## Abstract

Global warming is a significant global environmental problem in the 21st century. The problem is high in developing countries, particularly sub-Saharan countries in which the majority of the population live on rainfed agriculture. The present study aimed to undertake spatiotemporal analysis of seasonal and annual rainfall and temperature and its implications. The MK test, Sen's slope and precipitation concentration index (PCI) were applied. Finally, Pearson correlation analysis between climatic variables and crop production was analysed. The Mann–Kendall test results showed that the annual and seasonal rainfall trend was highly variable. The minimum and maximum temperatures have increased by 0.8 and 1.1 °C/year, respectively. Based on PCI results, rainfall during the summer and spring seasons is moderately distributed as compared to annual and winter season rainfall. Based on these observations, the rainfall pattern and distribution of the area could be classified as irregular and erratic distribution. Results of correlation analysis between monthly and seasonal rainfall with crop production were insufficient to conclude the impact of rainfall and temperature on crop production. In view of this, the incidence of food shortage is a common occurrence. Therefore, depending on the historical trend of rainfall variability and prolonged temperature increase, appropriate coping and adaptation strategies need to be encouraged.

## INTRODUCTION

Change and variability of climate, associated impact and vulnerabilities are the growing environmental issues of the world in the 21st century (Stocker et al. 2013; Pachauri et al. 2014). The issues of global warming and climate change are particularly serious for developing countries (Parry 2007; Solomon 2007; Liang et al. 2011; Pachauri et al. 2014). Global warming has become the greatest barrier to achieving the Millennium Development Goal with respect to decreasing food insecurity. Many countries of the world, particularly sub-Saharan African countries, are already affected by the variability of climatic conditions (Conway & Schipper 2011; Kløve et al. 2014). For instance, the variability, intensity and duration of temperature and rainfall affect crop production, especially for developing countries, particularly sub-Saharan countries in which the livelihoods of the population are dependent on subsistence and rainfed farming (Hulme et al. 2005; Batisani & Yarnal 2010; Randell & Gray 2016). For most developing countries of the world, agriculture is the basis of the economy. For more than 70% of the world's population, the primary source of their livelihood has originated from weather sensitive agriculture (Suarez et al. 2012; Fazzini et al. 2015). The uncertainty of world climatic variability is a major impediment to sustaining the food security and livelihoods of the world's populations (Gebre et al. 2013; Irannezhad et al. 2014).

According to a report made by the Intergovernmental Panel on Climate Change (Parry 2007; Pachauri et al. 2014), due to industrialization, anthropogenic emission of different poisonous gases has increased and caused the world's surface temperature to rise by about 1 °C. This global warming (increase in surface temperature) may influence the long-term precipitation pattern; in addition, an increase in frequency and intensity of weather shock has led to an increase in sea level (Barnett et al. 2005). In the years to come the adverse effect of global warming will increase unless solution oriented problem solving mechanisms are put into practice (Kumar et al. 2013; Pingale et al. 2014).

The analysis of vulnerability related to climate changes in Ethiopia implies that in the coming decades climate variability and volatility will threaten the social and economic order (damage to natural resources, agricultural productivity, water resources and ecosystems); therefore, the incidence and intensity of drought and famine occurrence is likely to increase. In the last few decades, incidence of climate change related hazards have manifested in the form of recurrent drought, erosive rain, rainfall variability and flood events (Kenabatho et al. 2012; Meshesha et al. 2016).

Annual and seasonal rainfall and temperature are influenced by the variability of the Intertropical Convergence Zone (ITCZ) which causes interannual rainfall variability over Ethiopia. Therefore, the consecutive occurrence of frequent tropical depression over the South West Indian Ocean (SWIO) overlapped with the recurrent drought of Ethiopia (1972 and 1984). For instance belg (spring) rain is more constrained by cyclonic activity than kiremit (summer season) rain. When the tropical depression is observed in the SWIO, the daily rainfall is significantly decreased. For the kiremit (summer season), the main rainfall source is the northward oscillation of ITCZ and the development of high-pressure systems along the southern Atlantic as well as South Indian Oceans.

Over the past decades, the minimum and maximum average temperature of Ethiopia have increased by about 0.25 and 0.1 °C, respectively. Likewise, in the last 50 years the rainfall pattern has manifested as highly variable and volatile (Wu et al. 2016). Climatic variability in the past has been increasing and from the trends suggested in different studies, may further increase in the near future, putting urgent emphasis on how the community perceives the extent of climate change in order to design coping and adaptation strategies (Belay et al. 2005). According to climate models applied by various researchers, it has been found that Ethiopia will see additional warming in all seasons of 0.7–2.3 °C by the 2020s and 1.4–2.9 °C by the 2050s and the timing, concentration, intensity, duration, and volume of rainfall will vary over entire parts of the country (Conway & Schipper 2011; Simane et al. 2012). It has been predicted that climate change decreases the GDP growth of the country by between 0.5 and 2.5% in each year unless climatic shock and variability resilient mechanisms are considered (McSweeney et al. 2010; Simane et al. 2012). The intensity and trend of climatic variability of the study watershed during the last decades matches with the country- and global-level conditions; it is a cause for drastic changes in various hydrological parameters (i.e. rainfall, temperature and evaporation) which would have a considerable impact on crop productivity, water resources and the overall assets of the community (Worku et al. 2017a, 2017b).

Therefore, long-term analysis of climatic trends has been used to characterize the situations (Singh et al. 2008; Subash et al. 2011; Jain & Kumar 2012; Suryavanshi et al. 2014; Kishore et al. 2016). Many researchers have undertaken trend analysis studies of the climate in some other parts of Ethiopia (Addisu et al. 2015; Wagesho & Yohannes 2016). Most of the studies about rainfall and temperature characteristics are limited by short-term and long-term time series available for most parts of the regions. Some of the studies conducted are based on areal averages of spatial climatic variability (Seleshi & Demaree 1995; Osman & Sauerborn 2001). Other studies have focused on very limited stations and arrived at a conclusion regarding the characteristics of spatial climatic variability for entire regions (Gamachu 1988; Meze-Hausken 2004). Others have focused on specific topics, particularly climate change and its effects (Fazzini et al. 2015). Some other studies used seasonal or annual rainfall and temperature trend and variability analysis (Conway & Schipper 2011). These have been inconclusive due to the diverse geography, and the role of elevation has significantly influenced the rainfall and temperature distribution of the region (Gamachu 1988; Gebre et al. 2013). However, local farmers evaluate climatic variability in relation to their crop productivity. Such studies ignored the localized trends of rainfall and temperature, particularly in most highlands of Ethiopia. Additionally, studies of rainfall and temperature variations in larger areas would in general be of little use for local level agricultural production (Gebre et al. 2013).

