UNIVERSITY OF NAIROBI DEPARTMENT OF METEOROLOGY

Transcription

1 UNIVERSITY OF NAIROBI DEPARTMENT OF METEOROLOGY DETERMINATION OF THE ONSET AND CESSATION OF MARCH-MAY (MAM) SEASON OVER THE CENTRAL RIFTVALLEY REGION OF KENYA. JELAGAT VICTORINE I10/1291/2012 SUPERVISORS: PROF. ININDA J.M. MS BOSIRE E. A RESEARCH PROJECT REPORT SUBMITTED IN THE PARTIAL FULFILMENT FOR THE AWARD OF THE DEGREE OF BACHELOR OF SCIENCE IN METEOROLOGY. APRIL 2016 i

2 ii

3 DECLARATION This research project is my original work and has not been submitted for a degree in any other University. Signature.. Date. Jelagat Victorine I10/1291/2012 This research project has been submitted for examination with our approval as university supervisors. Signature Date.. Prof. Ininda J.M. Signature. Date Ms. E. Bosire Department of Meteorology University of Nairobi P.O. BOX Nairobi. iii

4 iv

5 Dedication To Uncle Simon Kailel and his family for their support, encouragement and prayers. Also to all esteemed farmers who wish to improve their work and increased production. v

6 vi

7 Acknowledgement I wish to thank the Almighty God for care, protection, strength and faithfulness throughout the journey of my undergraduate. Secondly, my sincere gratitude goes to my supervisors Prof. Ininda and Ms Bosire for their untiring support, mentorship, and constructive criticism, making of corrections and giving suggestions on what to write on this manuscript thus a successful completion of the project. My gratitude also goes to the Kenya Meteorological Service (KMS) for providing rainfall data used in the study. My utmost gratitude also goes to the Department of Meteorology, staff, subordinate staff, students and friends for their guidance and unending support during my years of study at the University of Nairobi. vii

8 Abstract The area under study has high agricultural potential has it receives a substantial amount of rainfall. Therefore, examining the onset, cessation and the duration of the seasonal rainfall is very important for us to understand the performance of seasonal rainfall in the region. This enhances the crop production which plays a crucial role in the growth of our economy. Daily rainfall of four synoptic stations (Kericho, Kitale, Eldoret and Narok) for the period was considered. Instat software was used to determine the onset and cessation dates for the study. The obtain results were used to determine the mean onset and cessation date. The analysis showed that the onset and cessation occurred on 20 th march-3 rd April and 22 nd may-2 nd August respectively with early onset in the Kericho and late onset at Narok which is located at the southern part of the region. This is linked to the north-south oscillation of the zonal component of the ITCZ. Moreover, cyclones and the air mass for example the Congo air mass and Lake Victoria was other synoptic systems that lead to early onset during the year 2006 and the region experienced late onset in the year viii

9 Table of Contents Acknowledgement... vii Abstract... viii List of Tables... xii CHAPTER INTRODUCTION PROBLEM STATEMENT Main objective JUSTIFICATION OF THE STUDY AREA OF STUDY Geographical features of the study area...3 CHAPTER TWO LITERATURE REVIEW SYSTEMS THAT INFLUENCE THE RAINFALL OVER THE CENTRAL RIFTVALLEY Inter-Tropical Convergence Zone (ITCZ) Tropical Cyclones Monsoons Mesoscale Systems El-Niño Southern Oscillation (ENSO) Indian Ocean dipole Quasi-Biennial oscillation Subtropical anticyclones...11 CHAPTER THREE DATA AND METHODOLOGY Data type, source and quality control Data type and source...12 ix

10 3.1.2 Data Quality control Estimation of missing data METHODOLOGY Determination of onset and cessation dates Classification of onset and cessation into Early, Normal and Late Correlation Analysis Plotting of the analysis fields...16 CHAPTER FOUR RESULTS AND DISCUSSIONS DATA QUALITY CONTROL ANALYSIS Results from homogeneity test Mean Onset and Cessation Dates Time series analysis Temporal pattern of rainfall Classification of occurrence of onset and cessation dates into classes (early, normal and late) Correlation analysis...39 CHAPTER FOUR SUMMARY, CONCLUSION AND RECOMMENDATIONS Summary Conclusion Recommendations...42 REFERENCES...43 x

11 List of Figures Figure 1: Distribution of stations on the study area Figure 2: Cumulative mass curve for Kericho, Figure 3: Cumulative mass curve for Eldoret Met Figure 4: Cumulative mass curve for Kitale, Figure 5: Cumulative mass curve For Narok: Figure 6: Narok onset and cessation dates over mam season, Figure 7: Kitale onset and cessation over mam season, Figure 8: Kericho onset and cessation over mam season, Figure 9: Eldoret met onset and cessation over mam season, Figure 10: Frequency of onset of rainfall Figure 11: Factors influencing the synoptic systems thus early onset (A) sea level pressure (B) geopotential height (C) winds (anomalies) Figure 12: Factors influencing synoptic systems thus late onset: (a) geopotential height (b) sea level pressure (c) winds xi

