Analysis of Climate Variability, Trends and expected implication on crop production in a Semi-Arid Environment of Machakos County, Kenya

Advances in Agricultural Science 07 (2019), 01: 99-115

Analysis of Climate Variability, Trends and expected implication on crop production in a Semi-Arid Environment of Machakos County, Kenya

Emily Bosire 1*, Wilson Gitau 1, Fredrick Karanja 1 and Gilbert Ouma 1,2 

Department of Meteorology University of Nairobi, Kenya, P. O. Box 30197 – 00100, Nairobi Kenya.
Institute for Climate Change and Adaptation, University of Nairobi, Kenya, P. O. Box 30197 – 00100, Nairobi Kenya.


Climate variability and change are among the greatest hindrance to the realization of the Rio +20 declaration addressed by Sustainable Development Goals (SDGs), the New Partnership for Africa’s Development (NEPAD’s) Initiatives on “Food for All” and the planned Kenyan vision 2030 of transforming Kenya to self-reliance in food production since it directly influences agricultural production and community livelihoods. Knowledge of climate variability is currently of remarkable importance in Kenya, where agriculture is the driver of its economy contributing approximately 26 % of the Gross Domestic Product.  This study investigated the extent of the temporal variability and trends of rainfall, temperature and their effect on crop production for the period (1981 to 2013) in Machakos County, Kenya. The findings show that the March to May (MAM) season is characterized by high variations seasonal rainfall compared to the October to December (OND) season in the study area, but with low variability in both seasonal and annual temperatures. The results also showed insignificant trends for the intra seasonal rainfall characteristics. However, significant decreasing trends of rainfall at seasonal and annual scales were recorded at two stations out of the three studied (Iveti and Machakos DO station).  Results of the maximum (minimum) temperature show significant increase during the MAM (OND) season and at annual scale.

Keywords: Climate variabilityCrop productionIntra seasonal rainfallMachakos County

How to Cite: Bosire, E., Gitau, W., Karanja, F., & Ouma, G. (2018). Analysis of Climate Variability, Trends and expected implication on crop production in a Semi-Arid Environment of Machakos County, Kenya. Advances in Agricultural Science7(1), 99-115.   


Agriculture continues to be the most important driver of Kenya’s economy, contributing about 26 percent of the country’s Gross Domestic Product (GDP), accounting for approximately 65 percent of total exports, and employing formally and informally approximately 18 and 70 percent, respectively, of country’s total workforce (GoK, 2009; 2010). In spite of innovative advances such as adoption of high yielding crop varieties and use of more efficient irrigation systems, most of the agricultural practices are still rain-fed and therefore highly reliant on climate. This reliance of Kenya’s population on rain fed agriculture makes their food security highly exposed to the risks linked to climate, hence making adaptation to climate change a great necessity.

Assessing the variability of climate parameters over an area based on past records is vital in the quantification of their effects particularly on crop yields. This will allow for the best adaptation measures to be suggested (Tamiru et al., 2015).  While climate determines the types of crops grown in particular regions, the intensity and distribution of rainfall within a cropping season greatly affects the crop production and yields. The intensity and distribution of rainfall determine the need for irrigation and it also influences the agricultural calendar. On the other hand, temperature not only affects viability of pollen and number of seeds (Schoper et al., 1986, Prasad et al., 2006) but also affects crop yields primarily by influencing the rate at which biomass is accumulated and the growth cycle (Kimball et al., 2002; Ainsworth and Long, 2005; Fuhrer, 2003).

Agriculture is the main economic activity in Machakos County. It is leading sector with regard to food security, employment and income earnings. The commonly known factors that have limited agricultural production in the county include; high cost of inputs (fertilizer and hybrid seeds), few extension services, shortage of markets, and limited use of agricultural technology and innovation (such as enhanced crop varieties and irrigation systems) and unfavorable climatic conditions. The agricultural production in the County is dominated by cereals, pulses and root crops.

Machakos County is often characterized by high variability of rainfall amidst increasing temperatures that tend to increase the rate of evapotranspiration. Seasonal onset dates, cessation dates, number of rainy days and duration of growing season seem to be unpredictable in current decades. These have hindered efforts to improve food security situation in the county ( Mwandalu and Mwangi, 2013).

Several studies have examined trends in the climate variables across Eastern Africa (Camberlin and Okoola, 2003; Seleshi and Zanke, 2004; Tilahun, 2006; Mugalavai et al., 2008; Njiru et al., 2010; Recha et al., 2011; Wagesho et al., 2013; Kansiime et al., 2013; Omondi et al., 2013; Opiyo et al., 2014; Omoyo et al., 2015; Gitau et al., 2018).

Studies on the variability of intra-seasonal rainfall characteristics over Eastern Africa have shown that the variability of onset dates is always higher than cessation dates ( Camberlin and Okoola, 2003; Mugalaavai et al., 2008; Recha  et al., 2011; Omoyo et al., 2015). Tilahun (2006), obtained higher variability of annual rainfall in Ethiopia. In Uganda, Kansiime et al. (2013) found higher and lower variability within and between seasons in the highlands and lower lands areas, respectively.

Previous studies on the seasonal and annual rainfall trends over East Africa depicted different trends. Kansiime et al. (2013) found positive significant trends in the seasonal and annual rainfall for the highlands areas of Uganda and insignificant decreasing trends for the low lying areas. In the study conducted in Ethiopia, no trend was noted in the rainy days, seasonal and annual rainfall over Central, Northern and North West stations of the country. However, JJAS seasonal and annual rainfall showed significant decreasing trends (95% confidence level) over Eastern, Southern and South West stations (Seleshi and Zanke, 2004).

In Kenya studies have shown insignificant decreasing trends of rainfall at seasonal and annual scales (Njiru et al., 2010; Opiyo et al., 2014; Omoyo et al., 2015). On the other hand, temperature patterns exhibited warming trends at both seasonal and annual scales.