Therefore, clear information about the annual and seasonal rainfall distribution is highly important for policy planners and local users. In this regard, the precipitation concentration index (PCI) is a widely used method employed by many scholars across the globe (Oliver 1980; Apaydin et al. 2006; Rashid et al. 2015). Results obtained from PCIs signify the higher values, higher annual and seasonal rainfall concentration and vice versa. Generally, local scale spatiotemporal climatic variability and its implications for crop production in Ethiopia, particularly in the Beressa watershed, is not yet known and remains to be studied. Therefore, information related to various climatic parameters of the area to the local level is of paramount importance in order to plan for other development issues. Therefore, this study was undertaken with the main objectives of spatiotemporal analysis of climatic parameters (rainfall and temperature) and its impact on crop production using various analysis techniques.

## STUDY AREA DESCRIPTIONS

The study watershed lies between 39° 37′E–39° 32′E and 9 °40′N–9 °41′N. In administrative terms, it is located in Basona Worena District, in the North Showa zone of Amhara regional state (Figure 1), situated 180 km northeast of the capital city, Addis Ababa. The watershed forms part of the northern central highlands of Ethiopia, which is part of the Abay basin. The area is characterized by diverse topographic conditions such as mountainous and dissected terrain with steep slopes. The elevation ranges from 2,747 to 3,674 m a.s.l. The annual average temperature of the area is 19.7 °C. The annual minimum and maximum rainfall is 698.5 and 1083.3 mm, respectively. The most common types of soil are Cambisols (locally called Abolse), Vertisols (Merere), Andosols, Fluvisols and Regosols. Mixed crop-livestock is the production system of the area and is perhaps the only source of livelihood for the majority of the population. Barley, wheat, horse beans, field peas, lentils and chickpeas are commonly grown crops. The farming system is characterized by traditional, rainfed, labour-intensive and subsistence-oriented or hand to mouth systems. Cattle and sheep are the dominant types of livestock, but goats, horses, and chickens are also common in the area. Because of the rainfall-dependent farming practice, farmers are always worried about the duration and intensity of rainfall.

Figure 1

Location of the study area.

Figure 1

Location of the study area.

### Data used

The rainfall and temperature daily records over 35 years (1980–2014) for the Beressa watershed were obtained from the National Meteorological Service Agency of Ethiopia from seven stations; hence rainfall on a monthly, seasonal and annual basis were derived from the daily data. The long-term rainfall trend was assessed monthly, seasonally — i.e. kiremit season (June–September), belg season (March–May), bega season (October–February) — and annually for all subdivisions, while the long-term trend of temperature was assessed for annual average, annual minimum and maximum temperature. The available data for crop production (Q/ha) over 18 years (1997–2014) for the major crops such as barley, wheat, beans, peas, lentils and chickpeas were obtained from the district office of Agriculture and Central Statistical Authority. In this study, to manage the data quality, data series were plotted in order to identify the outliers. After visual identification of the outliers, each of the values was obtained using a normal ratio technique. Additionally, serial correlation was tested. Therefore, there were no gaps in the data series.

## METHODOLOGY

### Precipitation concentration index

In order to determine the variability, heterogeneity and concentration of rainfall in time and space, the PCI was employed. The guidelines for interpretation are presented in Table 1.

Table 1

Interpretation of PCI results

PCI valueInterpretation
<10 Uniform distribution of precipitation (lLow concentration)
11–16 Moderate distribution of precipitation (moderate concentration)
16–20 Irregular distribution of precipitation
>20 Strong irregularity of precipitation distribution
PCI valueInterpretation
<10 Uniform distribution of precipitation (lLow concentration)
11–16 Moderate distribution of precipitation (moderate concentration)
16–20 Irregular distribution of precipitation
>20 Strong irregularity of precipitation distribution
The PCI was used as an indicator of concentration and variability of rainfall was obtained as follows (Oliver 1980; Murugan et al. 2008; Iskander et al. 2014):
(1)
where Pi= the amount of rainfall in ith months; ∑ = summation of precipitation over 12 months.

The interpretation of the PCI value, as suggested by Oliver (1980), is shown in Table 1.

According to Michiels et al. (1992), the calculation and evaluation of the PCI has been carried out based on seasonal and annual records. For seasonal PCI, the Ethiopian seasons are divided into three seasons. Therefore, the PCI was obtained on a seasonal scale, namely belg season (spring, March–May), bega season (winter, October–February) and kiremit season (summer, June–September), using the following formula:
(2)
(3)
(4)

### Trend analysis

Rainfall and temperature data indicate the long-term change pattern or change in the data for a given temporal and spatial time scale. Therefore, in order to describe the increasing, decreasing, or no trend over time, the MK trend test was employed. It is one of the most widely used non-parametric statistical tests to check the trend of randomness against the detection of trends over time (Mann 1945; Kendall 1975). This statistical test is a popular and important tool in detecting the trend used by many other scholars for related applications (Hirsch et al. 1982; Burn & Elnur 2002; Yue et al. 2002; Suryavanshi et al. 2014; Mondal et al. 2015; Pingale et al. 2016).

The trends derived from the Mann–Kendall (S) statistic test are used to detect normalized p-value for the significant test. Sequential order data should be required to perform the MK trend analysis. First, decide the sign of sample results (difference between series successive samples). Xj–Xk is used to display the function resulting in the values 1, 0, or −1 on the basis of Xj–Xk, in which j>k, and the function is computed using the following equations:
(5)
where:
(6)
where Xj and Xk are the value of sequential rainfall and/or temperature in the month of j and k (j > k) respectively. While the positive value is used to show an increasing trend, the negative value reveals a decreasing trend.

Let X1, X2, X3…………. Xn signify ‘n’ data points (for monthly as well as annual), in which Xj signifies the data point at time of j.

These test statistics represent the difference between positive and negative difference. According to Helsel & Hirsch (1992), for the samples which are large in number (N > 10), the statistical test is employed using normal distribution with mean E(S) as well as variance Var(S), in which E(S) = 0:
(7)
(8)
where N = number of tied observations; m is the number of tied groups (in which a set of data samples has similar values in the observations); t is the data points tied in the kth group.
When the value of S is a positive one, it confirms signs of an upward trend, whereas when there is a negative value, a downward trend is specified. A standardized test statistic of Z was obtained by the following formula:
(9)
Z0 is a null hypothesis which signifies that the trend is not significant; it is recognized if Z statistics are insignificant statistically (−Zα/2 < Z < Zα/2), where Zα/2 is standard normal distribution. For the purpose of this study, two different levels of significance were considered (at 1 and 5% level of significance).

