal(3) distribution (denoted as
) AIC[Lognormal(3)]) is given by:
σ
µ y = ln( D ) − 0.5 * ln(h 2 + 1) n
h AIC [log normal ( 3 )] = − ∑ ln( D i − D 0 ) − n ln( σ 2π ) −
)
[
σ y = ln(h 2 + 1 ] … (7)
i =1
where The notation used in Equations 6 and 7

above was explained earlier.
1 − w2 / 3
h= Hosking (1986) shows that the L-moments
w1 / 3 which are based on the PWM's are an effi-
and cient diagnostic tool, having a good ability, to
identify the optimum model from a set of
− γ D + (γ D2 + 4) competing probability distribution models in
w= a given study case. In terms of the previously
2
) ) defined quartiles (M1, j, 0), the L-moments lj
In these estimates, µ D , σ D and gD are can be defined as:
the sample estimates of the mean, standard de- j
viation and skewness respectively of the un- λ j +1
= ∑ p
k = 0
j, k M 1, k ,0
transformed annual drought indices.
The AIC and the L-Moment ratio good- …………….…. (8)
ness-of-fit tests were used to identify the bet- In Equation 8
ter-fitting model amongst the two models
Pj , k =
(− 1) j − k * ( j + k )!
which are used in this study. These goodness-
of-fit tests are explained in the following sec-
( j − k )!*(k!)2 For any
tions. integer j³0 and k£j. …… (9)
The Akaike Information Criterion Goodness The expressions for the first four sample L-
of Fit Test moments from Equation 8 are:
The AIC test is given by (Mutua 1994):  λ1   1 0 0 0   M 1, 0, 0 
λ   − 1 2  
0   M 1,1, 0 
n
AIC = −2∑ log{ f ( y i )} + 2k  2 =  0
×
i =1  λ3   1 − 6 6 0   M 1, 2, 0 
…………………………….. (5)      
Where, f(y) is any one of the two distributions λ 4  − 1 12 − 30 20  M 1,3, 0 
under investigation each with k=3 parameters. ………………………… (10)
(i − 0.38The
)
PP
= ibest fitting model is the one which
= 1,2,..., N
( N + 0.2)
achieves the minimum value of AIC. Since each Cunane in WMO (1989) proposes the use of
of the two models has three parameters, then the plot of the dimensionless L-moment ra-
the term 2k contributes exactly the same tios t3 = l3/l2 versus t4 = l4/l2 rather than that
amount in Equation 5 and can safely be omitted of the L-skewness l3 and L-kurtosis l4 for the
in the test. Similarly, the constant coefficient (- identification of the best fitting probability
2) can also be omitted, but this time, the most distribution. The range for both t3 and t4 is (-
adequate model would be the one which con- 1, +1).
sistently gives the maximum of the modified The functional forms of the L-moment and L-
AIC. moment ratios in the real-domain for the
Consequently, the modified AIC for the log-GEV Log-GEV and Lognormal(3) distributions
distribution (denoted as AIC[log-GEV]) is given were not readily available. Consequently, the
by:
Table 1 The AIC Estimates for the Log-GEV and Log-Normal Distributions for the selected
Study-Stations.
AIC(Log- AIC(Log- AIC(Log-

STATION Normal) GEV) STATION Normal) AIC(Log-GEV)
DAGOR-
ETTI -24.6417 -20.0923 MALINDI -21.1718 -23.4807
ELDORET -22.9911 -25.6039 MANDERA 22.94262 21.16908
EMBU -23.4947 -28.7687 MARSABIT -20.7338 -17.7445
GARISSA -12.5477 -16.7133 MERU -12.7294 -13.9342
JKIA -13.3662 -17.4687 MOMBASA -15.1236 -16.6499
KAKAMEG
A -25.6632 -29.9772 MOYALE -32.1101 -30.9692
KERICHO -19.7137 -22.5501 NAKURU -25.9924 -26.0212
KISII -19.7338 -29.6639 NAROK -28.5916 -29.4673
NYAHU-
KISUMU -24.003 -27.0423 RURU -26.7439 -29.2918
KITALE -13.2372 -14.9948 NYERI -28.085 -31.9689
LAMU -20.0719 -22.4144 VOI -13.339 -14.706
LODWAR 19.58171 14.10155 WAJIR 2.413204 0.831579
MAKINDU -12.0014 -14.2658 WILSON -15.8346 -16.3397
samples (appropriately converted into the real- Log-GEV distributions.

domain values) of size 1000 in each station as The L-moments plot shows that the
follows: Log-GEV distribution approximates the sample
estimates better (R2=98%) than the Log-
l3(1000) l 4(1000)
τ 3(1000 ) = τ (1000 )
4 = normal(3) which has R2=94.5%.
l 2(1000) and l 2(1000) On the basis of the good performance of the
log-GEV in modeling the drought indices in
Where the ℓ’s are the estimates of λ’s as de-
Kenya, the distribution is therefore used
fined in Equations 8, 9 and 10 for a sample size
hereafter to estimate drought magnitudes
of 1000 simulated values.
for different return periods.
Table 1 gives a summary of the Log-GEV
Return Periods for the Computed Annual
and Log-Normal distributions for the selected
Droughts
study-stations.
The Log-GEV distribution was used
The log-GEV scores best in 23 cases
to estimate the drought magnitudes for re-
(bolded in Table 1) out of the total of 26 cases
turn periods of 50, 200 and 500 years in the
which accounts for 88.5% of the log-GEV suc-
stations used in this study. These return pe-
cess. Interestingly, the Log-GEV performs al-
riods were chosen arbitrarily. The estimates
most as well as the Lognormal(3) in the only
three exceptional cases in which Lognormal(3) ŷT of the log-transformed drought indices
performs better.. for different return periods (T) were esti-
3.2.2 The L-Moment-Ratio Goodness-of-Fit mated using Equation 13 given below:
α  1   
Test kˆ
 
Figure 7 shows the (t3, t4) scatter for the sam- yˆ T = 1 −  − ln  1 −   + µˆ
ple dimensionless annual drought indices, as k    T   
 
well as those for the fitted Lognormal(3) and
……………………………………………..……….. (13)
0 .5 0
S am p le L - M o m e n t s
0 .4 0
L o g- G E V ( R ^ 2 = 9 8 % )
L o g- n o r m a l( 3 ) ( R ^ 2 = 9 4 .5 % )
0 .3 0
L-Kurtosis
0 .2 0
0 .1 0
0 .0 0
L -M o m e n ts fo r A n n u a l D r o u g h t I n d i c e s
- 0 .1 0
- 0 .1 0 .0 0 .1 0 .2 0 .3 0 .4 0 .5 0 .6 0 .7
L -S k e w n e ss
Figure 7: The sample L-moment and the GEV distribution L-moments for the Annual Drought Indi-
ces data in Kenya.
6 6 6
5 5 5
4 4 4
3 3 3
2 2 2
1 1 1
0 0 0
-1 -1 -1
-2 -2 -2
-3 -3 -3
-4 -4 -4
RETURN PERIOD=50 YEARS RETURN PERIOD=200 YEARS RETURN PERIOD=500 YEARS
-5 -5 -5
33 34 35 36 37 38 39 40 41 42 33 34 35 36 37 38 39 40 41 42 33 34 35 36 37 38 39 40 41 42

