Assessing Rainfall and Temperature Changes in Semi-Arid Areas of Tanzania

This paper examines the variability of rainfall and temperature in Igunga and Kishapu Districts using time series data (1985 to 2016) from Tanzania Meteorological Agency. The regression analysis results show rainfall variability of R = 0.096 in Igunga and R = 0.186 in Kishapu which implies that about 0.96% and 1.86% of the changes in rainfall across the districts are associated with changes in weather variables. A considerable change of amount of rains was evident in Igunga than in Kishapu District. In both districts there was a change of months with the most rains. Generally rainfall showed a decreasing trend in both districts. The paper also examined temperature trends in the two districts; the findings showed an increasing trend throughout October in both districts. From this point of view, higher temperatures can increase evapo-transpiration that in turn can have an effect on moisture for the crops adversely affecting pasture productivity for livestock, and leading to a shortage of water for both crops and livestock. Annual rainfall variability trends, however, increased indicating that annual variability was somewhat a common feature in the study districts. So, districts efforts should be directed towards the support of crop and livestock adjustments in order to buffer impacts of rainfall and temperature variability during critical periods for growing of crops and pastures.

droughts in Tanzania, studies show that the frequency of droughts has increased over the past few decades, especially in semi-arid areas such as Dodoma, Shinyanga, Singida, Tabora and some parts of Arusha and Iringa [2] [5] [6]. Rainfall variability data analysed for the period between 1974 and 2005 in semi-arid of Shinyanga Rural District in Tanzania, reported no significant decrease over time. However, decreasing measured rainfall and increasing temperature for the period between 1992 and 2007 were reported in Manyoni, another semi-arid area in Tanzania [5] [7].
Climate studies indicate that mean temperatures and precipitation in the country have changed over time [8]- [13]. The Fifth Report of the Intergovernmental Panel on Climate Change (IPCC) on climate variability in Tanzania provides more evidence on the occurrence of the phenomenon than previous reports [14]. The average annual rainfall of Tanzania shows a very high level of variability over the past years [15]. Literature reveals a high degree of agreement that climate variability and change have already happened, and that they are global phenomena [12] [13] [16] [17]. The proponents of the phenomena are of the view that rainfall is decreasing, while temperature is increasing over time.
However, they fail to explain seasonal variability particularly within crop growing seasons over time. Some scholars are of the view that climate variability is not new in semi-arid regions; and that such variability in climate has been affecting smallholder farmers and pastoralists for many decades [18] [19] [20] [21] [22]. According to [6] for example, inter-annual variability of rainfall and temperature in Tanzania is common. Frequent dry spells have resulted in reduced crop yields and increased food shortages leading to food insecurity [5]. Furthermore, annual rainfall data analysis shows a decreasing trend at the rate of 3.3% per decade, more so in southern Tanzania, while the mean annual temperature has increased by 0.23˚C per decade during the period between 1960 and 2003 [23]. Both day time and night time temperatures show an increasing trend, particularly during January and February; but nighttime temperatures reveal an increasing trend at 19.8% per year relative to day time temperature, which increased at 13.6% per year between 1960 and 2003 [24].

Description of the Study Areas
Igunga and Kishapu districts were selected purposively for the study by being situated in semi-arid regions. Igunga District is located in Tabora Region and lies between latitudes 3˚51'S and 4˚48'S of Equator and longitudes 33˚22'E and 34˚8'E of Greenwich ( Figure 1). Igunga District [28]   mm to 700 mm per annum. The rainfall season spans the period from November to April. The southern and south western parts of the district get more rain than the northern and north eastern parts [28]. About three-fifth of the district's population cultivate cotton and sunflower, which are the main cash crops.
Kishapu District covers an area of 9226 km 2 , and lies between longitudes 36˚30'E and 33˚30'E and latitudes 3˚45'S and 5˚00'S. The total population is 272,999 and 35,500 households with an average household size of 8 people.
89.5% of the people live in rural areas [30], 75% own livestock and 39% practice mixed crop and livestock farming system [31]. The mean annual rainfall in Kishapu lies between 600 mm and 800 mm and surface temperature ranges from 16˚C in June to 30˚C in October. The area lies at an altitude 1000 -1200 m above sea level. The highest temperature is experienced in October, just before the onset of rainfall. A dry spell normally occurs between mid-January and February [31]. The rainfall regime in both districts is unimodal, which starts in November and ends in April [32]. The major cash crops are cotton, sunflower, groundnuts, green gram onions, pigeon peas and cowpeas. Livestock comprise cattle, goats, sheep and donkeys. The major food crops grown are sweet potatoes, sorghum and maize. Other economic activities are mining and sunflower oil processing.