In recent years, trend analyses of rainfall and temperature patterns in Machakos County have been carried out (Njiru et al., 2010; Omoyo et al., 2015). Njiru et al. (2010) analysed climate data and its associated risks on maize production in Machakos and Makueni Counties. Meteorological stations used were Katumani and Kambi ya Mawe. Rainfall and temperature data for Katumani meteorological station was for the period 1957 to 2008 and 1986 to 2008, respectively, whereas, for Kambi ya Mawe rainfall and temperature data was from 1959 to 2008 and  from 1971 to 2008, respectively. They observed high variability in the short rains (OND). The results also depicted a decline in both the seasonal and annual rainfall. Increasing trends in both maximum and minimum temperatures were revealed.

Omoyo et al. (2015) evaluated the effects of climate variability on maize grain yield in Machakos, Kitui, Mwingi and Makueni Counties for the period 1979 to 2009. In Katumani, Machakos County, the results showed insignificant decreasing trends in the seasonal and annual rainfall over the study period, whereas temperature patterns exhibited warming trends at both seasonal and annual scales. Both MAM and OND seasonal rainfall exhibited high variability (CV > 30%). It was also noted that the onsets were highly variable (CV= 98.1%).

However, such studies did not present comprehensive information on the variability and trends of the intra-seasonal rainfall characteristics within the county. Omoyo et al. (2015) only considered the onset, cessation dates and seasonal rainfall amounts in their study. Understanding the variability and trends of most of the intra-seasonal rainfall characteristics (duration, frequency of dry spells and number of rainy days and etc) is important because they have repercussions on the distribution of water within a growing season which finally affects crop yields.

In addition, none of the studies within Machakos County considered the non-parametric methods such as Mann-Kendall test to analyze the temporal trends. Although the non- parametric methods are less powerful compared to parametric methods, they don’t require the data to be normally distributed, they are insensitive to outliers (Shadmani et al., 2012). Relevant reviews on trend analysis of climate parameters using Mann-Kendall test include Modarress and Silva 2007; Boroujerdy 2008; Tabari et al., 2011; Mondal et al., 2012; Wagesho et al., 2013; Gitau et al.,2018.

The primary objective of this study was to understand the variability and trend of rainfall and temperature and their expected implications on crop production in the semi-arid environments of Machakos County, Kenya. The results will help guide the farmers within the communities in preparing themselves for any extreme events and thus enhance crop yields.  The findings are also intended to assist in designing appropriate adaptation strategies related to climate change.

2. Data and Methodology

2.1. Description of the study area

The area of study was Machakos County which borders Nairobi and Kiambu, Embu, Kitui, Makueni, Kajiado and Murang’a and Kirinyanga counties to West, North, East, SouthWest and North West, respectively. The County comprises of four sub-Counties, that is, Machakos, Kangundo, Mwala and Yatta (Figure 1). It covers an area of approximately 6,208 km2 with an estimated human population of 1,335,387 according to the 2009 Kenya population and housing census and the projected annual growth rate of 2.47% (KNBS, 2010).

The annual mean minimum and maximum temperature are 13.7oC and 24.7oC, respectively (Wamari et al., 2012). It has a bimodal rainfall pattern, the long rains which commences in March and lasts till May (MAM) and the short rains which commences in October until December (OND). The average seasonal rainfall for the long and short rains is approximately 277mm and 300mm, respectively (Shisanya et al., 2011), while annual mean is approximately 655mm (Wamari et al., 2012).

The soils at Katumani are said to be crusting sandy clay loams and are classified as Chromic Luvisols (FAO/UNESCO Classification in 1990). This type of soil always portrays a quick breakdown in its aggregates when exposed to strong rainstorm events because of its unstable structure owing to low organic matter content in the soils (0.5 to 1%). The soils are also slightly acidic with a pH range from5.7 to 6.9 (Gicheru and Ita, 1987)

Machakos County was selected for this study because in comparison with other ASAL in the country, the County has the potential to feed itself remarkably when compared to the Counties in the northern parts where crop production is intensely

Figure 1. Map of Machakos County, showing the four sub-counties including the meteorological stations used in the study

Figure 1. Map of Machakos County, showing the four sub-counties including the meteorological stations used in the study


Table 1. Details of the Meteorological Stations used in the Study

Meteorological Station Latitude Longitude Elevation Data
Rainfall Maximum Temperature Minimum Temperature
Katumani 1.58°S 37.23°E 1592m 1981-2012 1981-2012 1981-2012
Iveti Forest Rangers Post 1.47°S 37.28°E 1890m 1981-2012 Not Available Not Available
Machakos District Office 1.52°S 37.27°E 1646m 1981-2010 Not Available Not Available

inhibited by hot and dry climate and the livelihoods of the community are restricted to pastoralism. In comparison to other Counties in the southern and central parts of the country, the County’s close proximity to the country’s capital city was an added advantage, which allowed for ease of access to the research station.


2.2. Data

Quality controlled observed daily rainfall and monthly maximum and minimum temperature data for Katumani Agrometeorological station was employed in the study. Monthly rainfall data from two other stations (Iveti and Machakos DO) were also used. These sets of data were acquired from Kenya Meteorological Department (KMD). Details of these data sets are as shown in Table 1.


2.3. Methodology

2.3.1. Determination of Intra-Seasonal Rainfall Characteristics

The intra-seasonal rainfall characteristics; onset, cessation, duration and the number of rainy days were computed for each year. Katumani meteorological station was used to compute the intra-seasonal rainfall characteristics because it was the only station with observed daily rainfall.

In a season, rainfall onset and cessation, which signifies the start and end of the seasonal rainfall, have been defined in different ways in several studies (Stern et al., 1982; Sivakumarr, 1988; Kasei and Afuakwa 1991; Omotosho et al., 2000; Dodd and Jolliffe, 2001; Stern et al., 2003; Tsefaye and Walker, 2004; Kihupi et al., 2007; Tadross et al., 2009; Marteau et al., 2011). In this study onset and cessation criterions by Stern et al. (1982) were adopted. Hence onset date was defined as the day after 1st March for MAM or 1st October for OND that received at least 20mm of rainfall which had accumulated over 3 successive days and number of dry spell should not exceed 7 days within the following 30 days. The criterion for defining the onsets was informed by the broad perception that the MAM and OND seasonal rainfalls begin in the months of March and October, respectively. The state of having no dry spell of more than 7 days after the begining of the growing season reduces the likelihood of a false onset. An episode of 30 days represents mean length for the first growth stage (vegetative phase) of the majority crops (Allen et al., 1998). After 30 days majority of the crops would have surfaced and be well established. Similarly, cessation date was defined as any day after 1st May and 1st December for the MAM and OND season, respectively, when the water balance becomes zero (Stern et al., 1982; Tsefaye and Walker, 2004).