#### Sen's slope estimators

In the process of determining the trend magnitude and variability of rainfall and temperature throughout long-term time series, Sen's slope estimators was a widely used method (YUE et al. 2003; Partal & Kahya 2006), using the following equation:
(10)
For i = 1, 2, 3,…N
where the values of Xj and Xk are used to represent the data value at time j and k respectively, in which j is greater than k, the median of the N values of Ti termed as an estimator of Sen's slope, the value can be calculated by the following formula:
(11)
(12)
where if the β value is positive it shows an upward trend in reverse, while the negative value indicates a decreasing trend of climatic variables in the long-time series.
The percentage change over a period of time can be obtained from Sen's median slope and mean by assuming the linear trend in the long-term series using the following formula:
(13)
where β is Sen's slope.

### Moving average

In statistical terms, the moving average is also known as running average, used in order to explore a set of various data by creating an average value of various subsets for a data set. The moving average is possibly acquired by considering the initial subset average. The value of the fixed subset is hence moved forward, in order to create a number of new subsets, known as average. This process is repeated for the whole data sequence. The line connecting the fixed average is known as averagely moving. Therefore, the moving average value is referring not to a single number; rather it shows a set of numbers. Moving average rainfall and temperature can be obtained by using the following equation:
(14)
where Pt is the actual values in the given time t; N refers to the number of time periods in the moving averages.

### Rainfall and temperature spatial analysis

Inverse distance weighted interpolation methods (IDW) have been used in order to analyse annual and seasonal rainfall and temperature. According to Gemmer et al. (2004), the theory of interpolation surface would likely be more influenced by nearby stations and vice versa. Therefore, average annual and seasonal rainfall were interpolated using ArcGIS 10.2. The value of IDW (Yavuz & Erdoğan 2012) was obtained by the following general formula:
(15)
where the prediction value of Z(S0) is used for the location of S0;N is the number of measured points encompassing prediction location; λi is the weight, which is alloted for each of the measured points; and Z(Si) is used to indicate observed values in the Si location.
The value of weighting is referred to as the function of the inverse distance. To determine the weighting the following general formula was employed (Yavuz & Erdoğan 2012):
(16)
The IDW technique needs the choice of the parameters and a search radius. The power parameter (p) controls measured value significance on the value of interpolation on the basis of the distance of output points (d) (Yavuz & Erdoğan 2012). A selection of comparatively larger power parameters confirms a higher degree of local impact, hence giving the output surface in further detail. Finally, in order to obtain better insights on crop production, seasonal and annual rainfall and temperature relationships with crop production, the Pearson correlation analysis was used.

## RESULTS AND DISCUSSION

### Seasonal rainfall pattern

Over the last three and a half decades, the total annual rainfall of the Beressa watershed has varied from 698.5 to 1,100 mm. Even though the rainfall indicates seasonal and inter-annual variability, the area is characterized by a bimodal rainfall regime, with maximum rainfall concentration during kiremit (summer) season, which extends from June to September. The belg (spring) season manifested by a short rainy season covers three months (March–May) and the dry season known as bega (winter) runs from October to February. During these seasons, rainfall is more highly variable than the main rainy season of the area.

The results from the coefficient of variations shown in Table 2 revealed that in comparison with the kiremit rainfall season, during the bega and belg seasons rainfall varies considerably more. Elsewhere, in other parts of Ethiopia, similar conclusions are reached by Merasha (1999) and Seleshi & Zanke (2004) – that the bega and belg rainfall seasons are more highly variable than the main rainy season (kiremit season). In a study by Mekasha et al. (2014), it was concluded that a general tendency of increasing warm temperature, extreme variability and inconsistent precipitation trend was recorded in Ethiopia. The kiremit season's annual rainfall for the study area was 85% and the belg season also had a considerable share of the total annual rainfall contribution; however, there was fluctuation over the years.

Table 2

Summary of annual and seasonal rainfall, coefficient of variation and PCI

StationsAnnual
Kiremit
Belg
Bega
AverageCVPCIAverageCVPCI$AverageCVPCIAverageCVPCI DB 1,083 0.17 23.98a 740 0.277 12.13 135.9 0.471 11.31b 62.1 1.2 19.26c DBS 1,096 0.17 20.00b 188 0.18 13.20 57 0.55 12.40b 13 0.54 22.10a SH 1,012 0.51 21.00c 185 0.17 11.50 64 0.61 11.80b 15 0.78 20.50a GIN 1,089 0.21 19.50c 201 0.11 12.00 70 0.54 16.40c 28 0.91 20.90a ENW 1,097 0.13 20.70a 227 0.21 13.20 94 0.63 13.40b 18 0.66 18.60c HG 1,081 0.21 19.00c 212 0.31 11.60 87 0.50 15.00b 30 0.72 21.00a SD 1,007 0.11 20.00c 194 0.24 12.70 77 0.51 14.50b 20 0.82 19.20c StationsAnnual Kiremit Belg Bega AverageCVPCIAverageCVPCI$AverageCVPCIAverageCVPCI
DB 1,083 0.17 23.98a 740 0.277 12.13 135.9 0.471 11.31b 62.1 1.2 19.26c
DBS 1,096 0.17 20.00b 188 0.18 13.20 57 0.55 12.40b 13 0.54 22.10a
SH 1,012 0.51 21.00c 185 0.17 11.50 64 0.61 11.80b 15 0.78 20.50a
GIN 1,089 0.21 19.50c 201 0.11 12.00 70 0.54 16.40c 28 0.91 20.90a
ENW 1,097 0.13 20.70a 227 0.21 13.20 94 0.63 13.40b 18 0.66 18.60c
HG 1,081 0.21 19.00c 212 0.31 11.60 87 0.50 15.00b 30 0.72 21.00a
SD 1,007 0.11 20.00c 194 0.24 12.70 77 0.51 14.50b 20 0.82 19.20c

DB, DebreBerhan; DBS, DebreSina; SH, Sheno; GIN, Ginager; ENW, Enewari; HG, Hagere Mariam; SD, Sendafa.

aStrongly irregular.

bModerate.

cIrregular.

The calculated PCI for seasonal as well as inter-annual rainfall distribution for the spatiotemporal time series is shown in Table 2. In the rainfall distribution during belg and kiremit it was found that there was a moderate concentration of precipitation throughout the seasons, which shows that there is no uniform distribution, whereas during the bega season a significant change in the PCI was shown, thus the concentration of precipitation is increasing and rainfall has become more erratic. In line with Rashid et al. (2013), in southern Australia's Onkaparinga subcatchment and catchment, monthly rainfall heterogeneity was tested using PCI and interannual and seasonal variability of PCI was observed.

Likewise, as presented in Table 2, the distribution of annual rainfall has shown to be very low with high PCI. Therefore, the interannual rainfall distribution was very erratic.