12 List of Tables Table 1: Location And The Altitudes Of Stations Used In The Study...12 Table 2: Mean Onset, Cessation And Standard Deviation...20 Table 3: Onset And Cessation Dates Over Narok (E-Early, N-Normal, L-Late)...25 Table 4: Onset And Cessation Dates Over Kitale (E-Early, N=Normal, L=Late)...27 Table 5: Onset And Cessation Dates Over Eldoret Met (E=Early, N=Normal, L=Late)...29 Table 6: Onset And Cessation Dates Over Kericho (E-Early, N-Normal, L-Late)...31 Table 7: Correlation Coefficients For The Four Stations: Narok, Kitale,Kericho And Eldoret Met...40 Acronyms ITCZ-Inter Tropical Convergence Zone MAM-March, April, May OND-October, November, December FAO-Food and Agriculture Organization DJF-December, January, February ENSO-Elnino Southern Oscillation IOD-Indian Ocean Dipole xii

13 CHAPTER INTRODUCTION Rainfall is of importance to countries inclusive of Kenya whose economy is dependent on rain fed agriculture. Rainfall has strong bearing on plant growth. Lack of it or reduction during the critical stage may result to poor performance or none at all. The information on the spatial and temporal distribution of rainfall is essential to farming. In particular the information on onset and cessation is crucial to crop management. One of the most agricultural productive region in Kenya is Central Rift valley which produces tea, coffee, maize, beans, potatoes, tomatoes and wheat among many other crops. High rainfall variability often leads to reduce agricultural productivity leads to food insecurity. In much of Equatorial Eastern Africa, with the exception of western highlands areas, most precipitation fall within two rainy season that is the long rains MAM and the short rains OND. Each season coincides with the passage of the intertropical convergence zone when it migrates from the southern to the northern hemisphere and vice versa. The long rains generally provide more rainfall than the short rains and have a lower inter-annual variability (Camberlin et al.2009). The long rains (MAM) are crucial for most region of Kenya especially in maize growing areas that require a longer rainy period. The study determines the onset and cessation of long rains (MAM) in the Central Rift valley region of Kenya. 1.1PROBLEM STATEMENT Rainfall characteristics in terms of length of growing season have always been uncertain due to high variability of onset and cessation of the rains. Central Rift valley region of Kenya relies on a rain fed agriculture which can be limited by the variability of onset and cessation as it may lead to lack or little soil moisture. In a community like this which its economy is dependent on a rain fed agriculture, then knowledge on the onset and cessation dates are very important for planning purposes. The study is to address the frequently asked question about the onset and cessation of 1

14 rains with respect to growing period for the choice of suitable crop varieties and adequacy of water for livestock consumption. 1.2 OBJECTIVES 1.2.1Main objective To determine the onset and cessation of seasonal rains (long) over the Central Rift Valley region. Specific objectives include: (i) (ii) (iii) Determination of mean onset and cessation dates for March-May season in the study area Investigate the synoptic systems that are associated with early and late onset over the region. Determine the relationship between onset, cessation and duration on seasonal performance. 1.3 JUSTIFICATION OF THE STUDY The amount of rainfall and its distribution affects the type of farming activities carried out in an area. That is, yields may suffer significantly with either late onset or early cessation of the growing season, as well as with high frequency of damaging dry spells within the growing period. For farmers to better their produce, then knowledge on the onset and cessation is vital for them to know what to plant and when to plant. The study on the onset and cessation is also important to other stakeholders for planning purposes. 1.4 AREA OF STUDY Rift valley region of Kenya is enclosed by latitude N and longitude E with a square km. of It stretches from the county of Turkana and stretches along the Great Rift Valley to the south of Kajiado County. 2

15 FIGURE 1: DISTRIBUTION OF STATIONS ON THE STUDY AREA. 1.5 Geographical features of the study area The study region has vast geographical features consisting of mountains, escarpments, forests, and lakes, among other features. The water bodies that are found within the area of study include L.Naivasha, L.Nakuru, L.Bogoria, and Baringo among other inland lakes. The presence of the Rift Valley System that runs from the north to the south across the country and the Turkana Low-level 3

16 jet provide an environment where both local and global systems interact to create a climate that is quite variable in space and time. 1.6 Climatology of rainfall over the study area The climate of Rift valley region is mild, with temperatures usually below 28 C (82 F) and a trimodal rainfall regime (three peak season in a year).most rain is experienced during the March May (MAM) long rains, short rains in October December (OND) and the middle rains July- August. The rainfall are influenced by the north-south movement of the ITCZ which lags behind the overhead sun by 3-4 weeks Okoola (1999), Asnani (1993). Okoola and Ambenje (2003) have noted that ITCZ was the major synoptic system controlling the MAM and OND seasonal rainfall over the western part of Kenya which is inclusive of west highlands. The ITCZ over the Congo Basin was mainly meridional but had incursions eastward up to the Western part of Kenya. Ininda and Kihungu (1998) and Okoola (1999) among others have shown that most rainfall over western part of Kenya are associated with the mid-tropospheric westerlies over the Equatorial East Africa. The westerlies are associated with advection of moist Congo basin air into our country. The air mass upon reaching our region will encounter orographic lifting leading to convection activities. 4