Materials and Methods
Daily and monthly rainfall data for Igunga District from 1985 to 2016 (31 years) and Climatic data such as rainfall and temperature were analysed using Excel to generate tables and graphs. SPSS was employed to generate means and variances, skewness and kurtosis which were used to assess the changes in climate [33].
Maximum temperatures are recorded during the day time and thus, play a critical role in controlling evapo-transpiration and drying up of water bodies [23].
On the other hand, minimum temperatures are obtained during the night. For rainfall variability, the analysis focused on annual and seasonal variability trends because variabilty can reveal dry and wet periods over time .
During data analysis, both variability and monthly means were computed.
The variability was computed as a deviation from a long-term (annual) mean.
The rainfall and temperature annual variability are presented in (Tables 1-4).
The hypothesis that the study districts did not experience significant increasing trends in inter-annual rainfall variability for a period between 1985 and 2016 was tested using a p-value at 5% level of significance. To analyze meteorological data, R-Statistical package was used to perform simple regression analysis for rainfall and temperature data. Several studies have also used in this approach in the past to analyse evidence of climate change [33] [34].
The dependent variable [Y (j)] was the physical factor (mean rainfall, mean minimum temperature, mean maximum temperature) and independent va-

Trends in Monthly Rainfall and Temperature in Kishapu and Igunga Districts
Rainfall is a major climate parameter with the highest degree of spatial and  indicates that there were significant variations in rainfall between the months.
These results are in line with a study conducted by [36] when analyzing global monthly mean precipitation. It is important to note that variability of rainfall patterns leads to a redistribution of rainfall. Further, the precipitation had a decreasing trend over time, this led to an increase in extreme droughts and shortages of water during April to September. Therefore, an adaptation response to their perceptions would be appropriate and helpful to government efforts to avoid potential agricultural losses.  has had a consistent falling pattern. Consistent rainfall patterns suggest that it rained at the time farmers and agro-pastoralists expected it to rain [23]. Nevertheless, rainfall variability, punctuating the general trend, can be beneficial when the increasing annual rainfall trend exceeds the range for semi-arid areas. The common situation for Igunga and Kishapu districts is that there was seasonal rainfall variability in both districts and the crop growing season normally began November and went on to the end of January.
In Kishapu District, temperature on the other hand, shows little variation with increasing trends in terms of minimum and maximum temperature. Table 2 shows that October and December are the two months that indicate the highest temperature, implying that it was the hottest months throughout the period be- In Igunga District, the highest daytime temperature occurred in the months of October, November and December, while the lowest occurred in the months of November and January ranging from 24.0˚C to 33.6˚C (Table 4). Results showed that the mean maximum temperature for Igunga district in Tabora region, was 38.67˚C in February over the period between 1985 and 2016. This implies that the periods of highest temperature were also dry periods. It also implies that the increase in maximum temperature was accompanied by decreasing amount of rainfall in Igunga District in particular (Table 1). Maximum temperature is normally recorded during the day time, thus its increase most probably reduces soil moisture through evaporation and evapotranspiration, which in turn negatively affects crop and pasture development. This has been also reported in Singida District, Tanzania by [23].
The findings in Table 1 and  (Figure 2), maximum temperatures were increasing ( Figure   3). However, in the literature report [9] the semi-arid areas have rainfall which ranges between 200 and 800 mm per annum [9]. adversely affecting smallholder farmers and agro-pastoralists whose livelihoods largely depends on the rain-fed farming system. Seasonal variability of rainfall and temperature was similarly reported by [24] and [35] at the national level in Tanzania. Figure 3 in Kishapu the Inter-annual variability revealed R 2 of 0.078 for annual temperature. This translates into moderate temperature variability of 7.8% related to changes in time. For Igunga, the R 2 was 0.1724, which translates into strong temperature variability of 17.24% over time. However, in Kishapu change in time explained temperature variability by 7.8% and in Igunga temperature variability by 17.24%. This suggests strong annual trends for temperature variability for the period under concern. Table 5, the p-value was 0.007 (P > 0.005 ) implying that the change was not statistically significant at 5% level of significance. The R 2 was 0.1195, translating into 0.1% of the inter-annual temperature variability that was associated with change in time between 1985-2016.
Further, the correlation between temperature and time is significant as shown by the significance of Pearson Correlation test at the 1% probability level ( Table  5). The result indicates that there is no statistical significant trend in the aveage temperature over time in the study areas. The R-squared statistic also indicates a weak relationship between the variables average temperature and time (season).