### Trend and variability of seasonal and annual rainfall

As shown in Figure 2, during the period 1980–2014 the seasonal rainfall trend of the Beressa watershed for the kiremit season shows less rainfall variability throughout the study periods. In contrast to the kiremit season, the five years' moving average annual rainfall of the bega season during the period 1980–2014 was highly variable. However, in the belg season during the period 1980–2014 the five years' average moving annual and seasonal rainfall was considerably variable. The variation for the belg season is presented in Figure 2.

Figure 2

Five years moving average rainfall (1980–2014). NB: Kiremit: Summer; Belg: Spring; Bega: winter.

Figure 2

Five years moving average rainfall (1980–2014). NB: Kiremit: Summer; Belg: Spring; Bega: winter.

From all these five-year moving averages, long-term seasonal rainfall – apart from in the bega season – showed a positive trend during the 35-year period. During the belg (small rainy) season the subdivision indicates a slightly increasing rainfall trend and the bega season (dry season) shows a negative trend, as already presented in Figure 2. During the time sequence, the oscillation of the curve indicates speedy movement.

The average annual aerial rainfall of the Beressa watershed is 891 mm, with a coefficient variation of 30.6% and standard deviation of 227 mm. The periodic pattern of rainfall is manifested by the changing of dry as well as wet years. After total observation of the 35-year period, a record 16 years (45.7%) were lower than the total annual rainfall of the area. On the other hand, 19 years (54.3%) recorded more than the annual average rainfall. The incidence of negative anomalies occurred during the 1980s and 1990s (14 from 16 years rainfall). In the years between 1981 and 1984, the annual total rainfall was far lower than the mean long-term rainfall. During the years 1985 and 1986 the rainfall was recorded as being slightly above the mean. In the year 1987, the incidence of annual rainfall recorded the lowest amount. Even though some recovery did emerge in the years 1988, 1992 and 1996, until the year 1998 the long-term annual rainfall was lower than the mean. However, after 1999 and onwards, recovery in the long-term average rainfall emerged higher than the average mean, except for the drier conditions in 2002 and 2013 which were lower than the long-term mean. In line with the study by Wu et al. (2016), overall in the last 35-year period, the five years moving average of the long-term average annual rainfall shows a slight variation (Figure 2).

### Annual temperature trend and variability

The time series of five years moving average minimum and maximum temperature was analysed for the period 1980–2014. Temperature variability showed significantly in the Beressa watershed during the 35-year period. Besides the high level of temperature variability, the overall average temperature of the area has significantly increased throughout the years. Before this period, the maximum temperature was 19.40 °C and the minimum temperature was 6.20 °C, with an average temperature of 12.80 °C, while the time series maximum temperature has increased to 20.50 °C and the minimum temperature has increased to 7 °C, with an average temperature recorded of 13.75 °C (Figure 3). From Figure 3, it is confirmed that the maximum temperature has continuously increased by about 1.10 °C, whereas the minimum temperature has increased by about 0.70 °C. The trend of increasing maximum temperature is stronger than the minimum temperature. Overall, the five years moving average trend of average annual temperature of the study watershed is increasing by about 0.95 °C. The present results are in agreement with Parry (2007), who stated that due to a prolonged increase in the emission of gases through human activities and expansion of industry, the surface temperature has increased by about 1 °C. Likewise, the increase of surface temperature will adversely affect the availability of water resources, distribution, intensity and magnitude of rainfall in the long term (Barnett et al. 2005).

Figure 3

Five years moving average temperature (1980–2014).

Figure 3

Five years moving average temperature (1980–2014).

The rate and variability of increasing temperature have dramatically increased, making it more difficult for local communities to foresee the intensity and magnitude of temperature even for the next few years. Another study by Di Falco et al. (2012) found that due to global climate change the eastern part of Africa, including Ethiopia, was drying out. There has been a continuous decrease in the duration and distribution of rainfall during the last 35 years. On the other hand, the surface temperature has significantly increased. Generally, as can be seen from Figure 3, there has been a high increase in overall temperature, which may result in a decrease in productivity and food insecurity.

### Mann-Kendall test statistics

According to Anderson (1942), in order to exclude the influence of serial correlation, before using MK test statistics, serial autocorrelation is tested by Lag-I autocorrelation using different levels of significance (0.01, 0.05 and 0.1%). Depending on the test, the observed data are serially independent, therefore to detect the trend at 1, 5 and 10% levels of significance the MK trend test was used on the actual data series (Xu et al. 2007; Fu et al. 2009).

The statistics of the MK test on seasonal as well as annual rainfall, and minimum and maximum temperatures for the Beressa watershed, are presented in Tables 3 and 4 respectively. The spatiotemporal rainfall and temperature distribution are presented in Figures 4 and 5 respectively. Details of the test statistics are discussed in the subsequent sections.

Table 3

Summary statistic of MKs test (Zmk), Sen's Slope estimator (β) and change in % of annual and seasonal rainfall (1980–2014)

Rainfall (mm)
Annual
Seasons
Kiremit
Belg
Bega
Stations nameZmkβ%changeZmkβ%changeZmkβ%changeZmkβ%change
DB 0.26* 0.28 1.07 0.35* 1.62 31.79 0.07** 0.40 30.00 −0.04** −0.19 −53.00
DBS −2.00 −0.99 −3.16 0.42 0.33 6.13 0.06 0.23 13.69 −1.50* −0.21 −56.40
SH 0.42 1.20 4.14 0.60 0.96 18.10 0.31 0.20 10.57 −0.50 −0.13 −29.00
GIN −3.10 −2.10 −6.74 0.30 1.02 17.70 −1.12* −0.21 −10.00 −1.20* −0.06 −7.50
ENW 0.30 1.21 3.85 0.40 1.50 23.00 0.42 0.31 11.00 −1.70* −0.19 −35.00
HG −3.50** −8.62 −27.88 0.51 1.08 17.80 0.05 0.04 1.50 −0.64 −0.11 −12.70
SD 1.50 1.04 3.62 −0.12 −0.90 −16.20 0.80 0.06 2.80 −0.55 −0.05 −8.80
Rainfall (mm)
Annual
Seasons
Kiremit
Belg
Bega
Stations nameZmkβ%changeZmkβ%changeZmkβ%changeZmkβ%change
DB 0.26* 0.28 1.07 0.35* 1.62 31.79 0.07** 0.40 30.00 −0.04** −0.19 −53.00
DBS −2.00 −0.99 −3.16 0.42 0.33 6.13 0.06 0.23 13.69 −1.50* −0.21 −56.40
SH 0.42 1.20 4.14 0.60 0.96 18.10 0.31 0.20 10.57 −0.50 −0.13 −29.00
GIN −3.10 −2.10 −6.74 0.30 1.02 17.70 −1.12* −0.21 −10.00 −1.20* −0.06 −7.50
ENW 0.30 1.21 3.85 0.40 1.50 23.00 0.42 0.31 11.00 −1.70* −0.19 −35.00
HG −3.50** −8.62 −27.88 0.51 1.08 17.80 0.05 0.04 1.50 −0.64 −0.11 −12.70
SD 1.50 1.04 3.62 −0.12 −0.90 −16.20 0.80 0.06 2.80 −0.55 −0.05 −8.80

*5% level of significance.