17 CHAPTER TWO 2.1 LITERATURE REVIEW This part consists of various researchers who have carried out the study from different point of view on the definition of onset and cessation of rainfall. Various methodology were used to analyse the available rainfall data for specified areas to come with quality and useful information for end users. Some of the theories or literature reviews are as follows: Oshodi (1971) also grouped rainfall amounts into pentads, used 12-17mm of rainfall as a threshold value to establish the normal dates of onset and cessation of 14 rainy season while a higher threshold of 25.4mm or more was used to give corresponding onset and cessation dates of heavy rains. In Northern Nigeria, Kowal and Knabe (1972) grouped rainfall amounts into decades found the correlation coefficient as r = 0.88 and r2 =0.77 which imply that 77% of the rainfall received in a station could be caused by latitude. That is, the farther one is from the coast the latter the onset the earlier the cessation dates of rainy season. In East Africa, Alusa (1978) and Mushi (1974) basing on the work of Hesanmi (1972), they identified the onset of rains in East Africa stations as the first maximum curvature on cumulative rainfall plot which was the first pentad whose rainfall amount exceed 1/73 of the mean annual rainfall. FAO (1978) defined the start of the growing season as the date when precipitation exceeds half of the potential evapotranspiration Stern at el. (1982) proposed a daily water budget scheme, involving a rainfall as input for a realistic evaluation of cessation. They also proposed a simpler alternative which uses the first occurrence of a long dry spell of at least 15 days after mid-september. In northern Australia, Nicholls (1984) defined onset as the date on which 15% of the mean occurred at a given station. 5

18 Sivakumar (1988) defined the onset date as the date after 1 May when rainfall accumulated over three consecutive days was at least 20 mm and when no dry spell within the next 30 days exceeded 7 days. In the work of Omotosho (1990) on the onset of thunderstorms and precipitation over Kano in northern Nigeria, he showed that the onset of agriculturally sufficient and reliable rainfall is related to the vertical wind shears between surface to hpa. Pentads have been used by Joliffe et al. (1994) to determine onset of rains in the tropics. They defined wet pentad as that one with at least 25mm of rainfall with at least 3 days greater or equal to 0.85mm of rain to determine the start of season. Omotosho et al. (2000), it defines the onset and cessation based on the crop water requirement (CWR) where the date of rainfall onset is defined as the first day of any week having a cumulative rainfall of at least 20mm, one of which must be 10mm or more, followed by two other weeks, each with at least 50% of the weekly CWR. The first three to four weeks being critical for seed germination and crops establishment need some initial heavy rainfall (10mm) to moisten and soften the soil sufficiently for seed germination and ensure crop survival in the following twenty days. Marengo et al. (2001) defined regional scale onset dates for different of the Amazon Basin. The dates were computed by averaging daily rainfall data from many stations and constructing 5 days averages. Onset was the pentad in which rainfall exceeded a given threshold. In Eastern Africa, P Camberlin and R.E Okoola (2003) use the Principle Component Analysis (PCA) to depict the onset and cessation of the Long rains in Kenya and Tanzania. They did the analysis for all stations time series over a period longer than the rainy season, including parts of antecedent and posterior dry seasons. In Eastern Africa, P.Camberlin (2009) studied the components of rainy season on onset and cessation, rainfall frequency and intensity and found out that the onset and cessation dates for long rains are independent of the number of rainy days and daily rainfall intensity during long rainfall season. 6

19 2.2 SYSTEMS THAT INFLUENCE THE RAINFALL OVER THE CENTRAL RIFTVALLEY This section describes the synoptic systems that influence the rainfall over the central Rift valley region Inter-Tropical Convergence Zone (ITCZ) It is a narrow band of low pressure where air masses moving equator-wards from the southern and northern hemispheres converge. This zone is characterized by high humidity, heavy precipitation, deep cloudiness, slow winds, and low pressure. The ITCZ over the area of study have a meridional and zonal component. The converging westerlies from Atlantic Ocean and easterlies from the Indian Ocean form the meridional arm. The zonal (East-West) arm of the ITCZ moves North- South and brings rainfall in the areas it passes. The general passage of the ITCZ over our country Kenya has been known as the main cause of rains during the MAM and the OND over the Rift valley region. A north-south pattern of rainfall is experienced due to the movement of the ITCZ, with most rainfall occurring during the MAM season (Nicholson, 1996). ITCZ is, therefore, the main synoptic system known to control the East African seasonal rainfall (Asnani and Kinuthia, 1979; Asnani, 2005) Tropical Cyclones A tropical cyclones are cyclonic storms whose origin are almost invariably in the low latitude between 5 and 20 degrees north or south of the Equator. It is a low pressure zone which grows from a depression, storm and finally a cyclone with a speed of equal or greater than 64knots.Characterized by a counter-clockwise rotation in the Northern Hemisphere (NH) and a clockwise rotation in the Southern Hemisphere (SH), its formation is favored by sea surface temperatures of at least 26.5 C. A fully developed cyclone shows a center characterized by sinking winds, which forms an eye. This center has calm weather free of clouds. Outside this eye is the eyewall, which has the highest wind speeds, rising air, cloudiness, and heavy precipitation (Zehnder, 2013). Cyclones season over South West Indian Ocean (SWIO) spans from October to May with peaks in January and February. It has a greater influence on the rainfall over countries in the East Africa region compared to those over the East Indian Ocean (Elsberry, 2006) as it 7