Rainfall and Temperature Variation during Growing Season in Kishapu and Igunga
Annual crop yields for five major crops: maize, sorghum, cotton, sunflower and planting, weeding and harvesting, which are linked directly to rainfall events.
The period of planting is the time of greatest uncertainty and risk because planting decisions largely assemble the other activities and may not easily be implemented after certain time.
Results in Table 1 [23] in a study of climate variability in Shinyanga and Singida. Intra-seasonal factors, such as timing of the onset of first rains affecting the crop planting regimes [39], the distribution and length of rain during the growing season [40] and effectiveness of rains in each precipitation event [39], are the real criteria that determine the effectiveness and success of farming. Furthermore, in Figure 5, findings revealed that the linear rainfall trend in Kishapu was decreasing over time like Igunga rainfall anomaly trend. However, the R 2 was 0.00051 and 0.0237 in October and November respectively of rainfall variability and that variability were higher at Kishapu station compared to Igunga rainfall station. According to [23], climate variability and change occurred through increased rainfall unpredictability; increased frequency of drought; increased duration of dry spells and warmings. Such climatic changes increased farmers and agro-pastoralist mobility in search for water and pasture for livestock and also brought in changes of crop varieties to be grown.

Conclusions and Recommendations
This paper examined trends in seasonal variability of rainfall and temperature in Igunga and Kishapu districts. Statistically the temperature showed an increased trend while precipitation was characterized by large inter-annual variability and a decreased amount during rainy season. Low levels of precipitation were found from June to September while high levels of precipitation occurred in December.
These trends were confirmed by local perceptions, found from qualitative study.
Specifically the study assessed trends in the monthly temperature and rainfall variability. Based on the results and discussions, the paper concludes that, the meteorological data show increasing average temperatures and decreasing annual rainfall patterns for the past 31 years examined (1985 and 2016). The analysis indicates that the onset and end of rainfall during the growing period has become more erratic and unpredictable since 1985. Annual rainfall variability trends, however, increased minimally indicating that annual variability was somewhat a common feature in the study districts. The analysis indicated a shift on the onset of long rains from October/November to December/January with shortening of rainfall period and increased frequency of drought. As far as temperature was concerned, it had an increasing trend, and also the maximum temperatures showed considerable variability, a typical characteristic of semi-arid areas. Further, whereas annual variability showed rainfall variability in both districts, seasonal trends showed considerable rainfall and temperature variability within and between seasons. When analyzing climate variability, it is therefore important to consider both seasonal and inter-annual variability so as to have a clear understanding regarding the degree of climate variability and change.
Based on the results and conclusion, the paper recommends that the local government authorities and other independent initiatives should be directed towards the support of crop and livestock adjustments in order to cushion the impacts of rainfall and temperature variability during critical periods for growing of crops and pastures. Capacity building to the farmers and livestock keepers through training is still needed especially on appropriate adaptation strategies to climate change. Such strategies can be applied to address risks caused by insufficient rainfall, higher day time temperature, as well as rainfall and temperature variability, which have been demonstrated in this paper. Climate change policies should be actively incorporated in all developmental plans. For example water harvesting can be promoted for irrigation to supplement rain-fed agriculture and for consumption by people and livestock in both districts; as well as setting up policies for animal breeding and crop varieties in the study areas for providing highly tolerant breeds, pastures and crop varieties tolerant to drought.