**10% level of significant. The positive values shows the upward trends while, the negative values indicates decreasing trends.

Table 4

Summary statistic of MKs test (Zmk), Sen's Slope estimator (β) and change in % change of mean annual, annual minimum and annual maximum temperature (1980–2014)

Stations nameTemperature (°C)
Tmean
Tmin
Tmax
Zmkβ%changeZmkβ%changeZmkβ%change
DB 0.06 0.03 7.60 0.05* 0.07 34.12 0.05** 0.064 10.95
DBS 0.13** 0.11 27.60 0.06 0.12 52.40 0.31 0.12 21.10
SH 0.92** 0.09 21.50 0.20 0.03 11.20 0.50** 0.08 13.80
GIN 0.04* 0.04 9.40 0.40* 0.005 1.90 0.62 0.023 4.00
ENW 0.06 0.05 11.50 0.23** 0.06 37.90 0.54* 0.21 37.60
HG 0.01* 0.061 13.40 0.31* 0.05 24.20 0.33* 0.12 19.90
SD 0.12** 0.14 31.30 0.07* 0.06 22.80 0.07* 0.031 5.70
Stations nameTemperature (°C)
Tmean
Tmin
Tmax
Zmkβ%changeZmkβ%changeZmkβ%change
DB 0.06 0.03 7.60 0.05* 0.07 34.12 0.05** 0.064 10.95
DBS 0.13** 0.11 27.60 0.06 0.12 52.40 0.31 0.12 21.10
SH 0.92** 0.09 21.50 0.20 0.03 11.20 0.50** 0.08 13.80
GIN 0.04* 0.04 9.40 0.40* 0.005 1.90 0.62 0.023 4.00
ENW 0.06 0.05 11.50 0.23** 0.06 37.90 0.54* 0.21 37.60
HG 0.01* 0.061 13.40 0.31* 0.05 24.20 0.33* 0.12 19.90
SD 0.12** 0.14 31.30 0.07* 0.06 22.80 0.07* 0.031 5.70

Tmean, the mean annual temperature; Tmin, minimum annual temperature; Tmax, maximum annual temperature.

*5% level of significance.

**10% level of significance. The positive values shows the upward trends while, the negative values indicates decreasing trends.

Figure 4

Spatial rainfall distribution.

Figure 4

Spatial rainfall distribution.

Figure 5

Spatial distribution of temperature.

Figure 5

Spatial distribution of temperature.

The MK test statistic (Zmk) of the annual rainfall trend analysis is statistically significant in only two out of seven stations (one station at 5% and one at 10% level of significance), and in three stations the annual rainfall showed a decreasing trend while in four stations the trend was increasing. The details of these seven stations are presented in Table 2. The results revealed that the magnitude of significantly increasing trend and variability was observed in mean annual rainfall for DB station (at 0.28 mm/year and 1.07%). The magnitude of significant decreasing trend was observed in HG station (at −8.62 mm/year and −27.88%).

Seasonal analysis of rainfall obtained from MK test statistic results are presented in Table 3. Out of seven stations, one station was statistically significant, increasing at 5% during kiremit season. Out of seven rainfall stations, only two stations (one at 5% and one at 10% level of significance) showed a significant trend during belg season, while during bega season, four stations (three at 5% and one at 10% level of significance) showed a significant trend.

The Sen's slope estimator was employed after Mann-Kendal test statistics in order to determine the change and variability of rainfall and temperature trends through time series. As presented in Table 3, the Sen's slope estimator indicates an upward trend in four stations and a downward trend in three stations for annual rainfall. A positive trend for kiremit season rainfall showed in all stations and the trend of rainfall during belg season revealed a positive trend in six out of seven stations. However, during bega season the trend of all stations was downward. This is probably due to the fluctuation and variability of the seasonal and inter-annual rainfall pattern of the Beressa watershed during the last few decades, as indicated in Table 3, which is similar to other studies (Muhire & Ahmed 2015; Zhao et al. 2015). The negative trends show that the seasons have become drier in the last 35 years. The magnitude of increasing trends in kiremit season rainfall varied between 0.33 mm/year and a percentage change of 6.13% (DBS station) to 1.62 mm/year and 31.79% (DB). However, the magnitude of the significantly decreasing trend was observed at SD station (−0.90 mm/year and −16.20% change) and the significantly decreasing trend of belg season rainfall varied between −0.12 mm/year and −10.00% at GIN station to a significantly increasing trend of 0.40 mm/year and 30.00% at DB station. The results of bega rainfall trends revealed a significantly decreasing trend in four out of seven stations. The magnitude of the decreasing trend was found to be −0.06 mm/year and −7.50% in GIN station, −0.05 mm/year and −8.80% at SD station, −0.11 mm/year and −12.70% at HG station, −0.13 mm/year and −29.00% change at SH station, −0.19 mm/year and −53.00% at DB station, −0.19 mm/year and −35.00% and −0.20 mm/year and −56.40% change at DBS station. Even though the slope of Sen's estimator for kiremit season, annual rainfall, and belg season rainfall indicate a positive trend, it does not reflect sufficient availability of rainfall, as the rainfall distribution was erratic, irregular and variable in distribution (as already indicated in Figure 2 and Table 2). Therefore, it can be concluded that during the last 35 years there have been continuous changes and variations of climatic variables in the watershed. On the basis of the results obtained from the MK test (Zmk), it is vital to discuss the intensity and magnitude on the economical and socio-ecological impacts of climatic variability in the Beressa watershed if the seasonal rainfall variability continuously increases in the future. This is particularly the case for the local community, whose economy is susceptible to variability and the erratic nature of rainfall and water shortage; recurrent drought is a common phenomenon. Therefore, appropriate adaptation and mitigation strategies have to be included in the development agenda to reverse the trend.

Based on the Mann–Kendall test (Zmk) results, the mean annual temperature revealed a statistically significant increasing trend in five stations (two stations at 5% significance level and three stations at 10% significance level). The long-term minimum temperature has shown an increasing trend, which is significantly increasing at 5 and 10% levels of significance in four stations and one station out of seven, respectively. Out of seven stations, long-term annual maximum temperature has shown a significantly increasing trend (three stations at 5% significance level and two stations at 10% significance level).