20 heighten the pressure gradients between the south Atlantic and Eastern Africa, causing moist westerlies (Congo/Zaire air mass) to converge into our region Monsoons Monsoons are low-latitude winds, which change direction seasonally. Depending on where the monsoon flows from, either land or water, they cause drastic changes in temperature and precipitation patterns on the area they affect. Monsoons are mainly caused by temperature difference between the land and the ocean. Kenya experiences two monsoonal circulations; the Southeast (SE) and Northeast (NE) monsoons. Okoola (1999) found out that these monsoons are experienced when the ITCZ is away from Kenya hence bringing little rainfall. The NE monsoon occurs during the DJF season originating from Arabian Peninsula and curves south of Equator to become a north westerly wind. During this season, this monsoon drives dry continental air to Kenya hence, it is associated with little or no rainfall. The SE monsoon on the other hand occurs during the JJA season and emanates from the south Indian Ocean with cool and moist properties due to the Mascarene highs. Reaching the north of equator, this monsoon re-curves to become south westerly flow. These two monsoons flow parallel to the coast and are diffluent in the low levels Mesoscale Systems Mesoscale systems are small scale weather systems covering a horizontal distance of between 5 km and 1000 km and periods of not more than 1 day. Over the Rift valley region the systems are as a result of the availability of mountains and valleys, and the many inland lakes. Some episodes of heavy precipitation can be attributed to these local systems. Examples of these systems include sea/land breezes, low level jets, rain bands, thunderstorms, gap winds, orographic lifting, and organized convective systems among others. In the Northern Kenya, for instance, we have Turkana Jetstream which occur throughout the year (Asnani, 1993; Kinuthia and Asnani, 1982). It occurs due to air being channeled in a gap between the East African highlands and the Ethiopian highlands. The wetter southeast (Marsabit) area of the gap and the drier northwest areas are associated with the dynamics of the Turkana channel Jetstream. 8

21 2.2.5El-Niño Southern Oscillation (ENSO) El-Niño is an anomaly resulting from the warming of the central and eastern equatorial Pacific Ocean, where the sea surface temperature anomalies are of magnitudes greater than 0.5. If this anomalous condition persists for five or more months, the condition is termed as an El-Niño episode. Various studies have shown that there exist a relationship between East Africa rainfall with El-Niño (Ogallo and Suleiman, 1987, Janowiak, 1988, Indeje, 2000; Mutemi, 2003 and many others. Mutemi 2003 in his expository found a strong relationship between the rainfall over East Africa and the evolutionary phase of ENSO. It determines the monthly and seasonal rainfall pattern in East Africa. Mutemi (2003) also identified the shift in onset or cessation of rainfall pattern over some regions and reduction in seasonal peak as the significant evidence. Nicholson and Kim (1997) found a strong relationship between the ENSO and rainfall over Africa and suggested linkages through ENSO induced SSTS anomalies in the Indian Ocean, which in turn modulates interannual variability of rainfall over Africa. They further noted that ENSO has less influence on the long rains (MAM) and significant effect on the short rains (OND). Ogalllo (1988) found a significant instantaneous and time lagged correlation between East Africa seasonal and the Southern Oscillation Index (SOI). Significant negative correlation between SOI and precipitation occurred along coastal regions and western parts of Kenya during short rainfall period. Ogallo et al (1988) correlated the global SST anomalies within the latitude 3degree north and south of the Equator with the Rotated Principal Component Analysis (RPCA) modes of the Northern hemisphere autumn rainfall over East Africa for the period The study suggested that about 36 percent of the short rainfall variation in East Africa could be explained by SST variation in the western pacific and most of the Indian Ocean where correlation values are near 0.6. Using an atmospheric general circulation model forced with various combinations of Indian Ocean and Pacific SST anomalies, Goddard and Graham (1999) noted that while the SST variability of the tropical pacific exerts some influence over the African region, it is the atmospheric response to the Indian Ocean variability that is essential for the model simulating robust rainfall response over eastern, central and southern Africa. 9

22 2.2.6 Indian Ocean dipole This is a mode of climate variability that involves ocean and atmosphere interaction. The Indian Ocean Dipole (IOD), which is the difference between the sea surface temperature anomalies of the Western and South eastern equatorial Indian Ocean is taken from (50 E-70 E, 10 S-10 N) and (90E-110 S, 10 S-0 ) respectively. Past studies (Saji et al., 1999; Saji and Yamagata, 2003b; Owiti, 2005; Owiti et al., 2008) have shown that the anomalous warm sea surface temperatures are sometimes experienced over the western Indian Ocean and cold SSTs over the Eastern region. This condition has been found to have a great influence on the rainfall over the East Africa region. Reason (2001) in the study over the Southern Africa showed that the winds from the Indian Ocean are influenced by IOD SST anomalies especially on the amount of moisture that is pushed to the region. IOD is said to be positive if the western Indian Ocean is warmer than the Eastern side, and vice versa. The positive phase of IOD is known to be of great influence on the rainfall condition over the East and South Africa region. It activates atmospheric convection and leads to enhanced rainfall during the rainy season and wet spells during the dry season. This phase causes an anomalous cooling of SST in the eastern tropical Indian Ocean and brings droughts in the Indonesian and Australian region. On the other hand, the negative phase involves high SSTs in the eastern Indian Ocean, which brings more rainfall to Australian and Indonesian region and less is experienced over the East African countries. Studies have also shown that just like ENSO, the IOD may induce unusual circulation and rainfall pattern in countries beyond the proximity of the Indian Ocean. Owiti et al. (2008) analyzed the evolution phases of IOD, and showed that it starts developing in April and reaches the peak in October or November and starts decaying in January. This therefore implies that the SST anomalies pattern experienced during IOD events has a strong signal on the East African climate system during the OND rainfall season, hence indicating that some of the extreme events that have been experienced over the region have a close relationship with the negative and positive phases of IOD Quasi-Biennial oscillation Indeje and Semazzi (2000) examined the relationship between March May seasonal rainfall anomalies over Eastern Africa and the lower equatorial stratospheric zonal winds from Spatial correlation patterns were studied to understand the climatic association between regional rainfall and equatorial stratospheric zonal wind at the interannual time scale. Results showed that 10