Figure 8: Distribution of annual drought indices corresponding to the 50-year, 200-year and 500-
year GEV return periods
Values of the untransformed drought estimates gradients are observed in the dry and semi-
Dˆ T = e yˆT were estimated for the chosen return arid northern and eastern parts of the coun-
try. This may be attributed to the low varia-
periods in all the other rainfall stations. The tion in the annual rainfall in the dry and
spatial distribution for the estimated magni- semi-arid northern and eastern parts of
tudes for each of the chosen return periods is Kenya and the accountability of Di on the
shown with a corresponding map on Figure 8. deficit in rainfall from its average values. The
The patterns for the distribution of the spatial gradients of the annual droughts in
drought estimates for different return periods these parts are near zero, implying that
show that the drought intensities increase al- weak droughts of almost the same magni-
most logarithmically as the return period in- tude cover large regions.
creases. Further, the drought intensities are
generally stronger over the wet regions 4. Conclusions
(central districts) of the country than over the Droughts are common in Africa, al-
drier regions (eastern and northern districts). though some countries are more vulnerable
The spatial gradients of the estimates are also to drought hazards than others. Unlike other
large in these regions, implying that strong types of hazards, droughts are more difficult
droughts cover smaller areas in these regions. to define and therefore difficult to manage.
However, the weakest annual droughts and
They become disasters when the human in- Hosking, J.R.M., Wallis, J.R. and Wood, E.F.
duced environments become highly vulnerable (1985): Estimation of the General-
to the risks of the drought hazards. Drought ized Extreme-Value Distribution by
disasters enhance poverty and act as drought- the method of probability weighted
vulnerability catalysts. moments. Technometrics, 27, 251-
In Kenya, inadequate rainfall plays a 261.
key role in the development of drought. Studies International Strategy for Disaster Reduction
of the characteristics of drought in the country (ISDR) (2003): Turning the tide on
are often inhibited by the lack of practical disasters towards sustainable devel-
drought indices. In this study, a meteorological opment. World disaster reduction
annual drought index is developed. Monthly campaigns 2003.
rainfall data from 26 meteorological stations in McIntosh, D. H. (1972) Meteorological Glos-
Kenya is used in the study. Large values of the sary, Her Majesty's Stationery Office,
drought indices were found to indicate intensi- Met. Office - 842, A.P. 897.
fied annual drought conditions while low val- Mutua, F.M. (1994): The use of the Akaike
ues indicate mild or no annual drought condi- Information Criterion in the Identifi-
tions. The developed index succeeded to reflect cation of an Optimum Flood Fre-
the variability in the drought severity and tem- quency Model. Hydrological Sciences
poral distribution in the analyzed period of the Journal Vol 39, (3).
used stations. Results show that higher mean Mutua, F.M. (2000): Comparison of the
annual values of the computed annual drought Trend and GCM based Projections of
indices were prominent in the areas of higher Seasonal and Annual Rainfall in the
mean annual rainfall and vise versa. East African Region. African Journal
All the time-series of the annual drought of Science and Technology (AJST) Sci-
indices showed no significant serial depend- ence and Engineering Series Vol. 1,
ence. The study compares the adequacy of the No. 1, pp 26-38.
two commonly used 3-parameter extreme- Mutua, F.M. and Balint, Z. (2009): Analysis of
analysis distributions, namely, the GEV and the the General Climatic Conditions to
Normal probability distributions. The AIC and Support Drought Monitoring in So-
L-moment ratio goodness-of-fit tests collabo- malia. Technical Report No W-14,
rate in confirming that the GEV distribution fits FAO-SWALIM Nairobi, Kenya
the logarithmically transformed annual Ogallo, L.J. (1988): Relationship between
drought indices better that the Normal distri- seasonal rainfall in East Africa and
bution. The patterns for the distribution of the the Southern Oscillation. Int. Jour.
drought estimates for different return periods Climatol. 8:31-43
show that drought intensities are strongest in Okoola, R.E. (1998): Characteristics of the
the wet regions of the country while the weak- Inter-Tropical convergence Zone
est intensities occur in the dry and semi-arid (ITCZ) over Equatorial Eastern Africa
regions of the country. based on Station-Rainfall Records.
References Jour. Africa. Met. Society 3(1): 61-101.
Chakraborty, A, Nanjundiah, R.S. and Sriniva- Wilhite, D.A. (1993). Understanding the Phe-
san, J. (2008): Impact of African orogra- nomenon of Drought. Hydro-Review
phy and the Indian summer monsoon 12(5):136–148
on the low-level Somali jet. Int. J. Clima- Wilhite, D.A. 2005. Drought. In: J.E. Oliver
tol. DOI: 10.1002. (ed.) Encyclopedia of World Climatol-
Gommes, R.; and F. Petrassi. (1994). Rainfall ogy, pp. 338-341. Springer.
variability and drought in Sub-Saharan Dordrecht, The Netherlands.
Africa since 1960. Agrometeorology Se- Wilhite, D.A.; and M.H. Glantz. (1977): When
ries Working Paper No. 9, Food and Ag- is a Drought a Drought? Nature 267,
riculture Organization, Rome, Italy. 192-3.
Hosking, J.R.M., (1986): The Theory of Prob- Wilhite, D.A.; and M.H. Glantz. (1985): Un-
ability Weighted Moments. IBM Math. derstanding the Drought Phenome-
Report, RC12210, Yorktown Heights, non: The Role of Definitions. Water
New York, 160p. International 10(3):111–120.
Wilhite, D.A.; N.J. Rosenberg; and M.H. Glantz. World Meteorological Organization (WMO)
1986. Improving Federal Response to (1989): Statistical Distributions for
Drought. Journal of Climate and Ap- Flood Frequency Analysis. Opera-
plied Meteorology 25(3):332–342. tional Hydrology Report No. 33. Ed-
World Meteorological Organization (WMO) ited by C. Cunane. WMO – No. 718,
(1975): World Meteorological Organi- Geneva, Switzerland.
zation (WMO) 1975. “Intercomparison World Meteorological Organization (WMO),
of conceptual models used in opera- (2010): Press Release No. 887,
tional hydrological forecasting,” Op- “Experts Recommend Agricultural
erational Hydrology ReportNo. 7. Drought Indices for Improved Un-
WMO-No. 429. Geneva, Switzerland. derstanding of Food Production Con-
ditions” Geneva/Murcia, 8 June
DECEMBER 2010 J.Meteorol. Rel. Sci., 4 35–47 (2010) 35
Meteorological factors and measles occurrence in Akure,Ondo State, Nigeria
Akinbobola and J.Bayo. Omotosho

Federal University of Technology,
P.M.B 704 Akure.
Nigeria
Akinbobola
Federal University of Technology,
P.M.B 704 Akure.
Nigeria
E-mail- desmondbobola@yahoo.com
Phone no- +234 (0) 8034619238
(Manuscript received 10 September 2010, in final form 01 November 2010)
ABSTRACT
This study investigates the relationship between weather parameters,
such as temperature and relative humidity and the incidence of measles in
Akure, the Ondo State capital. Monthly data of (in and out patients) measles
patients between 1990 and 2003 were collected from Ondo state specialist
hospital, Akure. Also monthly relative humidity, rainfall and maximum tem-
peratures for the same years were collected from Nigeria Meteorological
agency, Oshodi, Lagos.
It was found that measles disease is more prevalent during the hot
dry season and lowest during the wet season. Analysis also showed that these
Meteorological variables have significant correlation with the occurrence of
measles. The disease is most prevalent when the temperatures falls between
32oC and 34oC and relative humidity of between 60% and 80%.Furthermore,
at temperatures below 32oC and relative Humidity of less than 60%, there
were very few measles patients. Finally, statistical analysis shows that simple
multiple regressions particularly rainfall, combination, of temperature and
rainfall, rainfall and relative humidity, and the combination of the three vari-
ables gave good estimates of measles patients. This will be of great use to
health planners and government agencies in the efficient delivery of health
care services.
Keywords: Measles; weather elements; hot dry season; multiple regressions
1. Introduction their habitat at the social, psychological and

Measles is one of the commonest dis- physiological levels, and this may manifest in a
eases in the tropical region of the world. In range of behaviors (Voracek et al. 2007). How-
Nigeria the occurrence of this disease is com- ever, little or nothing is in the literature about
mon among the children and it is considered the relationship between weather and measles
to be one of the diseases that cause death
in Nigeria. In Nairobi, Kenya, it has been ob-
among both the young and the old.
Weather varies from season to season and served that a high number of measles patients
influences populations living under these con- were reported between May and September
ditions. It also has impact on individuals in which are cold months, while between
December and February corresponding to vulnerable than rural populations because they
dry period, very few patients are reported. experience higher temperature. Furthermore, in-
(Nganga and Ngugi, 1986). Persinger, 1980 dividuals are more vulnerable to heat stress if
found that variability in occurrences of they are in poorly designed housing, with no ac-
many diseases is related to seasonal trends cess to air conditions or well ventilated and
in temperatures, (although significant year- cooler buildings. This is a common situation in
to-year differences do occur. Bronchitis, Nigeria.
peptic ulcer, adrenal ulcer, glaucoma, goiter This work investigates if any relationship
and eczema are related to seasonal varia- exists between temperature, relative humidity
tions in temperature. (Tromp, 1963). In the and rainfall and measles occurrences and if these
Sudan, Sharaf-Eldein, (1992) found that a parameters can be used to explain the monthly,
good correlations exist between some Me- inter-annual and seasonal variation in measles
teorological variables and asthma, malaria occurrence. The parameters which have the most
and typhoid fever. Futhermore, seasonal significant influence on the disease are further
occurrence of dust haze pollution during studied using statistical analysis to ascertain this
harmattan season has been documented by level of significance.
many research studies (Hamilton and Arch-
bold 1945, Aina 1972, Kalu 1978, Adebayo 2. Study Location
1978). Adefolalu, (1984) found that maxi- Akure is the Capital city of Ondo State
mum number of patients with respiratory with population of about 500,000 people. It is
problems occurred during the dry winter situated in the humid forest zone of Nigeria. It has
months while the minimum was in rainy an annual rainfall which exceeds 1400mm, and
season. The increasing intensity of thick falls between March and November. Average rela-
dust veils which laboratory analysis has tive humidity is about 55% during the dry season
shown to be carcinogenic (cancer causing) and 90% during the rainy (wet) season. The mean
apart from confirmed presence of DDT, Al- daily maximum temperatures ranges from 260C
derin and other chemicals used as pesti- to 350C, while the mean daily minimum tempera-
cides and herbicides is a major threat to tures range from 150C to 230C.
health conditions.
Temperature is a major factor in 3. Data and Methods
human well-being. Hot weather extremes The data set consisting of medical and me-
have a more substantial impact than cold teorological data were used for this study. The
weather and many heat stress indices have medical data consists of the monthly number of
been developed to assess the degree of im- reported cases of measles attack from 1990 to
pact (Quayle and Doehring, 1981, Kalkstein 2003. This was extracted from the records of the
1982, Steadman 1984). In most homes, the Ondo State Specialist Hospital, Akure which is the
indoor temperature ranges between 170C major government hospital located within the city
and 310C. Human being cannot comfortably and the only one that has continuous data for a
live in temperature outside this range. The long period of years.
tolerance range on any individual is usually Monthly maximum temperature (0C),
less than this and tends to get narrower Relative Humidity (%) and Rainfall (mm). from
with age of infirmity. The temperature of the Nigerian Meteorological Agency, Oshodi, La-
surrounding air is the most significant fac- gos were also used for the study. The measles
tor for human comfort. Outside the com- data and weather variables were analysed using
fortable range, as temperature increases, time frames ranging from the monthly, seasonal
thermal stress leads progressively to and annual levels. For each time frame, the mean
greater discomfort, physiological stress, ill of all the variables over the period were calcu-
health and even death. Heat can cause sev- lated. Also temperature and relative humidity
eral clinical syndromes. Heat stroke which ranges were formulated to establish the threshold
occurs when the body temperature exceeds values of these variables that correspond with the
40.60C, it is usually fatal. A record breaking peak of the occurrence of measles in the area. The
heat wave in June 1998 in central Russia ranges are, for Temperature, 26 to 28, 28.1 to 30,
caused more than 100 deaths (WMO, 30.1 to 32, 32.1 to 34 in 0C and for Relative Hu-
1999). Urban populations are more midity; 40 to 60, 60.1 to 80 and 80.1 to 100.(%)
DECEMBER 2010 AKINBOBOLA AND OMOTOSHO 37
These ranges were formulated based This analysis was carried out by the use of some
on the lowest and highest values and the statistical packages like, XLSTAT 2010, SPSS and
distribution of these variables. excel. The relationship was taken to be significant
at 0.05
4. Statistical analyses
Both simple and multivariate linear 5.0 Results and Discussions.
regression analysis was performed. (Anne 5.1 Monthly measles occurrence and
etal 2007) This was done in order to in- weather parameters
vestigate the individual contributions of the There is significant monthly variation in
various meteorological parameters and measles occurrence and weather variables as can
thus isolate the variable combination that be seen in Fig 1 (a-c)
best estimates the number of measles pa- While temperature has the same general pattern
tients for the station. The simple and multi- of variation with measles, decreasing from Janu-
variate linear relationships are of the form: ary to August and increasing thereafter to Decem-
M1 = a + b (TT)…………… ……………….…… (1) ber, an opposite pattern was observed with rela-
M2 = a + b (RH)……………………………..…… ((2) tive humidity and rainfall, suggesting a negative
M3= a + b (RR)…………………………...……..… (3) relationship.
M4 = a + b (TT) + c (RH)…………………….… (4)
M5 = a + b (TT) + c (RR)……………….……… (5) 5.2 Meteorological Parameters Threshold
M6 = a + b (RH) + c (RR)………………….…… (6) and Measles occurrence
M7 = a + b (TT) +c (RH) + d (RR)……..…… (7) Based on the general patterns of behavior
of measles variation with the meteorological pa-
Where, M1-M7 is the number of estimated rameters, they were divided into group of 20C and
patients the relative humidity into groups of 20% incre-
TT is monthly mean Temperature ments. The number of measles patients observed
RR is the monthly mean Rainfall in each temperature and relative humidity groups
RH is the monthly mean Relative were then tabulated and used to obtain Fig 2 (a-c)
Humidity.
Fig 1a: Monthly mean number of measles patients ( ) and Temperature (. )