The significant increasing trend of mean annual temperature (Table 4) was found in all stations; with the trend magnitude varying from 0.03 to 0.14 °C/year respectively. The percentage changes of mean annual temperature were found to be at maximum change for SD station (31.30%) and at minimum change for DB station (7.60%). Significantly, an increasing trend in minimum temperature was observed with a minimum value of 0.005 °C/year in GIN station to a maximum value of 0.12 °C/year in DBS station. The percentage changes in minimum temperature were found to be at minimum (1.90%) and maximum (52.40%) in GIN and DBS stations, respectively. Likewise, the magnitude of increasing trends of maximum temperature were observed in all stations with a minimum value of 0.023 °C/year in GIN station and a maximum value of 0.21 °C/year in ENW station. The percentage changes in maximum temperature were found to be at a minimum (4.00%) and maximum (37.60%) in the GIN and ENW stations respectively. Significantly, the increasing long-term annual minimum and maximum temperature during the study periods indicates that it is more likely this would contribute to the increase of mean annual temperature.

The details of these stations have already been presented in Table 2.

### Spatiotemporal rainfall and temperature distribution

Assessing the long-term spatiotemporal rainfall distribution pattern is the most significant component in the climate analysis of a given country, more specifically at the local and regional levels where the effect of climate change is worse. Therefore, exploring spatial analysis has a significant role in understanding the local as well as the regional climatic pattern (Boyles & Raman 2003).

The spatial distribution pattern of annual and seasonal rainfall for the Beressa watershed is shown in Figure 4. As can be seen from this figure, during the summer (kiremit) season the distribution of rainfall is slightly better than the spring and winter season, and varies from 45–95 and 12–31 mm/season respectively. The annual rainfall distribution is also variable in time and space. The coefficient of variation is higher during the rainfall in the bega and belg seasons than the rainy season (kiremit rainfall season), as shown in Table 2. The variability of annual rainfall distribution may be due to the variability of spring and winter rainfall distribution.

Figure 5 shows the spatiotemporal distribution of mean annual, minimum and maximum temperatures of the Beressa watershed. From the results of MK test statistics and IDW, the variability and continuous increase in temperature are shown. The mean annual temperature varied between 13 and 15.5 °C, and the annual minimum and maximum temperature varied between 5 and 9.5 °C, respectively.

### Climate–crop production relationships

The results of correlation analysis between crop production and climatic variables (rainfall and temperature) during the period 1997–2014 are shown in Table 5. All the given crops show considerably high correlation with belg rainfall. Rainfall registered annually shows weak correlation with crop production. Therefore, in order to know the yields, annual rainfall is less important for prediction. With respect to the statistically significant level, only barley and wheat crops are significantly related to belg and kiremit rainfall. The correlation between rainfall during the months of May–September and crops has a positive relationship, except in the cases of beans, peas and chickpeas, which are inversely correlated with rainfall during the month of June. Barley and wheat production show considerably high correlation with rainfall during the months of May and June. All crop production shows considerably high correlation with maximum temperature and stronger correlation with barley, while in the case of minimum temperature, poor correlation was observed for all crops.

Table 5

Correlation between crop production, and rainfall and temperature (1997–2014)

VariablesBarleyWheatBeanPeaLentilChickpea
Rainfall
Annual 0.05 0.02 0.12 0.12 0.12 0.02
Belg 0.35* 0.32* 0.29 0.24 0.25 0.25
Kiremit 0.61* 0.56** 0.12 0.07 0.13 0.03
May 0.46** 0.45** 0.28 0.19 0.14 0.28
June 0.48 0.48 −0.09 −0.04 0.02 −0.06
July 0.12 0.09 0.19 0.10 0.12 0.19
August 0.07 0.03 0.08 0.001 0.10 0.04
September 0.15 0.22 0.14 0.09 0.11 0.18
Temperature
Tmin 0.28 0.27 0.27 0.35 0.25 0.28
Tmax 0.51* 0.31 0.47* 0.41 0.41 0.48*
VariablesBarleyWheatBeanPeaLentilChickpea
Rainfall
Annual 0.05 0.02 0.12 0.12 0.12 0.02
Belg 0.35* 0.32* 0.29 0.24 0.25 0.25
Kiremit 0.61* 0.56** 0.12 0.07 0.13 0.03
May 0.46** 0.45** 0.28 0.19 0.14 0.28
June 0.48 0.48 −0.09 −0.04 0.02 −0.06
July 0.12 0.09 0.19 0.10 0.12 0.19
August 0.07 0.03 0.08 0.001 0.10 0.04
September 0.15 0.22 0.14 0.09 0.11 0.18
Temperature
Tmin 0.28 0.27 0.27 0.35 0.25 0.28
Tmax 0.51* 0.31 0.47* 0.41 0.41 0.48*

*0.5% significant level.

**0.01% significant level.

Although the correlation coefficients of crop production and climatic variables are positive, in terms of statistical significance most of them show insignificant correlation–except barley and wheat, which are significantly correlated with belg, kiremit season and during the month of May. Barley, bean and chickpea show significant correlation with maximum temperature. The possible reason may be monthly, sub-monthly time scale, temporal and spatial distribution of rainfall and temperature, which are determinant factors of production. According to Al-Bakri et al. (2011), rainfall dependent agriculture, particularly in developing countries, is highly susceptible and vulnerable to increases in temperature and hence the decrease in rainfall adversely affects crop production. Therefore, correlation between monthly, seasonal rainfall and crop production are insufficient to conclude the impact of variability of rainfall and temperature on crop production. In the study area, June is the sowing period for barley and wheat crops. This cereal crop shows stronger correlation with the kiremit rains. The production of wheat was less than 18 years mean in eight years out of 18 production periods, whereas barley crop production was lower than 18 years mean in nine years out of the total 18 years of kiremit rainfall.

Bean, pea, chickpea and lentil production are particularly related to kiremit rains in all stages because these crops are sown in the second week of June. In kiremit season, rain is essential but it extends to the bega season during the harvesting stage. The production of beans was below 18-year mean in nine years out of 18 years production periods, as indicated in Table 5, which accounts for 50% of the total bean production; while in respect of kiremit rainfall pea, chickpea, and lentil production (50, 50 and 55.5% respectively) were below the 18-year mean.

Fluctuating productivity and hence food insecurity for the area is due to long-term variability in the annual and seasonal rainfall. Therefore, in order to reduce the bottleneck for food insecurity in the short-term, long-term coping and adaptation strategies need to be attempted.

Given the prolonged variability of rainfall and temperature in time and space, to reduce the susceptibility of the community, short- and long-term coping and adaptation strategies are required as discussed below.