The duration of the seasonal rainfall was determined by subtracting the onset dates from the cessation dates (Stern et al., 1982; Tesfaye and Walker, 2004). Moreover, although rainfall measurements can be as low as 0.1 mm, a threshold value of 1mm was used to define a rainy day because a value of 0.1mm of rainfall more or less has no effect on growth of crops ( Robel et al.,2013) and is highly variable to instrumental errors. The rainy days were determined by counting all the days within the specific season with rainfall amounts exceeding or equal to 1mm (Stern et al., 1982; Segele and Lamb, 2005; Hadgu et al., 2013).

The percentage cumulative mean rainfall was also used to establish the onset and cessation of the seasonal rainfall. In addition, it was also used to estimate the rainfall amounts of and rainy days that had accumulated by time of start and withdrawal of the rains during both seasons. The initial step in this method was to obtain the mean rainfall amount and number of rainy days that occurred at each five day interval of each year for a specific season. The next step was to compute the percentage of the mean rainy days and rainfall amounts and that occurred at each pentad throughout each year for a particular season. Thirdly, the percentages of the pentads were cumulated. Finally, the cumulated percentage were plotted against the pentads, rainfall onset corresponded to the initial point of maximum positive curving of the graph, while the end of the rains corresponded to final point of maximum negative curving (Adejuwon et al., 1990; Odekunle, 2006).


2.3.2. Variability of Climate Parameters

To quantify the variability of the climate parameters, the coefficient of variation (CV) was computed as shown by Equation 1. Percentage Coefficient of variation values are classified as follows: < 20% as less variable, 20-30% as moderately variable, and > 30% as highly variable (Hare, 1983; Araya and Stroosnijder, 2011).


2.3.3. Trend Analysis of Rainfall and Temperature

Linear regression method was used to visualize and derive the magnitudes of trends in the intra-seasonal rainfall characteristics, seasonal and annual totals of rainfall and maximum and minimum temperatures. The statistical significance of the observed trends was determined using the Mann-Kendall test (Jaagus, 2006; Partal and Kahya, 2006; Yenigun et al., 2008; Hadgu et al., 2013; Gitau et al., 2018).


3. Results and Discussion

 3.1. Variability in the Intra-Seasonal Rainfall Characteristics

The onset for MAM and OND ranged from 1st March to 28th April and 8th October to 27th November, respectively. The long term mean onset for MAM and OND seasonal rainfall was 26th March and 1st November, respectively (Table 2). These dates could be taken as a reliable planting dates within the County. By the time of the onset, the station record 22% and 13% of the total seasonal rainfall amount and 19% and 14% of the total number of rainy days for both MAM and OND growing seasons, respectively (Figure 2).  Since rainfall in the study area is highly variable, this means that in a normal growing season sowing of seeds should be done prior to or on the onset. Planting later after the onset may result in crop failure, which may necessitate replanting or reduced crop yield as a result of unmet water requirements by the crop. The cumulated percentages of number of rainy days and seasonal rainfall amounts at the time of onset are enough to allow for the germination and establishment of the seeds. The attribution being that at the beginning of the growing season, less water is required (1 to 2.5mm/day) (Stichler and Fipps, 2003) because the plants are young and transpiration is low due to plants having small leaf surface area index.

The cessation for MAM and OND ranged from 1st May to 7th June and 1st December to 31st December, respectively. The mean cessation for MAM and OND seasonal rainfall was 14th May and 12th December, respectively (Table 2). By the time the rainfall was retreating the station received 94% and 87% of the total seasonal rainfall amount and 90% and 86% of the total number of rainy days for both MAM and OND growing seasons, respectively (Figure 2). This indicated that the susceptibility of the reproductive phase to water stress was minimal. This is because the water requirements in the reproductive phase (flowering and seed setting) increases to around 7 to 10mm/day (Stichler and Fipps, 2003) to allow for seeds to achieve their maximum weight.

The remaining percentages of the cumulative total seasonal rainfall amount and the total number of rainy days for both MAM and OND growing seasons was most likely enough for the crops to attain maturity. Daily water use by the crops is known to decrease steadily during grain filling stage as the leaves begin to senescence and the plant approaches physiological maturity.

Though, onset and cessation dates vary annually, the amplitude of variation of cessation at 15.9% and 15.3% for MAM and OND seasons is lower compared to the onset at 62.3% and 34.0% for MAM and OND seasons, respectively. This indicates that cessation dates are easily predicted in the area of study. Hence, decisions associated with harvesting of the crop, transportation, storage or selling are more easily made and that the decision connected to land preparation and crop planting should be taken with great caution.  Camberlin and Okoola (2003), Mugalavai et al. (2008) and Recha et al. (2011) reported similar larger inter-annual variability in onset compared to cessation dates for the wider Eastern Africa region.

Table 2 also shows that the duration of seasonal rainfall and number of rainy days within a season were highly variable (32.5% to 42.2%) with that for the MAM growing season being slightly higher (42.2% and 36.6%) as compared to the OND season (41.1% and 32.5%).  Higher variability on the seasonal rainfall duration within the study area could be attributed to higher variability in the onset

Table 2. Descriptive statistics of intra-seasonal rainfall characteristics in Katumani Meteorological Station during MAM and OND Seasons for the period 1981 to 2012.