Fig 1b: Monthly mean number of measles patients ( ) and Relative humidity ( )
Fig 1c: Monthly mean number of measles patients ( ) and Rainfall ( )

Fig 2a: Number of measles patients against relative humidity for temperature between 26 and 280C
Fig 2b: Number of measles patients against relative humidity for temperature between 28.1 and
300C.
Fig 2c: Number of measles patients against relative humidity for temperature between 30.1 and 32 0C
Fig 2d: Number of measles patients against relative humidity for temperature between 32 and 34 0C
Table 1: Regression coefficient and correlation values (equation 1 - 7) using monthly means of pa-
rameters significant at 0.05
Equation R Multiple Adjusted P-value

R R square
1 M1 = -42.43 + 1.92TT 0.70 0.43 0.012
2 M2 = 76.22 – 0.77RH 0.59 0.002

-
0.79
3 M3 = 25.33 – 0.07 RR - 0.81 0.000
0.90
4 M4 = 41.68 + 0.70TT – 0.60RH 0.81 0.58

TT*0.06
RH*0.4
13
5 M5 = 23.3 + 0.06TT – 0.06RR 0.91 0.79

TT*0.91
RR*0.0
02
6 M6 = 17.36 – 0.07RR+ 0.12RH 0.91 0.79

RR*0.0
RH*0.7
11
7 M7= 13.04 + 0.11TT-0.07RR+0.13RH 0.91 0.76

TT*0.70
RR*0.8
RH*0.0
22
The threshold values of Meteoro- on environmental factors as strong triggers.

logical parameters that correspond to vari- The correlations of the estimated number of pa-
ous levels of occurrences of measles were tients with the actual are shown in Fig.3 (a-g).
identified in Fig 2. The significant result Also shown in Fig. 4 are the monthly
here is that the highest attack of measles variations of the actual and estimated number of
occurs for temperatures between 32-34oC patients. It is clear from the figure that good esti-
with a corresponding relative humidity of mates of the number of patients could be ob-
between 60.1%-80%. At other tempera- tained using the various correlation equations.
tures, only few patients are recorded, par- Although there is good correspondence between
ticularly for humidity below 80%.. the actual and estimates from all the equations.
The results in Table 1 show that Equations 3,5-7 gave better monthly estimates of
correlation coefficients are generally high, measles patients.
suggesting a clear dependency of measles
Fig. 3a
Fig. 3b
Fig. 3c
Fig. 3d
Fig. 3e
Fig. 3f
Fig. 3g
Fig 3(a-g): Observed and estimated number of patients using equations 1-7 respectively. The
dashed lines are the trends for perfect agreement between observed and estimated values.
Fig 4a: Plots of Observed and estimated number of patients at Akure from Equations 1-3.
Observed ( ) Eq 1 ( ) Eq 2 ( ) Eq 3 ( )
Fig 4b: Plots of Observed and estimated number of patients at Akure from Equations 4-7
Observed ( ) Eq 4 ( ) Eq 5 ( ) Eq 6 ( ) Eq 7 ( )
6. Conclusions Research 82 (11) (1977)

Monthly temperature, relative hu- Anne E Jones, Ulrika Uddenfeldt Wort, An
midity and rainfall for the period 1990- drew P Morse, Ian M Hastings and Alex a n
2003 for Akure, a city in the southwest, Ni- dre S Gagnon (2007) Climate predict tion
geria have been used to establish a rela- of El Niño malaria epidemics in north-west
tionship with Measles occurrences. Also, Tanzania, Malaria Journal 2 0 0 7 , 6:162
threshold values of these Meteorological doi:10.1186/1475-2875-6- 162.
variables showed that high measles attack Hamilton R.A and Archbold, J.W (1945): Meteorol
occurred for a distinct temperature range ogy of Nigeria and adjacent territory, Quart. J.
of 32-340C if the relative humidity is not Royal Meteoro. Soc., 71, 231-265
too high (60%-80%). The high correlation Kalkstein, L.S., 1982: The weather stress index.
coefficients obtained suggest that the linear NOAA Technical Procedures Bulletin, 324,
multiple regression equations can be used 1-16.
to estimate the number of measles patients Kalu A.E 1978. The African dust plume: Its charac
in the city. These results would be very use- teristics and propagation across West Africa I
ful for more efficient health care planning n winter. In Saharan Dust- Mobilization,
and delivery, especially if the monthly val- Transport, Deposition, Morales C (ed). SCOPE
ues of temperature and rainfall could be 14 Publication, John Willey and Sons: New
predicted in advance. This is the subject of York; 95-118
an ongoing study. Nganga and Ngugi (1986): Influence of weatheron
measles and pneumonia in Nairobi, climate
References and human health WCAP No.1.38.
Adefolalu, D.O, (1984): On Bioclimatologi- Persinger, M.A., 1980: The Weather Matrix and
cal aspects of harmattan Dust Haze in Human Behavior, New York: Praeger, 327
Niger Arch. Met. Geophy. Biocli. Sc. B33 pp.
pp 387-404 Quayle, R., and F. Doehring, 1981: Heat stress: A
Aina, J.C., (1972). A contribution to the comparison of indices. Weatherwise, 34,
forecasting of ... of Saharan dust at solar 120-124.
wavelengths, Journal of Geophysical
Sharaf-Eldein H.I (1992): A study into the Tromp, S.W., 1963: Medical biometeorology, New
occurrence of some common diseases York: Elsevier.
in relationship with some atmos- Voracek M, Tran US, Sonneck G (2007) Facts and
pheric variables in Khartoum: An un- myths about seasonal variation in suicide.
published diploma project at Univer- Psychol Rep 100:810-814. Doi:
sity of Khartoum, The Sudan 10.2466/PRO.100.3.810-814
Steadman, R.G., 1984: A universal scale of WMO (1999) Weather, climate and health.
apparent temperature. Journal of Cli- World Meteorological Organization No.
mate and Applied Meteorology, 23, 892, Geneva, Switzerland
1674-1687. 36pp
DECEMBER 2010 J.Meteorol. Rel. Sci., 4 49 –60 (2010) 49
Effects of Urbanization on Climate of Nairobi City
Ongoma, V.
Department of Meteorology, University of Nairobi
Nzioka J. Muthama,
John K. Ng’ang’a
Victor Ongoma
P.O. Box 30197 00100, Nairobi
Email: victor.ongoma@gmail.com
(Manuscript received 23 July 2010, in final form 01 November 2010)
ABSTRACT
The Kenya’s capital city-Nairobi is one of the fastest growing cities in
the world. There is necessity to investigate the effects of urbanization on the
climate of our city. This study looked into the long-term urban modification
of annual weather conditions by studying the relation between the weather
elements and urbanization indicators.
The results showed that Nairobi city is growing at a high rate as evi-
denced by a population growth of about 6.9%. The temperature was found to
have been modified by urbanization such that there is a warming trend; the
stations that were more urban recorded a high rate of temperature increase
as compared to the less urban stations revealing the formation of an urban
heat island. Rainfall trend was similarly found to be increasing with time;
however, the change in humidity was found to be insignificant. It further re-
vealed that some of the long term changes in climate over the city were due
to changes in surface and atmospheric characteristics of the city such as sur-
face roughness, smoke and destruction of vegetation cover, as a consequence
of urbanization.
The environmental implications such as urban surface temperature
modification on human comfort are significant. It was suggested that in or-
der to control these trends and to obtain more favourable and healthy cli-
matic conditions, the acreage of green areas must be increased. Incorpora-
tion of the findings in the future planning of the city is hence recommended.
1. INTRODUCTION by higher temperature, lower humidity and