#### Coping strategies

Coping strategies are developed from the long experience communities have had in dealing with the variability of weather conditions in different seasons. Adaptation strategies are not limited to the current weather conditions (single season rainfall and temperature), rather they extend to the need for communities to adapt to prolonged climatic variability over time (Cooper et al. 2013; Muhire & Ahmed 2015). According to Griggs & Noguer (2002), Babel et al. (2011) and Manandhar et al. (2011), adaptation strategies are an important mechanism for managing climatic change and variability. Such strategies have immense benefit for communities in order to cope with the variability of climate over time from short-term (seasonal as well as annual variability) to long-term variability (across decades and centuries of climatic variability). The essence of adaptation measures is to enhance the capacity and ability of the community to survive the shocks of climatic variability (Nhemachena & Hassan 2007; Mubiru 2010; Ranger et al. 2011). Therefore, given the prolonged climatic variability of the Beressa watershed, the following coping and adaptation mechanisms are suggested.

Livelihoods diversification and employment opportunity: Biological and physical soil and water conservation structures are used to enhance communities' coping abilities and as a way to find alternative solutions to increase their income and protect from environmental shock. Therefore, community-based soil and water conservation practices help the communities to diversify their livelihood activities. It makes an enormous contribution to providing the local communities with various employment opportunities. Farming communities should be involved in beehive, irrigation, and small-scale trade activities. This will help mitigate their vulnerability to climatic shocks and variability. Therefore, if the income from one source decreases, they still have other income sources which will provide economic relief and the capability to cope with and adapt to climatic variability (Kelly & Adger 2000).

Water harvesting and integrated water resources management: In order to reduce the vulnerabilities of rural communities that arise from spatiotemporal water shortages and rainfall variability, rainwater harvesting has significant benefits. Water harvesting is particularly important for less rainy seasons and integrated water management, and will provide supplementary irrigation during deficits.

Soil management: Soil erosion and degradation reduces crop productivity for traditional farming practices (particularly for mountainous area like the Beressa watershed), as erosion and degradation occurs at a higher rate than fertile soil formation. Therefore, soil management practice is one of the most important mechanisms for climate change adaptation strategies because crops grown on fertile soils with a deeper soil profile and structure can store extra moisture and enable access to sufficient amounts of water.

Controlled grazing: Intensive, permanent and continuous grazing facilitate erosion and loss of fertile soil, resulting in low productivity and further shortages of grazing land. Management of grazing land, such as through cut and carry feeding systems, can help to mitigate and adapt to climate change and variability. Social fencing is another mechanism that can be adopted in the region.

Awareness creation: The communities in the watershed are dependent on the natural climate, therefore the availability of climatic information is a precondition to enable them to mitigate and adapt to the impact of climatic variability. Ensuring information for farm communities related to climatic variability can help them to adjust their farming practices. Improving awareness about climatic variability and its adverse implications for their environment enables farmers to modify their resources and management practices and make efficient use of available water for better crop production.

Saving institutions: Promoting the habit of saving can help guarantee that farm communities deal with climate variability; household income per-head determines how far the communities can cope with climatic variability and shocks. Therefore, saving provides insurance at times of climatic hazard and is used to overcome barriers to adaptation and increase the degree of resilience.

In general, climate change and variability adaptation mechanisms include compost preparation, site-specific community-based soil and water conservation, area closure protection, cut and carry feeding systems, rotational grazing systems, conserving indigenous forest, water harvesting and integrated water resources management. Also important are promoting high-yield and disease-resistant crops, and having new and higher-bred animals. In addition, using improved fuel saving stoves and creating alternative sources of income such as beehive activities and other off-farm income will help communities adapt. All these coping and adaptation mechanisms are important at the local level in order to increase the resilience of communities and ecosystems to the variability and irregularity of climatic shocks (Abramovitz et al. 2001; Kurukulasuriya & Mendelsohn 2008).

Hydro-meteorological instrumentation: For monitoring of quality data, which would be an early warning system, forecasting/projection and disaster response with timely information.

## CONCLUSIONS

This study involves the observation of climatic variables, i.e. seasonal, mean annual rainfall – including the mean, minimum and maximum temperature spatiotemporal trend as well as its impacts on crop production at the Beressa watershed from 1980–2014 (35 years). Both increasing and decreasing trends of climatic variables were observed. The magnitude of the significantly increasing trend of mean annual rainfall of 0.28 mm/year and 1.07% (DB station) was recorded, whereas a significantly decreasing trend of mean annual rainfall was observed with the values of −8.62 mm/year and −27.88% (HG station).

Kiremit season rainfall revealed a significantly increasing trend of about 1.62 mm/year and 31.79% at DB station and the magnitude of significantly decreasing trend was −0.90 mm/year and −16.20% at SD station. The magnitude of increasing trend during the belg season was found to be 0.40 mm/year and 30.00% in DB station and a significantly decreasing trend was found to be −0.12 mm/year and −10.00 in GIN station. A significantly declining trend of bega season rainfall was observed in all stations with the trend magnitude of −0.61 mm/year and −7.50% in GIN station to −0.21 mm/year and −56.40% in DBS station. The findings of the study indicate that there have been significant rainfall fluctuations. Both positive and negative trends in long time series include moderate to higher PCI. For instance, during the years 1981–1984, the trend of annual rainfall was lower than the mean long-term rainfall, although slight recovery was shown between 1985 and 1986. A significant increase in annual mean temperature was observed in all stations, with the magnitude varying from 0.03 °C/year and 7.60% in DB station to 0.14 °C/year and 31.30% at SD station. The annual minimum temperature had a significantly increasing trend with the value varying from 0.005 °C/year and 1.90% in GIN station to 0.12 °C/year and 52.40% in the DBS station. Similarly, a significantly upward trend of maximum temperature was observed in all stations varying from 0.023 °C/year and 4.00% in GIN station with a maximum value of 0.21 °C/year and 37.60% in ENW station.

Continuously increasing temperature, together with the variability and fluctuation of seasonal and inter-annual rainfall is a root cause for the decrease and fluctuation of crop production. Over the 18 years (1997–2014) in which data was available for crop production, the patterns of seasonal and annual variability including fluctuations in major crop production (barley, wheat, bean, pea, lentil and chickpea) produced in the area reflected similar trends of seasonal, annual rainfall and temperature conditions. Crop production showed high correlation with belg and kiremit rainfall; only annual rainfall and barley crops showed stronger correlation. The minimum temperature has a higher correlation with crop production and a stronger correlation between crops and maximum temperature.

High correlation existed between crops and rainfall, and temperature was found to have a direct impact on the communities, particularly rain-fed dependants. Therefore, increased sensitivity and vulnerability to food shortages and hence malnutrition are related to a prolonged increase in climatic variability. Therefore, there is a need for community-based coping and adaptation strategies such as adopting soil, water conservation and water harvesting strategies; and increasing diversified crops, high value and market oriented crops, fast growing crops and climate resistant crops, which are less susceptible to future climatic variability.