Seasonal Rainfall Characteristics Descriptive Statistics MAM OND
Onset Latest (days/dates) 119 (28th April) 332 (27th November)
Earliest (days/dates) 61 (1st March) 282 (8th October)
Mean (days/dates) 86 (26th March) 306 (1st November)
Standard Deviation (days) 16 11
CV (%) 62.3 34.0
Cessation Latest (days/dates) 159 (7th June) 365 (31st December)
Earliest (days/dates) 122 (1st May) 336 (1st December)
Mean (days/dates) 135 (14th May) 347 (12th December)
Standard Deviation (days) 12 11
CV (%) 15.9 15.3
Duration Longest (days) 90 83
Shortest(days) 20 18
Mean (days) 49 42
Standard Deviation (days) 21 17
CV (%) 42.2 41.1
Number of Rainy Days Maximum (days) 40 55
Minimum (days) 8 10
Mean (days) 24 28
Standard Deviation (days) 9 9
CV (%) 36.6 32.5


Figure 2. Cumulative percentages of rainfall amounts and number of rainy days for Katumani Meteorological station during the MAM and OND seasons. The yellow and black circles represent the cumulative percentages of rainfall amounts and rainy days respectively at the time of onset and cessation.

Figure 2. Cumulative percentages of rainfall amounts and number of rainy days for Katumani Meteorological station during the MAM and OND seasons. The yellow and black circles represent the cumulative percentages of rainfall amounts and rainy days respectively at the time of onset and cessation.


dates because the duration of the season is highly correlated to the onset of rainfall. Higher coefficient of variation on seasonal rainfall duration makes it difficult to plan for the type of crops to grow based on their growth cycle.

Since the intra-seasonal rainfall characteristics for the MAM season are characterized by high variations as compared to those of the OND season in Katumani, the MAM season is therefore said to be unreliable in terms of rain-dependant crop production resulting to less dependency on the rains for planning any agricultural activity.


3.2. Variation in the Monthly, Seasonal and Annual Rainfall Amounts

Analysis of rainfall amounts by months for all the three stations shows that the initial and last months of the two seasons (MAM and OND) are characterized by high variability compared to the peak months (April and November) (Table 3). Comparable results are reported by Recha et al. (2011) in which first (March and October) and last (May and December) months of the MAM and OND seasons are characterized by variations greater than the mid seasonal months (April and November) in Tharaka district, Kenya. These results are also consistent with Sivakumar (1987) in Sudano-Sahelian zone.

In addition, rainfall received during the seasonal peak months of April and November accounts for approximately 50% of the MAM and OND seasonal rainfall totals at the three stations considered in this study (Table 3).

Approximately 28% and 16.5% of the total seasonal rainfall in the three meteorological stations is received during the onset months (March and October). On the other hand, the cessation months (May and December) accounts for approximately 20% and 33% of the seasonal rainfall totals (Table 3).  The findings of this study are in agreement with Recha et al. (2011) where the cessation of OND season (December) accounted for greater percentage as compared to May the cessation month of MAM season. The results indicate that the OND seasonal rainfall amount is fairly well distributed throughout the three months of season at 16.5%, 50% and 33%, hence minimizing the effect of within season variability. Less amount of rainfall is received in May and might not be adequate for crop production in case of drought within the county where the water holding capacity of the crusting sandy clay loams is low.

Analysis of seasonal rainfall amounts (Table 3) shows that Iveti and Katumani receive more rainfall during the OND season. However, for Katumani the difference in rainfall amounts during the two seasons was not statistically significant.  These findings are in agreement with those of Amissah-Arthur et al. (2002), Barron et al. (2003) and Njiru et al. (2010)) which showed that some regions within the Eastern Kenya receive more rainfall during the OND season than MAM season. On the other hand, Machakos D.O station recorded slightly less rainfall during OND season compared to the MAM season. The seasonal amounts received at Katumani during the MAM and OND seasons were 9.6% and 1.2% less compared to the climatological mean for the period (1958-1980) an indication that the seasons are becoming  dry (KMD, 1984).

Considering the three stations in the study, both MAM and OND seasonal rainfall had their coefficient of variation exceeding 30%. According to Araya and Stroosnijder (2011) and Hare (1983), a value greater than 30% indicates higher variability of the parameter. However, the MAM season had slightly higher variability in comparison to the OND season (Table 3). The study results are consistent with other analysis (Mutai and Ward, 2000) where MAM seasonal rainfall variability in East Africa region is higher than OND seasonal rainfall. These results nonetheless contradict those of Recha et al. (2011) who established that in Tharaka District (Eastern Kenya) that the OND seasonal rainfall was highly variable.


Table 3. Mean Monthly, Seasonal and Annual Rainfall Totals (RT), their Respective Coefficient of Variation (CV) and Percentage Proportion for the three Stations used for the period 1981 to 2012.

Stations Parameter March April May October November December MAM OND Annual
Iveti RT 119.4 191.8 83.9 92.1 255.4 156.9 395.1 504.5 1035.5
CV (%) 78.2 67.9 91.5 104.1 53.3 59.1 56.5 52.6 39.2
Proportion (%) 30.2 48.5 21.2 18.2 50.6 31.1 38.2 48.7
Machakos D.O RT 79.4 195.6 69.7 54.6 172.9 115.4 344.7 342.9 816.6
CV (%) 75.6 71.5 94.5 90.2 58.2 73.7 57.8 51.6 36.7
Proportion (%) 23.0 56.7 20.2 15.9 50.4 33.6 42.5 41.9
Katumani RT 86.7 133.4 55.0 43.4 143.6 94.6 272.1 279.6 699.1
CV (%) 70.9 45.5 79.1 87.3 59.8 85.6 44.5 44.2 26.3
Proportion (%) 31.8 49.2 20.2 15.5 51.3 34.7 38.9 40.1



Table 4. Descriptive Statistics of Seasonal (MAM and OND) and Annual Temperature over Katumani

Parameter Descriptive Statistics Seasonal Temperature Annual Temperature
Tmax Maximum 27.0 26.2 25.9
  Minimum 24.7 24.2 24.0
  Mean 25.9 25.2 25.1
  CV 2.3% 1.9% 1.5%
Tmin Maximum 15.3 15.3 14.1
  Minimum 12.9 13.1 12.3
  Mean 14.5 14.2 13.3
  CV 3.4% 3.5% 3.0%


In the study area the annual rainfall was 1035.5mm, 816.6mm and 699.1mm for Iveti, Machakos D.O and Katumani stations, respectively. The highest amount of rainfall recorded at Iveti could be attributed to the higher amounts of seasonal rainfall received at the station. The annual variability was higher at Iveti and Machakos D.O stations (39.2% and 36.7%), while for Katumani the variability was moderate (26.3%).