Given the large and ever-increasing rainfall, and weaker winds than surrounding
number of urban dwellers globally, and the rural areas. These differences in climate
profound effects of cities and their inhabi- parameters vary depending on the factors such as
tants on the atmosphere, both within and the presence of industrial areas emitting exces-
beyond urban limits, ever-increasing atten- sive heat or air pollutants, urban density, the ori-
tion is being directed to the study of urban entation of streets, topography and population of
climates. cities, amount and type of green areas, form and
Urban climates are highly modified heat capacity of buildings.
local climates and are often characterized Urbanization is the process by which an
increasing proportion of the population urbanization on mean temperature fields, an-

comes to live in urban centres (Hardoy and nual rainfall and relative humidity patterns.
Satterthwaite, 1989). Densely populated Nairobi has experienced some of the highest
human settlements imply an increased use growth rates of any city in Africa. Since its
of fuels for various kinds of application and foundation in 1899, Nairobi has grown to be-
the result is increased release of heat and
come the largest city in East Africa, despite
pollutants into the environment. Industries
and factories release heat and pollutants
being the youngest large city in the region.
into the urban boundary layer. In addition, According to GoK (1999) the growth rate of
an increase in motor vehicles result in a Nairobi is currently at 6.9% and it is esti-
high consumption of fossil fuel resulting in mated that the population will reach 3 038
an increase in heat and air pollutants 553 in 2008, 3 138 295 in 2009, 3 240 155 in
emission into the urban boundary layer 2010 which has been reflected in the graph
(Ng’ang’a, 2003). This causes the local and 5 million in 2015.
climate to be substantially warmer than the Studies that have been done reveal
surrounding, a phenomenon known as the that, urbanization effects are noticed even for
Urban Heat Island (UHI) (figure 1). settings of populations as small as 10,000
Precipitation and humidity patterns are
(Karl et al., 1988) and the magnitude of the urban
also greatly modified depending on the
heat island effect is linearly correlated with the
area.
logarithms of the population (Oke, 1973).
An understanding of the urban According to Shepherd (2005), urban
climate is therefore important in the areas with rapid land cover and land use changes
planning and design of urban settle- suffer significantly from human activities.
ments for the comfort of the urban Understanding the human impacts on nature
dwellers. especially vegetation cover is a central
The overall objective of this component of global change studies and
study was to investigate the impacts of simulation of urban environments in a climate
model framework is a practical research
urbanization on the climate of Nairobi approach.
city in Kenya. The specific objectives There are no adequate recent studies that
were to examine the effects of have been done to observe the effect of Nairobi’s
Figure 1: Urban heat island profile

DECEMBER 2010 ONGOMA ET AL 51
urbanization and its impact on the local Addition city heat is given off by vehicle and fac-
climate and further proposed adaptation to tories, as well as by industrial and domestic heat-
the likely climate change-related dangers ing and cooling units. This effect causes the city to
over the next few decades to fit well within become (1 to 6oC) warmer than surrounding
a local development agenda and thus the landscapes. The impacts also include reducing
justification of this research. There is there- soil moisture.
fore need to study local climatic changes Every settlement is capable of generating
due to urbanization and future expecta- an UHI, regardless of its size (Mills, 2006). Obser-
tions in order to support decision making vations for UHI studies display common charac-
towards better urban planning presently teristics, the urban–rural difference reveal itself
and for the future. as a pool of warm air with largest values closest
Kenya is situated on the Eastern to the urban centre. At the urban edges, tempera-
coast of Africa, and lies astride the equator. ture changes are rapid and thereafter increase
It lies between latitudes 5.5° north and 5° more slowly. However in the vicinity of green
south and between longitudes 34° and 42° parks lower temperatures are observed. The
east. Nairobi city; Kenya’s capital city is lo- strength of the UHI is referred to by the maximum
0 0
cated at 1 17’S 36 49'E and has a land urban – rural difference recorded. Its magnitude
2 is greatest at night, under clear skies and with
area of 684 km (260 square miles).
little wind. Under such conditions, surface cooling
The urban heat island has become a
is associated with radiation exchange. While ex-
growing concern; for more than 100 years.
posed rural sites cool rapidly after sunset, urban
In rural areas, a large part of the incoming
sites cool more slowly .The difference between
solar energy is used to evaporate water
urban and rural sites grows with time after sun-
from vegetation and soil. In cities, where
set and reaches a maximum difference after about
less vegetation and exposed soil exists, the
4 hours. The maximum UHI value recorded is usu-
majority of the sun’s energy is absorbed by
ally found in the centre of the settlement is gener-
urban structures and asphalt. Hence, during
ally larger for bigger settlements (Karl, et al.,
warm daylight hours, less evaporative cool-
1988).
ing in cities allows surface temperatures to
Figure 2: The area of study

Source: www.unep.org/.../Images/nairobi_map.jpg
Studies on urban-rural climatic dif- is drier than surrounding areas in the afternoon
ferences have long been carried out since throughout the year .Similarly, in the study of Ro-
Howard (1833) , who reported that night baa (2003) in Cairo, it was found that on the basis
0
was 3.7 c warmer and day was 0.34 c
0 of relative humidity, urban atmosphere is drier
cooler in the city than the country in Lon- than its surrounding areas throughout the year.
don in early 1800s (1820). In recent stud- Fortuniak et al. (2006) analyzed data from two
ies, Karl, et al. (1988) stated that in the US automatic stations in Poland (one rural and one
annual mean temperature at stations in urban) for the period 1997-2007, he found out
populated area of 10 000 people or more that relative humidity is lower in town, some-
0 times by more than 40%.
were 0.1 c warmer than the nearby sta-
tions located in rural areas with a popula- 2. DATA AND METHODOLOGY
tion of less than 2000. 2.1 Data
Rainfall enhancement in urban ar- The data used in the study comprised of
eas has been attributed to higher condensa- two categories:
tion nuclei concentrations over the urban (i) The climate data which comprised
areas, the thermal and frictional effects of of monthly rainfall, humidity and tem-
the area (Changnon, 1996; Vogel and Huff, perature elements of Nairobi City for
1978).Rainfall may be enhanced by air pol- about forty years to date. The meteoro-
lution that increases the number of conden- logical stations are Muguga Forest Station,
sation nuclei through the atmospheric addi- Moi Air Base, Muguga KARI, Dagoretti
tion of smoke and dust particles. Also, the Corner, Kabete Agrometeorological Sta-
additional generation of heat within the tion, and Nairobi Wilson Airport.
city increases the number of convection (ii) Urbanization indicator: Nairobi
currents over that surface, the convection is City population that was obtained from
key in the development of thunderstorms. the Kenya National Bureau of Statistics for
Urbanization is strongly correlated about twenty years (GoK, 2007).
with increment in greenhouse gases emis-
sion, according to IPCC (2008), theoretical 2.2 Methodology
and climate model studies suggest that in The Single mass curve technique was used
climate that is warming due to increasing to test data homogeneity while Graphical Meth-
greenhouse gases; a greater increase is ex- ods were used in trend detection.
pected in extreme precipitation as com- Correlation analysis was used to deter-
pared to the mean. Hence, anthropogenic mine the relation of urbanization indicator;
influence may be easier to detect in ex- (population) to climatological parameters.
treme precipitation than in the mean. This
is because extreme precipitation is con- 3. RESULTS AND DISCUSSION
trolled by the availability of water vapour,
while mean precipitation is controlled by 3.1 Estimation of Missing Data
the ability of the atmosphere to radiate Spatial correlation and Arithmetic Mean
long –wave energy to space, and the latter methods were used to estimate missing data.
is restricted by increasing greenhouse Rainfall data of almost all the stations except Da-
gases. goretti Corner for the year 1998 that was esti-
Urban effect has been found to be mated.
associated with stimulation of rain output
in ongoing storms as opposed to an in- 3.2 Urbanization indicators
crease in frequency of storms (Huff and Vo-
gel, 1978). 3.2.1 Population
Generally, rural areas are more hu- Nairobi’s population is growing with time
mid than the urban ones almost throughout (figure 3).
the year with the exception of the wettest There are wide varieties of standards of
period of the year. Unkasevic et al. (2001) living in Nairobi. Most wealthy Kenyans live in
compared the urban –rural/sub urban wa- Nairobi City but the majority of Nairobi
ter vapour pressure and relative humidity population is poor. Half of the population has
in Belgrade and found out that urban area
been estimated to live in slums which cover goes hand in hand with increase in the increase in
just 5% of the city area. The growth of number of industries, increase in household, in-
these slums is as a result of urbanization crease in fuel consumption by the transport sec-
following increase in rural to urban migra- tor.
tion in search of employment and educa- Increment in fuel consumption in Nairobi
tion. The growth of slums is also due to this can be attributed to domestic and industrial
poor town planning. fuel consumption as well as increase in the num-
ber of cars. Increase in fuel consumption is
3.2.2 Energy Sales to Nairobi strongly correlated with an increment in Green
The consumption of energy in Nai- House Gases emission causing warming effect to
robi has generally been increasing with the city. Greater increase can also lead to increase
time; 1996 to 2005 (figure 4). There was in precipitation.
however a significant decline in energy The mean of the first data set was 24.33
while the second set had 24.47, i.e. σ II > σ I .
2 2
sales from 1999 to the year 2001 with the
lowest in the latter; 2001. This clearly indicates a positive trend (i.e., an in-
Energy consumption increase is a crease in the mean maximum temperature with
good indicator of urbanization in the sense time). The urban air temperature is therefore
that energy drives all forms of livelihood in gradually rising in Nairobi city as is the case in
which economy is included. Urbanization other cities of the world. One of the possible
Figure 3: Time series of population for Nairobi city
Figure 4: Time series of fuel sales in Nairobi