23 during march-may, there is a significant positive simultaneous and non-zero lag correlations between lower equatorial stratospheric zonal wind and rainfall over parts of East Africa. Significantly high correlations (+ 0.8) were found over the western regions of Eastern Africa. Furthermore, these month to month associations were observed to be outstanding during lag than in the simultaneous correlations and were observed after June preceding the onset of the March to May seasonal rainfall. The December February season were observed to have the weakest relationship Subtropical anticyclones These are synoptic-scale quasi-permanent high-pressure cells. They are characterized by anticyclonic circulations which gives rise to subsidence and low level horizontal velocity divergence of air masses. The four major anticyclones, which influence the flow of monsoonal winds over the region, are the Azores, Arabian, Mascarene and the St. Helena anticyclones (Griffiths et al. 1972; Boogaard, 1977). The strength, location, horizontal and vertical structure of the anticyclones closely determine the characteristics of the monsoonal winds and the associated air masses. The Azores and the Arabian anticyclones control the flow of the north easterlies from the northern hemisphere. While the flow of the southeast monsoons from the southern hemisphere is controlled by Mascarene and St. Helena anticyclones. The characteristics of the anticyclones together with other weather systems over the Indian Ocean and the Congo Basin (like the easterly waves, equatorial westerlies and tropical storms) are major sources of moisture into the Equatorial Eastern Africa (EEA). During the northern hemisphere summer (June-August) the Mascarene anticyclones intensifies while the Arabian anticyclones weakens. The ITCZ is pushed northwards and hence the convergence zone. During the equinoxes (March and September) the two anticyclones are somewhat balanced. During the southern hemisphere summer (December-January), the Arabian anticyclone intensifies while the Mascarenes weakens. The ITCZ moves southwards with the rain belt. Intensities of St. Helena anticyclones over the southern Atlantic reactivates the Congo/Zaire air mass, which causes the third peak in rainfall within a year over the western parts of the EEA region centered on July/August. 11

24 CHAPTER THREE 3.0 DATA AND METHODOLOGY This chapter describes the data and methods that were used to obtain the useful results for the goals targeted to be achieved in the study. 3.1 Data type, source and quality control This section discuss the type of data used resources and the methods to carry the data quality control Data type and source The data used include the daily rainfall data and the reanalysis fields. The daily rainfall data for the analysis was obtained from the Kenya Meteorological Service (KMS) from 1980 to 2015 (35yrs) for the following stations: Kericho, Eldoret-met station, Narok and Kitale. Table 1 show the location and altitudinal details of the four rainfall stations TABLE 1: LOCATION AND THE ALTITUDES OF STATIONS USED IN THE STUDY STATIONS LATITUDE LONGITUDE ALTITUDE KERICHO M KITALE M NAROK M ELDORET-MET M The monthly data for the reanalysis field was extracted from the Earth Science Research Laboratory (ESRL) ESRL : PSD : Monthly/Seasonal Composites 12

25 3.1.2 Data Quality control This consisted of data validation, estimation of missing and test for homogeneity Data validation This was done to ensure the truthfulness of the observed data and to verify that it is a true representation of the actual observation. This was achieved by identifying any unusual values, outliers and checking observations of nearby station Estimation of missing data Many long-term meteorological data have a number of gaps where data were not recorded. This problem among others has been the major setback in research. The missing data should be filled with the highest precision and their quality checked in order to provide reliable near-homogeneous series.with less than 10% of the missing monthly data were filled using arithmetic mean method shown in equation (1) the method used by De Silva et al. (2007). X Bi = X Ai X B X A (1) Where: I = 0, 1, 2.n X Bi is the station with missing data X Ai is the station with data X B is the mean of a station with the missing data X A is the mean of station with the complete data Homogeneity Test The consistency of the data was determine by using single mass curve. This involve plotting annual rainfall values of the observed rainfall against time. 13

26 3.2 METHODOLOGY The methodology used to achieve the objectives of the study are describe in the following subsections Determination of onset and cessation dates To meet the first objective a statistical analysis of daily rainfall data for the period for the stations used was done. The data was organised in a format required by Instat Software Program. Instat is a statistical program which was developed by the University of Reading in 2002 in United Kingdom. The tools offered by Instat and the under-mentioned criteria were used to answer the questions implied by the objectives Onset Several researchers have determine onset using different criteria, however the one adopted in the present study is according to Sivakumar (1988) The following criterion was used to determine onset; -Rainfall total of 20 mm or more totaled over 3 days - No dry spell of duration of 9 days or no more in the next 30 days: - should occur with an earliest starting day of March: - A threshold value of 0.85 mm Cessation The following criterion was used to determine cessation; The estimation of the end of the rainy seasons are based on the earliest possible Day of 1 st May, the capacity of soil to persist precipitation with a water balance Equal to zero is 100mm. 14