3.3. Variation in the Seasonal and Annual Temperature

During the MAM season the mean maximum and minimum temperatures were higher as compared to the OND and annual, with values ranging from 25.9 ± 0.6 and 14.5 ± 0.5, respectively (Table 4). Comparing to the climatological mean for the period 1958 to 1980, maximum temperature during the MAM and OND season has increased by 2.8% and 1.6%, respectively. On the other hand, minimum temperature during the two seasons decreased 3.9% and 1.4%, respectively. At annual scale maximum and minimum temperatures have increased and decreased by 1.6% and 2.9%, respectively.

Both maximum and minimum temperature depicted less variability (<5%) at both seasonal and annual timescales. However, the variability for maximum temperature was less compared to that of minimum temperature.

Comparing the variability of rainfall and temperature, rainfall characteristics showed greater variability while temperature depicted less variation both at seasonal and annual scale. This indicates that seasonal and annual minimum and maximum temperature are almost stable.


3.4. Trend Analysis of Intra-Seasonal Rainfall Characteristics

Decreasing trends in the onset and cessation were observed during the MAM and OND season, depicting early onsets and early cessations an indication that the seasons were shifting (Figure 3 and Table 5). However, the observed trends in the intra- seasonal rainfall characteristics were insignificant.

On the other hand, the duration for both seasons and number of rainy days during the OND season depicted an insignificant increasing trend. An increasing change in the duration and number of rainy days during the OND season indicates the suitability of the season for rain-fed agriculture in the study area because of enhanced soil moisture content owing to slightly lower variability of the onset of the OND seasonal rainfall.


3.5. Trend Analysis of Seasonal and Annual Rainfall

The trends of seasonal and annual rainfall and their magnitudes for three weather stations (Iveti, Machakos DO and Katumani) for the period 1981-2012 obtained by the Mann-Kendall and the linear regression analysis are shown in Table 6. During the MAM season, the trend test detected a significant and insignificant decreasing trend at Machakos DO and Iveti stations, respectively. On the contrary, Katumani was characterized by a slight positive trend, although it was not statistically significant (Table 6). The trend tests detected statistically significant decreasing trend in the OND seasonal rainfall for both Iveti and Machakos DO (Table 7). The rate of change of the significant decreasing trend during the OND season ranged between (-) 14.06 and (-) 8.1mm per year at Iveti and Machakos DO, respectively (Table 6).

Generally the OND seasonal rainfall depicted declining trends within the study area. The findings of this study agree with Njiru et al. (2010) where the OND seasonal rainfall revealed a decreasing trend in Katumani. However, the results are contrary to other seasonal analysis (Shisanya et al., 2011; Schreck and Semazzi, 2004) where the OND seasonal rainfall depicted a slight positive trend in part of the ASALs of Kenya. These contradicting results could be attributed to the methods of analysis used whether parametric or non-parametric, length of the data sets, whether the data was gridded or point observation and the area of study. Thus there is need for performing location specific analyses of rainfall trends to ascertain arguable affirmations on the same. Amissah-Arthur et al. (2002);  Hansen and Indeje (2004) suggested that OND seasonal rainfall Eastern Kenya represent the major growing season in on which most annual crops are reliant on. Hence, its decline has repercussions on production of crops and related livelihoods. This calls for an assessment of existing crop cultivars so as to determine their performance in the recent rainfall regime.

Analysis of annual rainfall revealed that there was a decreasing trend in the three stations. However, the trends were only statistically significant for Iveti and Machakos DO stations (Table 6). These findings are consistent with other earlier studies conducted in the ASALs of Kenya where the trends in the annual rainfall were decreasing significantly (Shisanya et al., 2011; Omoyo et al., 2015).

The significant observed decreasing trends in the OND seasonal rainfall and annual rainfall at Iveti and Machakos D.O imply that OND seasonal rainfall is a considerable determinant of variability of annual rainfall within the County.


Figure 3. Trends for onset and cessation during the MAM season for the period 1981 to 2012 at Katumani Meteorological station

Figure 3. Trends for onset and cessation during the MAM season for the period 1981 to 2012 at Katumani Meteorological station


Table 5. Observed trends in the MAM and OND intra-seasonal rainfall characteristics at Katumani for the period 1981 to 2012

Season Parameter Mann-Kendall a
tau p-value
MAM Onset -0.16 0.16 -0.22
  Cessation -0.09 0.45 -0.06
  Duration 0.03 0.81 0.17
  Number of rainy days -0.02 0.80 -0.05
OND Onset -0.08 0.49 -0.21
  Cessation -0.06 0.58 -0.06
  Duration 0.06 0.57 0.15
  Number of rainy days 0.07 0.51 0.07

a: Slope of linear regression (mm/decade), tau: Statistic of the Mann-Kendall test


3.6. Trend Analysis of Maximum and Minimum Temperature

The results of the trend analysis for the seasonal and annual maximum (TMax) and minimum (TMin) for Katumani station are presented in Table (7). From these results it is observed that both seasonal maximum and minimum temperature exhibited increasing trends. This affirmation can be compared with other studies in the greater Horn of Africa region that emphasized a general warming trend (Schreck and Semazzi, 2004). The results were also similar to those by Collins (2011) where seasonal maximum temperature showed a rapid warming in Kenya.