Figure 5 : Maximum temperature trend for Nairobi
Figure 6 : Minimum temperature trend for Nairobi
Figure 7 : Rainfall Trend for Nairobi

causes may be the drastic reduction in the 3.4 Correlation Analysis

greenery area in cities, implying that land The correlation coefficient was found to
use planning becomes critical in determin- be 0.7587 between population and minimum
ing the environment quality. temperature. A coefficient of 0.3578 was realized
Since Nairobi does not have a for the relation of maximum temperatures and
large evenly distributed industrial areas population. The coefficient of correlation between
that could emit excessive heat and smoke population and humidity was 0.0282 and 0.0879
except in a small location, the main reason for rainfall.
for climate change especially in tempera- In all cases, correlation was positive. Ur-
ture may be modified surfaces and atmos- banization was found to be highly correlated with
pheric features in urban areas. Urban en- minimum temperature and least correlated with
vironments mainly consist of impervious rainfall. One parameter (minimum temperature)
surfaces which have lower albedo than can therefore be used to estimate or predict ur-
rural surfaces. banisation given the high correlation coefficient
of 0.7587. Rainfall on other hand was found to be
Trend analysis of minimum Tempera- not strongly correlated to population and can
ture therefore not be used as an index of urbanisation.
The graphical method clearly sug-
gests an increasing trend of minimum 3.5 Regression Analysis
temperature (figure 6) further supported
the view of a warming urban environ- Minimum Temperature and Population
ment. From the plot of population and minimum
temperature (figure 9), it was evident that there
Trend analysis of Rainfall was a direct positive relationship between the
Rainfall the trend is observed to two.
be increasing (figure 7) The value of coefficient of determination
This can be attributed to a collec- is 0.575, it was only about 58% of the variability
tion of aerosols with a typical size be- in minimum temperature that can be explained
tween 0.01 and 10 micrometer that reside by population growth. The remaining percentage;
in the atmosphere for at least several of about 42% of the variability in minimum tem-
hours, affecting the climate directly perature can not be explained by human popula-
through scattering and absorbing radiation.
tion and through acting as cloud conden-
sation nuclei. 3.6 More urban and Less Urban Compari
The increase of heat in Nairobi city son Rainfall
leads to increase in the rate of convection Comparing the trends (figure 10), Moi Air
over the city. Increase in convection leads Base has a positive slope of 0.132 while Muguga
to increment in rainfall frequency as well Forest Station has a weak negative slope of 0.108.
amounts. The rainfall trend in the more urban setting; Moi
Air Base, increases at a low rate compared to the
Humidity Trend Analysis station in the rural setting that shows a slightly
A plot of humidity against the re- decreasing trend.
spective time in years indicates a decreas-
ing of humidity trend (figure 8). Minimum temperature
It is reduction of humidity with Owing to the incompleteness of Muguga
time that shows the city is becoming drier Forest Station and Moi Air Base minimum tem-
as an effect of urbanization. Due to urbani- perature data, NWA and KAS stations were used
zation, the extent of water seepage into as a representation of more urban and more rural
the earth surface is generally low thus re- setting respectiv
ducing the amount of water available for The rate of change of minimum tempera-
evaporation. Similarly, the amount vegeta- ture change over a station in a more urban set-
tion cover is low thus leaving the earth ting: NWA is more than the station in a rural set-
surface exposed to intense heating and ting Kabete Agrometeorological Station (figure
Figure 8 : Humidity Trend for Nairobi
Figure 9 : Nairobi population-minimum temperature regression
Figure 10 : Rainfall trends of Moi Air Base and Muguga Forest Station
Figure 11 : Nairobi Wilson Airport (NWA) and Kabete Agrometeorological Station (KAS) Minimum
Temperature trend
Figure 12 : Nairobi Wilson Airport (NWA) and Kabete Agrometeorological Station Maximum Tem-
perature trend
Figure 13 : Kabete Agrometeorological Station (KAS) and Nairobi Wilson Airport (NWA) Humidity
trend
Humidity
of population growth on minimum tem- Humidity in the more rural settingKabete
perature in less urban setting in compari- Agrometeorological Station (KAS) is increasing
son to in a more urban setting. while in more urban setting; Nairobi Wilson Air-
port (NWA), it is decreasing with time (figure 13).
Maximum Temperature It therefore affirms that the city is becom-
The rate of maximum temperature ing drier in comparison to its environs.
increase at Nairobi Wilson Airport which is
more urban is higher than at Kabete 4. CONCLUSIONS
Agrometeorological Station (figure 12). The main purpose of this study was to
The city is generally getting determine the urbanization effect on the climate
warmer than the more rural locations. This elements over Nairobi City using historical tem-
comes down to a difference between the perature, relative humidity and rainfall data.
energy gains and losses of each region. The findings of this study have estab-
There are a number of factors that lished that Nairobi City is growing at a high rate
contribute to the relative warmth of cities: of about 6.9%. Population growth is directly pro-
During the day in rural areas, the solar portional to energy use and an increased of fossil
energy absorbed near the ground fuel energy implies an increase in emission of
evaporates water from the vegetation and heat from domestic use and industries, green
soil. Thus, while there is a net solar energy house gases and dust particles into the environ-
gain, this is compensated to some degree ment. All these have the effect of increasing air
by evaporative cooling. In cities, where temperature.
there is less vegetation, the buildings, The trend test results show that there has
streets and sidewalks absorb the majority been a significant warming trend in both the
of solar energy input. Also, the city has less mean maximum and minimum temperature in
water and higher runoff rates because the the urban setting. This was found to be more sig-
pavements are largely nonporous (except nificant in the surface minimum temperatures as
where there are potholes). Thus, evapora- compared to maximum temperatures. In contrast,
tive cooling is less which contributes to the the maximum temperature at the less urban set-
higher air temperatures. ting does not show any significant trend. This
Waste heat from city buildings, cars therefore ascertains that there is an urban in-
and trains is another factor contributing to duced modification (warming) in the surface tem-
the warm cities. Heat generated by these perature.
objects eventually makes its way into the Trend test for rainfall showed an increase
atmosphere. This heat contribution can be in rainfall amounts, thus, there is modification of
as much as one-third of that received from rainfall by urbanization at least within the city.
solar energy. The thermal properties of Humidity was tested and found to be in-
buildings also add heat to the air by con- significantly decreasing with time over the city. A
duction. Tar, asphalt, brick and concrete further comparison between the more urban and
are better conductors of heat than the vege- the less urban station showed that there is a de-
tation of the rural area. crease in humidity levels in the more urban set-
The canyon structure that tall buildings ting than in the less urban setting. This implies
create enhances urban warming. During urban settings are becoming drier as result of ur-
the day, solar energy is trapped by multiple banization.
reflections off the buildings while the infra- The environmental implications of such
red heat losses are reduced by absorption. urban surface temperature modification are very
However, urban heat island effects can also sensitive to on human comfort and therefore the
be reduced by weather phenomena al- findings in this study should be incorporated the
though it is not applicable in the study. The current and future planning programs of the city.
temperature difference between the city
and surrounding areas is also a function of 5. RECOMMENDATIONS
the synoptic scale winds. Strong winds re- Green areas in cities have been consid-
duce the temperature contrast by mixing ered as potential measure in mitigating the urban
together the city and rural air. heat island effect (Wong and Yu, 2005). Urban
forest areas which serve as urban lungs Secretariat, Geneva, 210 pp.
should be enlarged and be well maintained. Changnon, S. A., 1996a: Defining the flood: A chro-
Related professionals whose work fields nology of key issues. The Great Flood of
are planning and designing, e.g. landscape 1993, S. Changnon, Ed., Westview Press,
architects and urban planners, should seek 3–28.
new ways of providing healthy environ- Fortunaik K., K. Klysik and J. Wibig, 2006.Urban-
ments for people considering the basic Rural Contrasts of Meteorological pa-
principles of their occupations, e.g. leaving rameters in Lodz. Theoretical Application
enough vegetated spaces. Another ap- in Climatology, 84, 103-116.
proach for the construction of thermally Fortuniak K., K. Klysik and J. Wibig, 2006. Urban-
comfortable environments may be the clus- rural contrasts of meteorological pa-
tering of newly developing cities by placing rameters in Lodz. Theor. Appli. Clim. 84,
large green areas between them and by re- 108 – 116
ducing the size of population and surface GOK 1999, National population and housing cen-
areas of the settlements. Since the number sus. Vol 1. Ministry of Planning and N a -
of such studies in developing countries like tional Development Central Bureau of Sta-
Kenya is insufficient, their quality and tistics.
quantity should be increased in order to Hardoy, Jorge E. and Satterthwaite, David (1989).
provide local people with more liveable Squatter Citizen; Life in the Urban Third
urban environments. World, Earthscan Publications, London.
Decentralization of resources to Houghton J.T and Callander B.A; IPCC, 1992: Cli-
minimise rural – urban migration, reduced mate Change 1992: The supplementary
urban population, reduces pressure on ur- report to the IPCC scientific assessment.
ban resources. This reduces the rate of de- Houghton J.T, Callander B.A and Maskell K.; IPCC,
pletion of natural resources such as clear- 1994: Climate Change 1994: Radiative
ing of vegetation to create room for settle- forcing of Climate Change and an Evalua-
ment and business centres. Reduction in tion of the IPCC Emission scenarios.
population goes a long way in reducing Howard, L., 1833; Climate of London De-
amount of waste disposal. duced from meteorological observations,
Intensification of the network of 3rd ed; in 3 Vols. Harvey and Darton,
meteorological stations so as to enhance London.
their coverage of the city and other urban Huff FA and Changnon SA Jr (1973) Precipitation
centres. Availability of sufficient climate modification by major urban areas. Bull
data from the stations will facilitate more Amer. Meteor Soc 54:1220-1232
accurate research in future. Huff, F. A., and J. L. Vogel, 1978: Urban, topog-
raphic, and diurnal effects on rainfall in
REFERENCES the St. Louis region. J. Appl. Meteor.,
Aesawy, A.M and Hasanean, H.M., 1998. An- 17, 565 -577
nual and Seasonal Climatic Analysis IPCC (Intergovernmental Panel on Climate
of Surface Air Temperature Change), 2007a:Climate Change 2007:the
Variations at Six Southern Mediter- Physical Science Basis .Contribution of
ranean Stations. Theoretical a n d Working Group I to the Fourth Assessment
Applied Climatology, 61, 55-68. Report of the IPCC, S. Solomon D. Qin, M.
Awuor, Cynthia B., Victor A. Orindi and An- Manning Z. Chen, M. Marquis, K.B.
drew Adwerah (2008), Climate Averyt, M. Tignor and H.L. Miller,Eds.
Change and Coastal Cities: The Cambridge University Press, Cambridge
Case of Mombasa, Kenya, Environ- 996 pp.
ment and Urbanization, Vol. 2 0 , Karl T.R., Diaz, H.F and Kukla, G.,
No. 1. 1988.Urbanization: Its detection and Ef-
Bates, B.C., Z.W. Kundzewicz, S. Wu and J.P. fects in the United States Climate re-
Palutikof, Eds., 2008: Climate cord. Journal of Climate, 1, 1099- 1023.
Change and Water. Technical Karl, T. R., H. F. Diaz, and G. Kukla, 1988: Urbani-
Paper of the Intergovernmental zation: Its detection and effect in the
United States Climate Record, J. Cli- the Impact of Urbanization on the Mi-
mate, 1, 1099-1123. croclimate over Nairobi Area, PhD Thesis,
Mills, G. 2006: Progress toward sustainable University of Nairobi.
settlements: a role for urban clima- Opijah F. J., and Mukabana, J., 2004. On the Influ-
tology. Theoretical and Applied Cli- ence of Urbanization on the Water
matology 84, 69–76, doi: 10.1007/ Budget in Nairobi City: A Numerical Study.
s00704–005–0145–0. GeoJournal, 61, 121-129.
Mitchell, J.M., Dzerdzeevskii, B., Flohn, H., Robaa, S. M, 2003. Urban - Suburban/rural differ-
Hommery, W. L., 1996. Climatic ences over Greater Cairo, Egypt. At-
Change WMO TN-79, WMO No. mosfera 16, 157-171.
195. TP-100, WMO, Geneva, 79 pp. Shepherd JM (2005) A review of current investi-
Ng’ang’a, J. K., 2003. Use of Weather and Cli- gations of urban-induced rainfall and rec-
mate Information for Mitigation of ommendations for the future. Earth In-
Air Pollution Related Disasters teractions 9: paper 12, 27 pp.
(A Case of Nairobi). Proceedings of Unkašević M., O. Jovanović and T. Popović,
the Sixth Kenya M e t e o - 2001. Urban-suburban/rural vapour pres-
rologoical Society (KMS), KMS-KMD, sure and relative humidity differ-
pp 86-92. ences at fixed hours over the area of Bel-
Oke, T.R., 1973. City Size and the Urban Heat
grade city. Theor. Appl. Climatol. 68,
Island .Atmospheric Environment 7:
67-73.
796- 779.
Oke, T.R. 1979. Review of Urban Climatol-
Vogel, J. L., and F. A., Huff, 1978: Relations
ogy 1973-76.TN-134, WMO No. 383, between the St. Louis urban precipitation
WMO, Geneva. anomaly and synoptic weather factors. J.
Okoola, R. E., 1980 .The Nairobi Heat Island. Appl. Meteor., 17, 247 - 251
Kenya Journal of Science and Tech- Wong N. H. and C. Yu, 2005. Study of green
nology. Series A, Vol 1, No. 2, 53-65. areas and urban heat island in a tropical
Opijah, F. J., 2000: Numerical Simulation of city. Habitat Int. 29, 547-558.
DECEMBER 2010 Information for Authors 61
…………………………...continued from page 2
The appropriate form for the transfer of the dress and e-mail. These items should ap-
copyright to KMS can be obtained from the pear on the first page by themselves, with
Editor's office or from the Society. The writ- the abstract beginning on page 2. The date
ten, signed transfer of the copyright is re- of receipt of the manuscript will be sup-
quired in order for KMS to have valid rights to plied by the Editors.
continue its wide dissemination of research 2). Abstract. A concise (250 words) abstract
results and other scientific information. Edito- is required at the beginning of each article
rial action on a manuscript that is not accom- and, at the discretion of the chief Editors,
panied by the completed copyright transfer at the beginning of appropriate shorter
form, signed by all authors, will be delayed un- contributions. Authors should summarize
til the form is received. Peer review can begin their conclusions and methods in the ab-
after receipt of faxed or scanned versions, but stract. First person construction should
the originals must be provided as soon as pos- not be used in the abstract, and references
sible and before accepted manuscripts can should be omitted because they are not
enter the production process. All authors need available per se to abstracting services.
to sign a copyright transfer form for the manu-
script that is submitted, but it is acceptable for 3). Text. The text should be divided into sec-
each author to provide a separate form with tions, each with a separate heading and
his or her signature rather than requiring a numbered consecutively. The sec-
single form signed by all, which greatly eases tion/subsection headings should be typed
the logistical problems associated with col- on a separate line [e.g., 1. Introduction, a.
laborations among authors of different institu- Data, 1) RADIOSONDE, and (i) Experiment
tions. 1].
3. Manuscript preparation 4).. Acknowledgments. Omit the word