27 Classification of onset and cessation into Early, Normal and Late This was used to attain specific objective 2 where onset dates were classified into either early, normal or late. In order to carry out this exercise the onset and cessation dates were transformed into standardized anomalies. If the value obtained is less than -1 then early onset, greater than 1 is late onset and if less or equal 1 and greater or equal than Correlation Analysis This method was used to examine the simple relationship between pairs of variables. In this case, SYSTAT was used to compute the correlation coefficients which shows whether correlation is either negative or positive. A very high correlation implies a direct dependence of the variables, if it is zero then no correlation and if negative then the correlation is indirectly dependent. The simple correlation coefficient is given by the equation below: r xy n i n i ( x ( x i i x)( y x) 2 n i i ( y i y) y) 2 ( 2 ) Where: r xy is the computed value of coefficient. If r xy = +1 x and y have a strong positive linear correlation If r xy = -1 x and y have a strong negative linear correlation If r xy = 0 nonlinear relationship between the two variables n is the number of variables 15

28 The student t-test was used to test the significance of the values Plotting of the analysis fields In order to determine the systems associated with early and late onset, composite of the wind field, mean sea level pressure (mslp) and geopotential height were plotted. 16

29 Rainfall cummulative(mm) CHAPTER FOUR 4.0 RESULTS AND DISCUSSIONS This chapter consists of results and discussion from the analysis done in this study 4.1DATA QUALITY CONTROL ANALYSIS Results from homogeneity test Examples of single mass curves are presented in Figures 3, 4, 5 and 6. All the data showed that a single straight line could be fitted on cumulative rainfall amounts for any station. This is an indication that the data was homogeneous hence used without any adjustment KERICHO MASS CURVE Years FIGURE 2: CUMULATIVE MASS CURVE FOR KERICHO,

30 Rainfall cumulative(mm) Rainfall cumulatives(mm) ELDORET-MET MASS CURVE Years FIGURE 3: CUMULATIVE MASS CURVE FOR ELDORET MET KITALE MASS CURVE Years FIGURE 4: CUMULATIVE MASS CURVE FOR KITALE,

31 Rainfall cumulatives(mm) NAROK MASS CURVE Years FIGURE 5: CUMULATIVE MASS CURVE FOR NAROK:

32 4.2 Mean Onset and Cessation Dates Table 2 below shows the mean onset, cessation and duration for the 4 stations used in the study. It is shown that Kericho experiences the earliest onset (79) then Eldoret met, Kitale and Narok (86, 89 and 94) respectively that is between mid-march to the end of March It also shows that Kericho had largest duration, this is depicted by the mean duration. The onset is associated both with eastward and meridional progression of the ITCZ. It also shows that the other stations had marked cessation except for Kericho which depicted a progression to late August. For the Narok the onset day determine is contrary to that obtain by Mugo.R, M., Ininda J.M and Okoola R.E in the Interannual variability of onset and cessation of long rains in Kenya. The change may be attributed to climate change and other factors which needs to be investigated. TABLE 2: MEAN ONSET, CESSATION AND STANDARD DEVIATION Stations Mean onset day Standard deviation Mean cessation day Standard deviation MEAN DURATION Kitale Eldoret-met Kericho Narok Time series analysis Time series shown in figure 6,7,8,9 shows variability of the onset and cessation dates for individual years for the stations used in the study. Figure 6 shows the variability over Narok where early onset was experienced in the year 1988, 1994, 1995 and Late onset occurred in the year 1983 and The region experienced a late cessation in the year 1990 and Figure 7, shows the onset and cessation dates for Kitale station. Early onset occurred in the year 2006 and Late occurred in the year 1983, 1993 and 1995 while late cessation occurred in

33 Figure 8, depicts the time series for the Kericho station. Early onset occurred in the year 1995, 1999 and 2006, late onset in the year 1984, 1993,2000,2008,2012 and 2014 and late cessation occurred in the year Figure 9, shows the variability of onset and cessation for Eldoret met station. It had an early onset of 1995 and 2006, late onset in the year 1993,2000,2010,2012 and 2014 and a late cessation in the year This may be attributed to the Elnino of 1997/1998. All the stations showed a common early and a late onset during the year 2006 and 1993 respectively. FIGURE 6: NAROK ONSET AND CESSATION DATES OVER MAM SEASON,

34 FIGURE 7: KITALE ONSET AND CESSATION OVER MAM SEASON,

35 FIGURE 8: KERICHO ONSET AND CESSATION OVER MAM SEASON,

36 FIGURE 9: ELDORET MET ONSET AND CESSATION OVER MAM SEASON,

37 TABLE 3: ONSET AND CESSATION DATES OVER NAROK (E-EARLY, N-NORMAL, L-LATE) YEAR DAY ONSET DAY CESSATION DURATION May L Jun L Mar N May E Apr L May N May L May N Apr N May E Mar N May N Apr L May N Apr N May E Mar E May N Mar N May N Mar N Jun L Mar N Jun L Mar N May N May L May N Mar E May N 77 25