The positive trends for maximum and minimum temperature were statistically significant during the

Table 6. Observed Trends for the Seasonal and Annual Rainfall Totals for the three Stations used for the Period 1981 to 2012. The bold values represent statistically significant trends at 95% confidence level

Parameter Station
Katumani Iveti Machakos DO
MAM Rainfall
Mann Kendall tau 0.08 -0.21 -0.33
p-value 0.55 0.09 0.008
Linear Regression a 1.43 -8.98 -10.21
OND Rainfall
Mann Kendall tau -0.088 -0.36 -0.379
p-value 0.53 0.003 0.024
Linear Regression a -1.66 -14.06 -8.11
Annual Rainfall
Mann Kendall tau -0.02 -0.38 -0.427
p-value 0.91 0.002 0.0006
Linear Regression a -1.23 -24.09 -18.99

a: Slope of linear regression (mm/decade), tau: Statistic of the Mann-Kendall test



Table 7. Observed Trends for the Seasonal and Annual Maximum and Minimum Temperature for Katumani over the period 1981-2012. The bold values represent statistically significant trends at 95% confidence level

Parameter Statistical Test Seasons Annual
Tmax Tau 0.10 0.33 0.41
  p-value 0.53 0.02 0.01
   a 0.02 0.03 0.04
Tmin Tau 0.35 0.27 0.04
  p-value 0.02 0.07 0.01
   a 0.04 0.03 0.03

a: Slope of linear regression (°C/decade), tau: Statistic of the Mann-Kendall test


OND and MAM season (Table 7). From the slopes of the linear regression lines OND maximum and MAM minimum temperatures were increasing at the rate of (+) 0.03 and (+) 0.04 ° C per year, respectively.

The trend test showed an increasing trend in the annual maximum and minimum trends which was statistically significant. This indicates a warming trend in the annual temperatures. These results are consistent with other prior studies which discovered an increasing trend in the annual maximum temperature in Katumani (Njiru et al., 2010). The rate of increase of the annual maximum and minimum temperatures were 0.04°C and 0.03°C per year respectively.

Such warming trends of temperature are expected to result in the increase in the rate of photosynthesis, which in turn enhances the growth and development of any crop. On the other hand, warming lessens the number of days for the crop to attain maturity given that other vital resources are not limiting (Baviskar et al., 2017). The reduction in the growth duration will require farmers within the County to shift from planting long duration crops cultivars to the early maturing cultivars, whose grain yield potential is low (Wylie, 2008).



This study aimed at quantifying variability and trends of rainfall and temperature in Machakos County.  High degree of temporal variability linked to both MAM and OND seasons was noted in all the rainfall parameters analyzed except cessation dates. This expresses high level of unpredictability associated with planning of any cropping activity and adds to the threats for farming practice in the study area. The variability of seasonal rainfall characteristics was higher during the two growing seasons (MAM and OND).  In addition, the results of seasonal and annual rainfall amounts showed less stability at all the stations studied. Though the names of the two rainfall seasons (long and short rainfall seasons) are confusing, from the study the short rainfall season usually receive more rainfall and tend to be most reliable rainfall season compared to the long rainfall season. For that reason, the growing period during the short rainfall season is more significant for any agricultural production in the study area. Temperature depicted less variation both at seasonal and annual scale. Observed trends depicted statistically non-significant trends in the intra-seasonal rainfall characteristics. The decreasing trends of onset dates indicate a backward shift in the onset of seasonal rainfall resulting to backward shift in the timing of land preparation and planting of most crops. This makes farmers to experience uncertainties in their planting activity. Crop failure owing to early planting is as high as the possibility of total crop failure due to early cessation of seasonal rainfall, since the duration of the season is highly connected to the onset of both short and long rainfall seasons. On the other hand, seasonal and annual rainfall totals in most of stations studied showed decreasing trends. However, statistically significant trends were noted at Iveti and at Machakos DO stations. Low levels of rainfall in combination with declining condition of the soils’ would certainly have adverse impacts on crop production. Both maximum and minimum temperature showed positive trends at seasonal and annual scales. Such warming trends of temperature are expected to result in quick accumulation of heat units, which in turn enhances the growth and development of crops by reducing the number of days for crops to reach physiological maturity provided that other vital resources are not limiting.



  1. Because of high temporal variability of the intra-seasonal rainfall characteristics, the government should invest greatly in early warning systems (Famine Early Warning System Network (FEWSNET), Livestock Early Warning Systems (LEWS)) so as to aid smallholder farmers in the county in planning or adjusting their farm operations. Likewise, effectual communication of climate information and services is important for adaptation by households because communication enhances understanding and awareness. On this matter, suitable communication channels such as the utilization of local radio stations disseminating in local language may be utilized to guarantee that such climate related information and early warnings get to the anticipated farmers. In addition, based on level of awareness and access to supplementary irrigation, farmers could complement crops by irrigation during episodes of deficit rain.
  2. Due to warming trends , delayed onset and little possibility of irrigating crops, growing and breeding of rapidly maturing crop varieties that are tolerant to heat stress and drought  (sorghum, millets, pigeon pea, cowpea and green gram) may also help in cushioning farmers from impacts of climate change such as complete crop failure.  For a lifelong solution, government, policy makers and donors should build up/support irrigation amenities and water harvesting skills under variable and changing climate.
  3. Farmers should be encouraged to maximize their crop production in the OND season because of high variability of MAM seasonal rainfall characteristics compared to the OND seasonal rainfall characteristics


Competing Interests

The authors do not have any competing interests.


Authors’ Contributions

Emily Bosire: data collecting and analyzing, manuscript writing and references search; Wilson Gitau, Fredrick Karanja and Gilbert Ouma manuscript reviewing.



This research paper is part of the PhD thesis for the corresponding author at the University of Nairobi, Kenya. The authors are grateful to Kenya Meteorological Department for providing the climate data.



Adejuwon, J. O., Balogun,. E. E., and Adejuwon, S. A., 1990:  On the annual and seasonal patterns of rainfall fluctuations in Sub-Saharan West-Africa. International Journal of Climatology, 10: 839–848.

Ainsworth, E. A. and Long, S. P., 2005: What have we learned from 15 years of free-air CO2 enrichment (FACE)? A meta-analytic review of the responses of photosynthesis, canopy properties and plant production to rising CO2. New Phytologist, 165: 351-372.

Allen, R. G., Pereira, L. S., Raes, D. and Smith, M., 1998:  Crop evapotranspiration: a guideline for computing crop water requirements. FAO Irrigation and Drainage Paper No 56. FAO Water Resources, Development and Management Service, Rome, Italy. 300pp

Amissah-Arthur, A., Jagtap, S., Rosenzweig, C., 2002: Spatio-temporal effects of El Niño events on rainfall and maize yield in Kenya. International Journal of Climatolology, 22:1849–1860.