Each manuscript must be complete and final "number" from grant or contract ac-
when submitted. Article length should be less knowledgments.
than 7500 words (including appendixes but
not references or figure captions) or about 26 5)..Appendix. Lengthy, mathematical analy-
double-spaced typed pages. KMS authors are ses whose details are subordinate to the
encouraged to submit their manuscripts elec- main theme of the paper should normally
tronically (see section 5), and in many cases appear in an appendix. Each appendix
KMS can use files prepared in Microsoft Word, should have a title.
and other formats directly in the typesetting
process. If not submitting electronically, four 6)..References. References should be ar-
copies of the text and photocopies of large ranged alphabetically without numbering.
drawings are required. All copy (including ta- The text citation should consist of the au-
bles, references, and list of figure captions) thor's name and year of publication, [e.g.,
must be double spaced, on one side only, and “according to Rossby (1945),” or “as
have wide margins, and all pages must be shown by an earlier study (Rossby
numbered consecutively. The font used should 1945)”]. When there are two or more pa-
be no smaller than 12 point and the line spac- pers by the same author in the same year,
ing should be double spacing. the distinguishing suffix (a,b, etc.) should
be added. More information on preparing
a. Components of a manuscript and arranging references is provided in
Each manuscript should include the following section 3d of this document.
components, which should be presented in the
order shown. 7).Figure captions. Each figure must be pro-
vided with an adequate caption; all cap-
1). Title, name and affiliation of each author, tions should be listed together, double
dateline, any current or additional af- spaced, for typesetting. Authors must in-
filiations, and corresponding author ad- clude single-spaced captions directly on
62 Information for Authors Volume 4
cally should be submitted as high-quality hard