38 Mar E Jun N Apr L May E Mar N Jun N Apr N May E Mar E May E Apr N May E Apr N May N Apr N Jun N Mar N Jun L Mar N Jun N Mar N May E Mar E May N Apr N May E Apr N May N Apr N May E Apr L May N Apr N May E Apr N Jun N Mar E May N Mar E May N 54 26

39 Apr N Jun L 52 TABLE 4: ONSET AND CESSATION DATES OVER KITALE (E-EARLY, N=NORMAL, L=LATE) KITALE YEARS DAY ONSET DAY CESSATION DURATION May L Jun N Mar N Jun N Mar N Jun N Apr L May E Apr N May E Mar N Jun N Apr N May N Mar N Jun N Apr N Jun N Mar N Jun N Mar N Jun N Mar N May E 33 27

40 Apr L May N Apr L Jun N Mar E May N Apr L Jun N Mar N Jun N Mar N May N Mar E Jun N Apr L May N Apr N Jun N Mar E May N Apr N May N Apr N Aug L Mar N May N Apr N Jun N Mar E May E Apr N Oct L Mar E May E Apr N May E Mar E Jun N Mar N May N 74 28

41 Apr N Jun N Mar N May N Mar N May E Apr N Jul L 100 TABLE 5: ONSET AND CESSATION DATES OVER ELDORET MET (E=EARLY, N=NORMAL, L=LATE) YEAR DAY ONSET DAY CESSATION DURATION May L Jun N Mar N Jun N Mar N Jun N Mar N May N Apr N May N Mar N Jun N Apr L May N Mar N May N Apr L May N Mar N Jun N Mar N May N May L May N Apr L Jun N 36 29

42 Apr L May N Mar N Sep L Mar E May N Mar N May N Mar N May N Apr N Oct L Mar N May N Apr L Jun N May L May N Apr L May N Apr L Jun N Mar N May N Apr N Jun N Mar E May N May L Jul N Apr N May N Apr N May N Apr L May N Mar E May N Apr L Jun N 65 30

43 May N Jun N Apr L May N Mar N Jun N 84 TABLE 6: ONSET AND CESSATION DATES OVER KERICHO (E-EARLY, N-NORMAL, L-LATE) YEAR DAY ONSET DAY CESSATION DURATION Apr L Sep L Mar N Oct L Mar N Jul N Mar N Jul N Mar N Jul N Mar N Jun N Mar N May E Mar N May E Mar N Nov L Apr L Jul N Mar N Sep N Mar N Jul N Mar N Jun N Mar N May E 55 31

44 Apr N Aug N Mar E Jul N Apr L Sep N Mar N Aug N Mar N Jun N Mar N Jul N Mar N May E Mar N Sep L Mar E Oct L Mar N Nov L Apr L Sep N Mar N Jun N Mar N Aug N Mar N Oct L Apr L Jun N Mar N May E May L Jul N Apr L Jul N 85 32

45 KITALE 3 KERICHO STANDARDIZED VALUES ABOVE NORMAL BELOW NORMAL -2 STANDARDIZED VALUES ABOVE NORMAL ELDORET-MET STANDARDIZED VALUES ABOVE NORMAL BELOW NORMAL NAROK STANDARDIZED VALUES ABOVE NORMAL BELOW NORMAL FIGURE 10: ANOMALIES FOR THE STATIONS 33

46 Frequency Frequency 4.4 Temporal pattern of rainfall From the frequency graphs observed below for the four stations; Kericho, Eldoret met, Kitale and Narok (figure 10; a, b, c and d), it depicts that the onset is skewed to the right for Kericho and Narok, figure 10a&c respectively. This shows that there is normalcy in the onset for both the station that is to mean that, the occurrence of onset has been on early March for most of the years. For the Eldoret met station, figure 10b, it is reflected that the onset shifted to the center attaining a normal distribution curve. Implying that the onset of rainfall occurred during mid-march for most of the years in the region. However, Kitale station figure10d, has no marked onset frequency. This is shown by the station having two peaks implying that there is no normalcy. Kitale is more west and north and the unpredictability of its occurrence is due to the ITCZ migration. If it shifts meridionally then early onset is expected and therefore skewed to the right and if it is zonally then late onset and thus skewed to the left. Congo air mass and L.Victoria air mass are other attributing factors KERICHO ELDORET-MET onset date onset dates (a) (b) 34

47 Frequency Frequency NAROK KITALE Onset date Onset dates (c) (d) FIGURE 11: FREQUENCY OF ONSET OF RAINFALL 4.5 Classification of occurrence of onset and cessation dates into classes (early, normal and late) This summary statistic was done to obtain the classification which will in turn help us achieve objective two under section 2.1. Table 3, 4, 5 and 6 shows the various classification. It can be seen from the tables that stations had early, normal and late onset as well as cessation. This classification helped us to identify the common years where the stations had early/late onset so as to plot the analysis field of winds, mean sea level pressure and the geopotential height anomalies as shown in figure 10 and 11. Figure 10-A and B shows sea level pressure and the geopotential height anomaly. Negative values as indicated by the scale corresponds to low sea level pressure implying that there is probably a cyclone or a storm near the surface of the earth. Figure 10-C shows the vector wind at 850mb composite anomaly. There was influx of south easterlies from the Indian Ocean and north westerlies from Azores High, Atlantic Ocean. This leads to convergence forming meridional component of the ITCZ into our region which facilitates 35