Araya, A. and Stroosnijder, L., 2011: Assessing drought risk and irrigation needs in northern Ethiopia. Agricultural and Forest Meteorolology, 151: 425–436.

Barron, J., Rockstrom, J., Gichuki, F., Hatibu, N., 2003: Dry spell analysis and maize yields for two semi-arid locations in East Africa. Agricultural and Forest Meteorolology, 117:23–37.

Baviskar,S. B.,  Gadakh, S. R.,  Haldavnekar,  P. C., Munj,  A. Y. and Raut,  R. A., 2017: Heat units and heat unit efficiency influenced by environment effect on yield and dry matter of rabi sorghum. International Journal of Chemical Studies, 5(3): 395-398.

Boroujerdy, P. S., 2008: The analysis of precipitation variation and quantiles in Iran 3rd IASME/WSEAS; International Conference on Energy & Environment, University of Cambridge UK February 23-25.

Camberlin, P. and Okoola, R. E., 2003: The onset and cessation of the “long rains” in Eastern Africa and their interannual variability. Theoretical and Applied Climatology, 75(1-2): 43-54.

Dodd, D. E. S. and Jolliffe, I. T., 2001: Early detection of the start of the wet season in semiarid tropical climates of Western Africa. International Journal of Climatology 21(10): 1251–1262.

Fuhrer, J., 2003: Agroecosystem responses to combinations of elevated CO2, ozone, and global climate change. Agriculture, Ecosystems and Environment, 97: 1–20.

Gicheru, P. T. and Ita, B. N., 1987: Detailed soil survey of the Katumani National Dryland Farming Research station farms (Machakos District). Detailed soil survey report No. D 43, 1987. Ministry of Agriculture – National Agricultural Laboratories. Kenya Soil Survey.

Gitau, W., Camberlin, P.,  Ogallo,  L. and Bosire, E., 2018: Trends of intraseasonal descriptors of wet and dry spells over equatorial eastern Africa. International Journal of Climatology, 38: 1189–1200.

Government of Kenya (GoK), 2009: 10 Years of Fight against Poverty (1999–2009): A Report of the Poverty Eradication Commission. Nairobi, Kenya: Government of Kenya. (Accessed January 20, 2014).

Government of Kenya (GoK), 2010: Draft National emerging crops policy. Ministry of Agriculture, Kilimo House, Cathedral Road, Nairobi, Kenya. ( (Accessed January 20, 2014).

Hadgu, G., Tesfaye, K., Mamo, G., and Kassa, B., 2013: Trend and variability of rainfall in Tigray, Northern Ethiopia: Analysis of meteorological data and farmers’ perception. Academically Journal Environmental Science, 1(8): 159-171.

Hansen, J. W. and Indeje, M., 2004: Linking dynamic seasonal climate forecasts with crop simulation for maize yield prediction in semiarid Kenya. Agriculture and Forest Meteorology, 125: 143–157.

Hare, F. K., 1983: Climate and Desertification. Revised analysis (WMO-UNDP) WCP-44 pp 5 20. Geneva, Switzerland

Jaagus, J., 2006: Climatic changes in Estonia during the second half of the 20th century in relationship with changes in large-scale atmospheric circulation. Theoretical and Applied Climatology, 83: 77–88.

Kansiime, M. K., Wambugu, S. K., and Shisanya, C. A., 2013:  Perceived and actual rainfall trends and variability in Eastern Uganda: Implications for community preparedness and response. Journal of Natural Sciences Research, 3(8), 179-194.

Kasei, C.N., Afuakwa, J.J., 1991: Determination of optimum planting date and growing season of maize in the northern savanna zone of Ghana. In Soil Water Balance in the Sudan Sahelian Zone, Sivakumar MVK, Wallace JS, Renard C, Giroux C (eds). IAHS Publication, No. 199. International Association of Hydrological Sciences: Wallingford.

Kenya National Bureau of Statistics (KNBS), 2010:  Kenya 2009 Population and Housing Census .

Kihupi, N. I., Tarimo, A. K. P. R., and Dihenga, H. O., 2007: Spatial and temporal variation of growing season characteristics in Tanzania. Journal of the Geographical Association of Tanzania, 32: 33 – 49.

Kimball, B.A., Kobayashi, K. and Bindi M., 2002: Responses of agricultural crops to free-air CO2 enrichment. Advances in Agronomy, 77: 293-368.

Marteau, R., Sultan, B., Alhassane, A., Baron, C. and Traoré, S.B., 2011: The onset of the rainy season and farmers’ sowing strategy for pearl millet cultivation in Southwest Niger. Agricultural Forest Meteorology, 151(10): 1356-1369.

Modarres, R., and Silva, V.P.R., 2007: Rainfall trends in arid and semi-arid regions of Iran. Journal of Environment, 70: 344–355 Arid .

Mondal, A., Kundu, S., and Mukhopadhyay, A., 2012: Rainfall trend analysis by Mann-kendall test: a case study of Northern- Eastern part of Cuttack district, Orissa. International Journal of Geology Earth and Environmental sciences, 2:70-78.

Mugalavai, E. M., Kipkorir, E. C., Raes, D., and Rao, M. S., 2008: Analysis of rainfall onset, cessation and length of growing season for western Kenya. Agricultural and forest meteorology, 148(6): 1123-1135.

Mutai, C. C. and Ward, N. M., 2000: East African Rainfall and the Tropical Circulation/Convection on Intra-seasonal to Inter-annual Timescales. Journal of Climate, 13(22):3915-3939.

Mwadalu, R. and Mwangi, M., 2013: Potential role of sorghum in enhancing food security in semi-Arid Eastern Kenya. Journal of Applied Biosciences, 71: 5786-5799.

Njiru, E. N., Kironchi, M. N. G., Kironchi, G., Mbuvi, J. P., Nguluu, S., 2010: Analysis of climate data and the associated risks to maize production in Semi-Arid Eastern Kenya.

Odekunle, T. O., 2006: Determining rainy season onset and retreat over Nigeria from precipitation amount and number of rainy days. Theoretical and Applied Climatology, 83:193-201.