the figures used for the reviewer cop- copies that can be electronically scanned by the
ies. Authors are encouraged to remove printer for optimal reproduction.
captions from the final print quality The print-quality original electronic figure
original vesions of the hard copy fig- files must be in EPAuthors should attempt to visu-
ures . alize mathematical expressions as they will
appear in print. Avoid built-up fractions and other
8). Illustrations and tables.Each figure complicated equation structures in text. Instead,
and table must be cited specifically in the have complicated expressions appear as display
text. The figure number will be typeset equations, that is, as equations centered on their
and should not be part of the illustration. own line. Display equations are usually numbered
Authors should label figure panels consecutively to facilitate their citation in text,
with lowercase lettering, preferably in the which is done by using the equation number in
upper-left corner within the figure panel. parentheses set flush right.
All tables should have a double-spaced Because of KMS typesetting requirements,
caption, and table text and headers also authors who use Microsoft Word to prepare their
should be double spaced. manuscripts are asked to use MathType version 5
to prepare their display equations, rather than
b. Figures making entries from the keyboard, and to avoid
All figures printed in the journal are the use of MathType entirely in running text, using
placed on the composed page as an elec- the keyboard exclusively except to create over-
tronic image file. The highest quality of barred variables or variables with stacked su-
reproduction is possible if authors supply per/subscripts that cannot be easily created from
electronic files of any images that were the keyboard. Following this practice will greatly
originally created in an electronic form. reduce production time for mathematics-heavy
Instructions for transmitting both the papers.
original and review (captioned) versions Authors can facilitate the correct typesetting of
of figure files are provided on the KMS their equations by using the correct typeface for
manuscript upload page at the KMS Edi- variables. Scalar variables are set as italic (with S,
tors e-mail. If the original files change in TIF or MS word format.
number or content during the review Halftone reproductions of photographs and dense
process and have already been uploaded, stippling patterns do not scan well and should be
they can be uploaded again. Authors must avoided. Authors should strive to submit their fig-
still send at least one set of high-quality ures at the size they will appear (but do not sub-
hard copy versions of the original figures mit photocopy reductions in place of larger,
to KMS Offices, even if some or all are higher-quality originals), with the understanding
available in electronic form, because they that the technical editors will reduce the figures to
aid in evaluating the quality of the typeset the maximum extent possible while still preserv-
figures and serve as a backup for scanning ing legibility. Note any unusual layout require-
if the electronic files have problems. These ments. Multiple-panel figures should be combined
hard copies must contain only one num- into a single electronic file and/or hard copy when
bered figure per page, but that figure can possible in order to avoid additional handling
contain multiple panels. They no longer charges. The printing of color figures in the jour-
have to be captionless, although it is still nal represents a significant additional expense
preferred. Authors will be reminded by that is passed on to authors in the form of in-
the field editor during the review process creased author charges (see section 6).
to send figures to KMS. If a field editor
needs hard copy versions of the peer- c. Mathematical formulas, units, and time and
review (captioned) figures for any date
reason, the author also will be notified. the exception of multiple character vari-
More information on electronic figure sub- ables, e.g., RH or SST), vectors are set as boldface
mission is available in the KMS Editor’s e- roman (e.g., V), and matrices and tensors are set
mail. Illustrations and photographs that as boldface sans serif (e.g., A). If the author cannot
were not originally produced electroni- reproduce these typefaces, he or she should
DECEMBER 2010 Information for Authors 63
indicate vectors with a single wavy line under Wallace, J. M., and P. V. Hobbs, 1977: Atmos-
the character and matrice(in print versions) or pheric Science: An Introductory Survey.
provide a list of variable types (if submitting Academic Press, 350 pp.
electronically).
No other mathematical symbols should 2) FOR A BOOK
be underlined. Subscripts or superscripts are Reference must consist of last name and ini-
usually set as lightface, even when applied to tials of author(s), year of publication of
vectors or matrices, and are set italic unless the book, title of book (italicized or underlined),
subscript or superscript is an acronym or abbre- publisher’s name, and total pages. For exam-
viation (e.g., T L, TLCL, Vg, Vobs). ple:
Units should be SI with the exception of
a few approved non-SI units of wide meteoro- Wallace, J. M., and P. V. Hobbs, 1977: Atmos-
logical or oceanographic usage that are de- pheric Science: An Introductory Survey.
scribed in the AG. Units should be set in roman Academic Press, 350 pp.
font using exponents rather than the solidus (/)
and with a space between each unit in a com- 3) FOR A CHAPTER IN A BOOK
pound set (e.g., m s-1 rather than m/s or ms-1). For a book or monograph that is a collection
Day, month, and year are written in the of papers written by independent authors,
form “26 May 1998” in KMS publications. Do not the reference must be made to the authors
abbreviate the names of months except in figure of a particular
captions or tables. The recommended time zone chapter and consist of last name and initials
annotation system is universal time and is ab- of author(s), year of publication of book, title
breviated UTC. Time, time zone, day, month, and of the chapter, title of book (italicized or un-
year are written in the form “1619 UTC 26 May derlined), name of editor(s), publisher’s
1998.” The use of other time zones is permissi- name, and inclusive pages for the chapter.
ble—for instance, EST, EDT, PST, LST (local For example:
standard time), or LT (local time). Do not use Z Anthes, R. A., 1986: The general question of
or GMT in place of UTC. Astronomical or military predictability.
time (i.e., a 24-hour clock) is required. Mesoscale Meteorology and Forecasting, P.
S. Ray, Ed., Amer. Meteor. Soc., 636–656.
d. References
A complete “Guidelines for Preparing Refer- For a chapter in a book that is part of a
ences” may be obtained from KMS Headquarters monograph series, the format is similar but
or online in the AG. A few of the most common includes the volume and number of the
reference types are shown here. In order for the monograph. For example:
cross-reference linking now possible through
the KMS Journals Online to work properly, refer- Arakawa, A., 1993: Closure assumption in
ences must be complete and properly formatted. the cumulus parameterization problem.
Authors are encouraged to invest the time The Representation of Cumulus Convec-
needed to prepare the references according to tion in Numerical Models, Meteor.
KMS style. Monogr., No. 46, Amer. Meteor. Soc., 1–16.
1) FOR A JOURNAL ARTICLE 4) CONFERENCE PREPRINT OR PRO-

Reference must consist of last name and initials CEEDINGS
of author(s), year of publication of journal, title Reference must consist of last name and ini-
of paper, title of journal (italicized or underlined tials of author(s); year of publication; title of
and abbreviated, volume of journal (boldface), paper; indication of the publication as a pre-
number of issue (only if required for identifica- prints, proceedings, or extended abstracts
tion), and first and last page numbers of the pa- volume; name of conference volume
per. For example: (italicized or underlined); city and state
where conference was held; conference
Charney, J. G., and A. Eliassen, 1964: On the sponsor’s name; and pages of the paper. For
growth of the hurricane depression. J. Atmos. example:
Sci., 21, 68–75.
62 Information for Authors Volume 4
Kalnay, E., and Z. Toth, 1994: Removing growing authors (see section 2), and, 3) if
errors in the analysis cycle. Preprints, 10th submission is by hard copy rather
Conf. on NumericalWeather Prediction, Port- than online, five complete copies of
land, OR, Amer. Meteor. Soc., 212–215. the manuscript and figures with cap-
tions below (three for J. Phys. Ocean-
References should be to peer-reviewed lit- ogr.). In addition, KMS still requires
erature whenever possible. Technical reports, at least one set (two if electronic files
conference proceedings, and other “gray litera- are not available) of original hard
ture” should be referenced only when no other copy figures, preferably without cap-
source of the material is available, and an tions, by the time the paper is ac-
“available at” address should be provided for cepted. A summary of requirements
reports and dissertations. for successful qualification of manu-
scripts is given at the end of this
4. Manuscript submission document. For online submission
All articles in meteorology and its appli- requirements, see the next section.
cations are welcome for submission and authors
are encouraged to study carefully the sections 5. Online manuscript submission
on manuscript preparation. The KMS journal now accepts manuscripts
The manuscript should be submitted to KMS in electronic form through an online submis-
through online submission at the KMS Editor’s sion process. There are constraints that
e-mail rokoola@uonbi.ac.ke or must be met before a manuscript can be
kmsmails@yahoo.com. Note that hard copy submitted online. These constraints and the
manuscripts and related material for all journals submission process itself are discussed in
are now submitted directly to KMS Editor. If a this document. Authors may submit a soft
manuscript meets the KMS submission qualifica- copy as an attachment in ms word.
tions (described in the table below), it will be
turned over to the chief editor’s office to begin The KMS upload software will automatically
peer review. The chief editor and the journal’s create for peer review a PDF file that con-
editorial board will oversee the peer review of tains the double-spaced manuscript,the ta-
the manuscript and will correspond directly bles, and the figures with captions below.
with the author concerning the disposition of The author can view it and approve it for
the submission. submission to the chief editor.
The submission must include the follow-
ing components: 6. Publications charges
1) a cover letter that includes the manu- KMS has not developed a page
script title and full contact information, charge policy. These page charges are de-
including mailing address, phone and fax signed to cover the cost of editorial, compo-
numbers, and e-mail address, for one of sition, and related work needed to prepare
the authors (usually the lead author), as an article for publication. Reproduction of
well as any additional information re- color figures is significantly more expensive
quired for the manuscript,(see section and results in higher publication charges.
2); Payment of publication charges may be ex-
2) the copyright transfer forsigned by all pected by the Society in the near future in
view of the color printing that is necessary.
DECEMBER 2010 A Brief Guide for Authors 61
Additional items that must be in place be-

A BRIEF GUIDE FOR AUTHORS fore a manuscript can be accepted for pub-
____________________________
lication and begin the production process:
Qualification of Manuscripts 1) Signed original copyright forms from
Items that must be in place before a all authors.
submitted manuscript package can begin the
peer-review process: 2) References and citations in proper
Items that must be in place before a sub- KMS format.
mitted manuscript package can begin the peer-
review process: 3) Separate double-spaced figure caption
1) Properly signed copyright forms from all lists provided.
authors (submitted individually or to-
gether). Faxed or scanned signatures are 4) Abstract, references, and figure cap-
acceptable initially. tion list begin on new pages.
2). Electronic, preferably) cover letter stat- 5) Double-spaced abstract, main text, ap-
ing type of submission (e.g., article) and pendixes, references, figure caption
whether it has been considered previ- list, and table captions and body text.
ously by any other journal and contain-
ing contact information of the corre- 6) Tables set one per page, with captions
sponding author. set above the table text.
3).Length of no more than 7500 words 7) One set of press-quality hard copy fig-
(approximately 26 double-spaced pages, ures, if electronic figure files are
counting abstract through appendixes, available in eps or tiff format. If elec-
but not list of figure captions or refer- tronic figure files cannot be provided,
ences). The author is required to request two sets of press-quality hard copy
chief editor approval and provide justifi- figures are required.
cation to obtain an exception to the
length limit. Captions no longer must be removed
from the press-quality hard copy fig-
4).Double-spaced text [three typed lines per ures, but it is recommended in order
inch (2.5 cm) measured from anywhere to reduce processing time and costs if
on the page], including abstract through they must be scanned. Figures must
appendixes. be placed one figure per page, with
individual elements (e.g., panels)
5).Type that is 12-point font or larger. grouped together as much as possible
in order to minimize publishing costs.
6).Figures (with captions below each figure
for the review process) and captioned Notice to authors: The following statement
tables placed at the end of the manu- must be signed by all authors (use multiple
script, rather than embedded in the text. forms if needed) of a manuscript and re-
ceived by the editor to which the manuscript
7).The following elements in the proper or- is submitted. Request for further information
der: title page, abstract, body text, ap- should be addressed to the Editor, Kenya Me-
pendixes (if any), references, figures, teorological society, P. O. 41959, 00100 GPO,
and tables. Some manuscript types, such Nairobi, Kenya. E-mail:rokoola@uonbi.ac.ke
as Comments and Replies, may not re- or kmsmails@yahoo.com.
quire an abstract. Please number all www.kenyametsociety.or.ke
pages sequentially.
66 Agreement to transfer copyright Volume 4
Agreement to transfer copyright