48 the early onset. ITCZ is, therefore, the main synoptic system known to control the East Africa seasonal rainfall (Asnani and Kinuthia, 1979; Asnani, 2005). Moreover, the north westerlies winds play an important role. The winds are moist, and thus modifying the Congo air mass bringing substantial rainfall in the Central Rift valley region. This may have contributed to early onset experienced that year. For the year 1993, we experienced late onset because the conditions wasn t favorable. As seen from the figure 11, the geopotential height and the sea level pressure anomalies was zero and had no impact on the synoptic systems. As for the winds, the westerlies blow from Atlantic Ocean reaching West Africa they curve back and towards north whereas the easterlies from the Indian Ocean blow towards southern Tanzania, towards Somalia and the horn of Africa. (B) (A) 36

49 (C) FIGURE 12: FACTORS INFLUENCING THE SYNOPTIC SYSTEMS THUS EARLY ONSET (A) SEA LEVEL PRESSURE (B) GEOPOTENTIAL HEIGHT (C) WINDS (ANOMALIES) 37

50 (A) (B) (C) FIGURE 13: FACTORS INFLUENCING SYNOPTIC SYSTEMS THUS LATE ONSET: (A) GEOPOTENTIAL HEIGHT (B) SEA LEVEL PRESSURE (C) WINDS 38

51 4.6 Correlation analysis Table 7 below shows the correlation between rainfall totals, onset, cessation and duration of rainfall over the MAM period. It was noted that the onset were negatively correlated with the rainfall amount/totals. This imply that early onset would favor above normal rainfall. On the other hand cessation dates were positively correlated with rainfall indicating that a poor season is marked by early withdrawal of rainfall. Moreover, it was also observed that duration of rainfall was negatively correlated with the onset, this shows that early onset marks a late withdrawal of rainfall thus a long duration of rainfall. The t-test was used, it showed that t-computed for Narok, Kitale, Kericho and Eldoret met is 4.1, 3.0, 1.6 and 2.4 respectively and the t-tabulated is It shows that the correlation was significant 39

52 TABLE 7: CORRELATION COEFFICIENTS FOR THE FOUR STATIONS: NAROK, KITALE, KERICHO AND ELDORET MET NAROK ONSET CESSATION DURATION TOTALS ONSET CESSATION DURATION TOTALS KITALE ONSET CESSATION DURATION TOTALS ONSET CESSATION DURATION TOTALS KERICHO ONSET CESSATION TOTALS DURATION ONSET CESSATION TOTALS DURATION ELDORET MET ONSET CESSATION TOTALS DURATION ONSET CESSATION TOTALS DURATION

53 CHAPTER FOUR 4.0SUMMARY, CONCLUSION AND RECOMMENDATIONS 4.1 Summary Rainfall has a significant effect on the economy of our country. This is because our country economy depends on a rain fed agriculture. As crop growth spans from land preparation to harvesting, appropriate knowledge on the expected seasonal performance is of importance. The development stages of crop growth until maturity and general efforts to curb crop failure cannot be achieved without a reliable knowledge on the onset, its duration and the cessation of the rains. Onset and cessation dates are very important for countries inclusive of Kenya who rely on rain fed agriculture. For us to achieve the set objectives, data from for the 4 stations in central Rift valley (Kericho, Narok, Eldoret met and Kitale) was used. The data was analysed using the INSTAT software which generated the onset and cessation dates which were further categories into early, normal and late. It was observed that most years were grouped under the normal category however a few years lied on the extreme cases as shown in Table 3-6. Moreover, the time series analysis were plotted to show the temporal pattern of onset and cessation dates (figure 6-9) over the period. The onset occurrence depends highly on the meridional and zonal movement of the ITCZ. From the analysis, it was observed that Kericho had the earliest onset followed by Eldoret met, Kitale and Narok respectively (Table 2). This is attributed to the both the eastward and meridional progression of the ITCZ. From the frequency graphs plotted figure 10a-d, we observed the normalcy in the station (a) and (c) while the other stations had undeterminable skewness. Correlation was also done between the onset, cessation, totals and duration of rainfall 4.2 Conclusion From the analysis it was observed that the mean onset range between 20 th March- 3 rd April. The stations located to the west were noted to receive an early onset compared to the rest. The study further showed that the mean cessation range between 22 nd may-6 th June. However, for stations with trimodal rainfall distribution they do not have distinct date of cessation. 41

54 Onset is negatively correlated with the seasonal performance suggesting that when the rains comes earlier (late) then the season is likely to be above (below) normal. From the plot field analysis, it showed that there was a high pressure at the North Atlantic and a low in the West Africa. The warm air from the high leads to the east ward movement of the meridional arm of the ITCZ. With these dates in mind, we can therefore advise farmers to prepare well in advance in relation to planting and harvesting among other farm activities to be carried out. 4.3 Recommendations Study should be done in the study area with higher resolution (more dense) station network than used in the present study to provide more detailed spatial distribution of onset and cessation. This could be achieved by complimenting the observation with satellite derived data. Studies should be done in the region on the characteristics of JJA rainy season since they play a major role for the crops that extend beyond the 3-month season. More studies need to be carried out to unearth the various courses on the interannual variability of onset and cessation. 42