Omondi, P., Awange, J.L., Forootan, E., Ogallo, L. A., and Barakiza, R., 2013: Changes in temperature and precipitation extremes over the Greater horn of Africa region from 1961 to 2010. International Journal of Climatology, 34: 1262-1277.

Omotosho, J. B., Balogun, A. A. and Ogunjobi, K., 2000: Predicting monthly and seasonal rainfall, onset and cessation of the rainy season, in West Africa, using only surface data. International Journal of Climatology 20: 865–880.

Omoyo, N. N., Wakhungu, J., and Oteng’i, S., 2015: Effects of climate variability on maize yield in the arid and semi-arid lands of lower eastern Kenya. Agriculture and Food Security, 4:8.

Opiyo, F., Nyangito , M., Wasonga, O.V.,  and Omondi, P., 2014: Trend Analysis of Rainfall and Temperature Variability in Arid Environment of Turkana, Kenya. Environmental Research Journal, 8(2): 30-43.

Partal, T., and Kahya, E., 2006: Trend analysis in Turkish precipitation data. Hydrological Process, 20: 2011-2026.

Prasad, P. V. V., Boote, K. J. and Allen,L. H. , 2006: Adverse high temperature effects on pollen viability, seed-set, seed yield and harvest index of grain-sorghum [Sorghum bicolor (L.) Moench] are more severe at elevated carbon dioxide due to high tissue temperature. Agricultural and Forest Meteorology, 139: 237-251.

Recha, C. W., Makokha, G. L. Traore, P. S., Shisanya, C., Lodoun, T. and Sako, A., 2011:  Determination of seasonal rainfall variability, onset and cessation in semi-arid Tharaka district, Kenya. Theoretical and Applied Climatology, 108:479–494 .

Robel, E., DeGiuli, C., Schoof, and Tziperman, E., 2013: Dynamics of ice stream temporal variability: Modes, scales,and hysteresis.

Schoper, J. B., Lambert, R. J., and Vasilas, B. L., 1986: Maize pollen viability and ear receptivity under water and high temperature stress. Crop science, 26: 1029-1033.

Schreck, C. J. and Semazzi, F. H. M., 2004: Variability of the recent climate of eastern Africa. International Journal of Climatology, 24: 681-701.

Shadmani, M., Marofi, S., Roknian, M., 2012: Trend analysis in reference evapotranspiration using Mann-Kendall and Spearman’s rho tests in arid regions of Iran. Water Resources Management, 26: 211–224.

Segele, Z. T. and Lamb,P. J., 2005: Characterization and variability of Kiremt rainy season over Ethiopia. National Meteorological Services Agency of Ethiopia, Addis Ababa, Ethiopia and Cooperative Institute for Mesoscale Meteorological Studies and School of Meteorology, Norman, OK, USA . Meteorology and Atmospheric Physics, 89: 153–180.

Seleshi, Y. and Zanke, U., 2004: Recent changes in rainfall and rainy days in Ethiopia. International Journal of Climatology, 24: 973–983.

Sivakumar, M. V. K., 1987: Agro-climatic aspects of rain-fed agriculture in the Sudano- Sahelian Zone. In: ICRISAT (ed) Soil, crops and water management systems for rain-fed agriculture in the Sudano-Sahelian zone. Proceedings of an international workshop, Niamey, Niger, 11–16 January 1987, pp 17–38.

Sivakumar, M. V. K. 1988: Predicting rainy season potential from the onset of rains in southern Sahelian and Sudanian climatic zones of West Africa. Agricultural and Forest Meteorology, 42: 295-305.

Shisanya, C. A., Recha, C. and Anyamba, A., 2011: Rainfall Variability and Its Impact on Normalized DifferenceVegetation Index in Arid and Semi-Arid Lands of Kenya. International Journal of Geosciences, 2: 36-47.

Stern, R., Dennett, M., and Dale, I., 1982: Analyzing daily rainfall measurements to give agronomically useful results. Experimental Agriculture, 18:223-236.

Stern, R., Knock, J., Rijks, D. and Dale, I., 2003: INSTAT Climatic Guide. 398pp.

Tabari, H., Somee, R. S., and Zadeh, M. R., 2011: Testing for long term trends in climatic variables in Iran. Atmospheric Research, 100: 132-140.

Tadross, M. A., Suarez, P., Lotsch, A, Hachigonta, S., Mdoka, M., Unganai, L., Lucio, F., Kamdonyo, D. and Muchinda, M., 2009: Growing-season rainfall and scenarios of future change in southeast Africa: implications for cultivating maize. Climate Research, 40: 147- 161.

Tamiru, S., Tesfaye, K. and Mamo, Girma 2015:  Analysis of Rainfall and Temperature Variability to Guide Sorghum (Sorghum Bicolar) Production in Miesso Areas, Eastern Ethiopia.   International Journal of Sustainable Agricultural Research, 2(1): 1-11.

Tsefaye, K., and Walker, S., 2004: Matching of the crop and environment for optimal water use: the case of Ethiopia. Physics and Chemistry of the Earth, 29:1061-1067.

Tilahun, K. 2006:  Analysis of rainfall climate and evapo-transpiration in arid and semi-arid regions of Ethiopia using data over the last half a century. Journal of Arid Environments, 64(3): 474-487.

Wagesho, N., Goel, N. K., and Jain, M. K., 2013: Temporal and spatial variability of annual and seasonal rainfall over Ethiopia. Hydrological Sciences Journal, 58: 354-373.

Wamari, J. O., Sijali, V. I., Kheng, L. H., Miriti, J. M. and Obutiatia, E. A., 2012: Use of Aqua crop model to predict maize yields under depleted rainfall and elevated temperature in a semi-arid environment. Journal of Meteorology and Related Sciences, 6: 23-32.

Wylie, P., 2008: Managing sorghum for high yields: A blueprint for doubling sorghum production. Grains Research and Development Corporation, Canberra, 24pp.

Yenigun, K., Gumus, V. and Bulut, H., 2008:  Trends in stream flow of the Euphrates basin, Turkey. Water Management, 161: 189 – 198.