Copyright to the article entitled:…………………………………………………………
……………………………………………………………………………………………
by…………………………………………………………………………………………
assigned and transferred exclusively to the Kenya Meteorological Society (hereinafter re-
ferred to as KMS) effective if and when the article is accepted for publication
in The Journal of the Kenya Meteorological Society
……………………………………………………………………………………..
The authors, however, reserve 1) all proprietary rights other than copyright, such as patent
rights, and 2) the right to use all or part of this article in future lectures, press releases, and
reviews of their own. Certain additional reproduction rights as granted by the Society. All
other uses will be subject to the limitations in the copyright statement of the KMS journal
in which the article is published. To be signed by all authors or if the manuscript is a “work
made for hire”, the employer, who is the “legal author” under the copyright law.
……………………………………………………….
Signature
……………………………………………………….
Print name
………………………………………………………
Title if signed by employer
………………………………………………………….
Date
……………………………………………………….
Signature
……………………………………………………….
Print name
………………………………………………………
Title if signed by employer
………………………………………………………
Date
DECEMBER 2010 The Kenya Meteorological Society 67
The Kenya Meteorological Society

BRIEF HISTORY Membership fees: Membership fees in the vari-
The Kenya Meteorological Society ous categories are minimal and are only meant to
(KMS) was registered in 1987, as a non- indicate commitment of a given member:
profit making professional/scientific soci-
ety, in a bid to promote the understanding ACTIVITIES
of meteorology and its applications, among The Society is set to attain its goals by
other functions, in Kenya and beyond. Dur- facilitating communication of important findings,
ing its short span of existence, the Society which contribute towards the advancement of
has undertaken its duties with dedication. meteorological knowledge to Kenyans and the
The objectives of the Society include global community at large. In this regard it is the
the advancement of meteorology in Kenya. policy of the Society to encourage research in
In this respect, the role of the KMS is, there- Meteorology and other related sciences; organize
fore, to enhance the services of the Kenya local and international Meteorological work-
Meteorological Department (KMD) in the shops, conferences, symposia; introduce public
promotion of meteorology. education; and, sensitize the public on the impor-
tant aspects of Meteorology such as drought, de-
MEMBERSHIP sertification and climate change.
The KMS draws its membership
from meteorologists, environmentalists, Activities in Progress:
agriculturalists and other scientists from The Kenya Meteorological Society (KMS), in con-
related disciplines in Kenya and from re- junction with the Kenya Meteorological Depart-
nowned institutions the world over. ment (KMD), the University of Nairobi (UON),
and the Kenya Agricultural Research Institute
Member: Any Kenyan who holds at least a (KARI), has organized Workshops on Meteoro-
post-secondary education diploma in Meteor- logical Research, Applications and Services. In
ology or related sciences from a recognized addition to dealing with pressing global issues, it
institution of higher learning. is also intended to facilitate sustained interaction
of Meteorologists and users of Meteorological
Associate Member: Any person who is dis- products. This will eventually enable Meteorolo-
tinguished by his/her work in Meteorology or gists to better tailor their products to the needs
a distinguished person whom the Society may of the users.
choose to honor for his/her services to the Among the on-going projects by KMS is
Society. the publication of ‘The KMS Bulletin’, a quar-
terly newsletter, which seeks to inform scien-
Fellow: Any member who has been elected tists and the general public on what is happen-
to the fellowship by the Committee. ing in the field of weather and climate. This
Newsletter, whose aim is the promotion of pub-
Foreign Member: Any non-Kenyan who has lic education, is distributed to schools, other
been elected to the membership of the Society institutions and individuals in Kenya and
by the Committee. abroad. In addition, the Society sponsors lec-
tures by experts in schools and community-
Corporate Member: Any school, college, based groups on weather and other meteoro-
university or department of a school, college logical topics of current interest.
or university, research institution, company, The KMS launched a-half-yearly publi-
national meteorological services or any other cation, Journal of the Kenya Meteorological So-
organization with interest in meteorology. ciety, in September 2007 in a bid to facilitate
dissemination of research findings by scientists
Student Member: Any person enrolled in a within Kenya and beyond.
recognized post-secondary institution of
learning.
ENTRANCE, ANNUAL SUBSCRIPTION AND LIFE MEMBERSHIP FEES:

Registration Fees Life Membership
Ordinary Members Ksh.100.00 KSh 500.00 KSh 4,000.00
Fellows Ksh. 100.00 KSh 500.00
Foreign Members US$. 20.00 US$ 50.00 US$ 300.00

Corporate Member - Local - KSh 500.00 KSh 10,000.00
Corporate Member - Foreign - - US$ 2,500.00
Student Members Ksh. 100.00 KSh 100.00 -
CURRENT OFFICE BEARERS

Chairman: Prof. Nzioka. J. Muthama University of Nairobi
Vice-Chairman: Mr. Joseph N. Kagenyi Kenya Meteorological Department
Secretary: Mr. Samuel. W. Kahuha Kenya Meteorological Department
Vice-Secretary: Mr. Nicholas. W. Maingi Kenya Meteorological Department
Treasurer: Mr. Francis. Nguatah Kenya Meteorological Department
Vice-Treasurer: Ms. Pamela R.N. Kaithuru Kenya Meteorological Department
EXECUTIVE COMMITTEE MEMBERS

NAME INSTITUTION
Dr. Joseph. R. Mukabana Director, Kenya Meteorological Department
Prof. Nzioka John Muthama University of Nairobi
Mr. Joseph Kagenyi Kenya Meteorological Department
Mr. Samuel Kahuha Kenya Meteorological Department
Mr. Nicholas Maingi Kenya Meteorological Department
Mr. Francis Nguatah Kenya Meteorological Department
Ms. Pamela Kaithuru Kenya Meteorological Department
Maj. Stephen Sane National Operation Centre, NOC
Prof. John K. Ng’ang’a University of Nairobi
Prof. Laban O. Ogallo IGAD Climate Prediction and Application Centre, ICPAC
Dr. Raphael E. Okoola University of Nairobi
Ms. Emmah Mwangi IGAD Climate Prediction and Application Centre, ICPAC
Mr. John Rao Nyaoro Ministry of Water and Irrigation, Department of Water Development
Prof. Silvery Oteng’i Masinde Muliro University of Science and Technology, MMUST
Mr. Alexander L. Alusa Climate Consultant
Dr. Sharif I. Ahmed UN-HABITAT
Dr. George. A. Keya Kenya Agricultural Research Institute, KARI
Ms. Alice Bett Kenya Wildlife Service, KWS
Mr. Claudio Achola University of Nairobi
UNESCO Representative Vacant
IOC Representative Vacant
KMA Representative Vacant
ISDR Representative Vacant
ICAO Representative Vacant

Journal of Meteorology and Related Sciences
(ISSN 1995—9834)
Contents
Volume 4, December 2010
ARTICLES
Spatial and Temporal Rainfall characteristics Over Sey-

chelles……………...……V.J. Amelie , J.M. Ininda and R.E Okoola.
3-20
Analyses of Annual Droughts in Kenya Using an Objective Annual
Rainfall Drought Index…………………………………Francis M. Mutua and
Abdel A.F. Zaki.
21-33
Meteorological factors and measles occurrence in Akure, Ondo

State, Nigeria…………...A. Akinbobola and J.Bayo. Omotosho. 35-47
Effects of Urbanization on Climate of Nairobi 49-60

City…………………...……O. Victor, N.J. Muthama, J.K. Ng’ang’a.
View publication stats

Effect of Urbanisation On Nbi Climate

Uploaded by

Copyright:

Available Formats

You might also like

Effect of Urbanisation On Nbi Climate

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Effect of Urbanisation On Nbi Climate

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Effects of Urbanization on Climate of Nairobi City

Article · December 2010

Victor Ongoma Nzioka John Muthama

SEE PROFILE SEE PROFILE

Journal of Environmental and Agricultural Sciences editor.jeas@outlook.com View project

Agroecology View project

The user has requested enhancement of the downloaded file.

A Journal of Kenya Meteorological Society

December 2010 - Volume 4

Dr. C. Mutai R. Ojala P. Ambenje,

Information for Authors

Spatial and Temporal Rainfall characteristics Over Seychelles

J.M. Ininda and R.E Okoola

Department of Meteorology, University of Nairobi, Kenya

(Manuscript received 16 September 2009, in final form 26 June 2010)

1. Introduction (Figure 1). It comprises of 115 islands with a

clude: graphical, polynomial, and statistical

2.2.1.2 Spectral Analysis In order to obtain consistent estimates of ¦

Figure 4: Showing the plot of Eigenvalues v/s Factor number

3.3 Onset and cessation of seasonal rains

3.4 Results from trend and spectral

Accumulative frequency for year 1

800 r/f accu

Accumulative frequency for year 4

1000 r/f accu

Table 2: Loading of the Three Dominant Components on Various Stations

RAINFALL STATIONS FACTOR

Frequency accumulation for year 9

2000 r/f accu

Table 3: Loading of each rainfall station on the Rotated Principal Component.

STATION Loading on the Three rotated PCA

ROCHON 0.694 0.278 0.465

STATION COMPONENT C A S - 0.899

Anse Forbans (Region 1)

Figure 11: Examples of spectral density over Seychelles.

spheric circulation patterns. Meteorol. Okoola, R,E.A ,1996: Space-Time Character

Abdel A.F. Zaki

(Manuscript received 27 July 2010, in final form 01 November 2010)

period with no rain, especially during a moisture conditions respond to precipitation

1.2 Drought Types

Figure 2: The Extreme positions of the ITCZ over the globe

-4 Annual Rainfall (mm) -4 Coefficient of Variation -4

Longitude (Degrees East) Longitude (Degrees East) Longitude (Degrees East)

• drought. The opposite applies 6

Longitude in degrees (North is positive)

ness coefficient of t h e annual 3

co effi cient of variation. This K IT A L E KENYA

episodes of heavy rainfall is a -3 T A N Z A N IA VOI

strong phenomenon in the dry -4 M O M BASA In d ian

parts of the country.

Figure 3 shows the distribution of the mean, Figure 4: Map of(Degrees

Where, 3. Results and Discussion

Annual Drought Indices

Longitude (Degrees East) Longitude (Degrees East) Longitude (Degrees East)

and sy are the mean and standard deviation of

where The notation used in Equations 6 and 7

AIC(Log- AIC(Log- AIC(Log-