## REFERENCES

Abramovitz
J.
,
Banuri
T.
,
Girot
P. O.
,
Orlando
B.
,
Schneider
N.
,
Spanger-Siegfried
E.
,
Switzer
J.
,
Hammill
A.
2001
Adapting to Climate Change: Natural Resource Management and Vulnerability Reduction
.
Background paper to the Task Force on Climate Change, Adaptation and Vulnerable Communities
.
S.
,
Selassie
Y. G.
,
Fissha
G.
,
Gedif
B.
2015
Time series trend analysis of temperature and rainfall in lake Tana Sub-basin, Ethiopia
.
Environ. Syst. Res.
4
(
1
),
25
.
Al-Bakri
J.
,
Suleiman
A.
,
Abdulla
F.
,
J.
2011
Potential impact of climate change on rainfed agriculture of a semi-arid basin in Jordan
.
Phys. Chem. Earth A/B/C
36
(
5
),
125
134
.
Anderson
R. L.
1942
Distribution of the serial correlation coefficient
.
Ann. Math. Stat.
13
(
1
),
1
13
.
Babel
M. S.
,
Agarwal
A.
,
Swain
D. K.
,
Herath
S.
2011
Evaluation of climate change impacts and adaptation measures for rice cultivation in Northeast Thailand
.
Clim. Res.
46
(
2
),
137
146
.
Barnett
T. P.
,
J. C.
,
Lettenmaier
D. P.
2005
Potential impacts of a warming climate on water availability in snow-dominated regions
.
Nature
438
(
7066
),
303
309
.
Belay
K.
,
Beyene
F.
,
Manig
W.
2005
Coping with drought among pastoral and agro-pastoral communities in eastern Ethiopia
.
J. Rural Dev.
28
(
2
),
185
210
.
Boyles
R. P.
,
Raman
S.
2003
Analysis of climate trends in North Carolina (1949–1998)
.
Environ. Int.
29
(
2
),
263
275
.
Burn
D. H.
,
Elnur
M. A. H.
2002
Detection of hydrologic trends and variability
.
J. Hydrol.
255
(
1
),
107
122
.
Conway
D.
,
Schipper
E. L. F.
2011
Adaptation to climate change in Africa: challenges and opportunities identified from Ethiopia
.
Glob. Environ. Change
21
(
1
),
227
237
.
Cooper
P.
,
Stern
R. D.
,
Noguer
M.
,
Gathenya
J. M.
2013
Climate change adaptation strategies in Sub-Saharan Africa: foundations for the future
. In:
Climate Change-Realities, Impacts Over Ice Cap, Sea Level and Risks
.
InTech, London
.
Di Falco
S.
,
Yesuf
M.
,
Kohlin
G.
,
Ringler
C.
2012
Estimating the impact of climate change on agriculture in low-income countries: household level evidence from the Nile Basin, Ethiopia
.
Environ. Resour. Econ.
52
(
4
),
457
478
.
Fazzini
M.
,
Bisci
C.
,
Billi
P.
2015
The climate of Ethiopia
. In:
Landscapes and Landforms of Ethiopia
.
Springer
,
Dordrecht
.
Fu
G.
,
Charles
S. P.
,
Yu
J.
,
Liu
C.
2009
Decadal climatic variability, trends, and future scenarios for the North China Plain
.
J. Clim.
22
(
8
),
2111
2123
.
Gamachu
D.
1988
Environment and development in Ethiopia
. In:
Beyond the Famine: an Examination of the Issues Behind Famine in Ethiopia
(
Penrose
A.
, ed.).
International Institute for Relief and Development and Food for the Hungry International
,
Geneva
, pp.
54
96
.
Gebre
H.
,
Kindie
T.
,
Girma
M.
,
Belay
K.
2013
Trend and variability of rainfall in Tigray, northern Ethiopia: analysis of meteorological data and farmers’ perception
.
1
(
6
),
88
100
.
Gemmer
M.
,
Becker
S.
,
Jiang
T.
2004
Observed monthly precipitation trends in China 1951–2002
.
Theor. Appl. Climatol.
77
(
1
),
39
45
.
Helsel
D.
,
Hirsch
R.
1992
Statistical Methods in Water Resources
.
Elsevier
,
New York
.
Hirsch
R. M.
,
Slack
J. R.
,
Smith
R. A.
1982
Techniques of trend analysis for monthly water quality data
.
Water Resour. Res.
18
(
1
),
107
121
.
Hulme
M.
,
Doherty
R.
,
Ngara
T.
,
New
M.
,
Lister
D.
2005
Global warming and African climate change: a reassessment
.
Clim. Change Afr.
29
40
.
M.
,
Marttila
H.
,
Kløve
B.
2014
Long-term variations and trends in precipitation in Finland
.
Int. J. Climatol.
34
(
10
),
3139
3153
.
Iskander
S. M.
,
Rajib
M. A.
,
Rahman
M. M.
2014
Trending regional precipitation distribution and intensity: use of climatic indices
.
Atmos. Clim. Sci.
4
(
3
),
385
.
Jain
S. K.
,
Kumar
V.
2012
Trend analysis of rainfall and temperature data for India
.
Curr. Sci.
102
,
37
49
.
Kelly
P. M.
,
W. N.
2000
Theory and practice in assessing vulnerability to climate change and facilitating adaptation
.
Clim. Change
47
(
4
),
325
352
.
Kenabatho
P.
,
Parida
B.
,
Moalafhi
D.
2012
The value of large-scale climate variables in climate change assessment: the case of Botswana's rainfall
.
Phys. Chem. Earth A/B/C
50
,
64
71
.
Kendall
M.
1975
Rank Correlation Measures
.
Charles Griffin
,
London
.
Kishore
P.
,
Jyothi
S.
,
Basha
G.
,
Rao
B.
,
Rajeevan
M.
,
Velicogna
I.
,
Sutterley
T. C.
2016
Precipitation climatology over India: validation with observations and reanalysis datasets and spatial trends
.
Clim. Dyn.
46
(
1–2
),
541
556
.
Kløve
B.
,
Ala-Aho
P.
,
Bertrand
G.
,
Gurdak
J. J.
,
Kupfersberger
H.
,
Kværner
J.
,
Muotka
T.
,
Mykrä
H.
,
Preda
E.
,
Rossi
P.
2014
Climate change impacts on groundwater and dependent ecosystems
.
J. Hydrol.
518
,
250
266
.
Kumar
P.
,
Wiltshire
A.
,
Mathison
C.
,
Asharaf
S.
,
Ahrens
B.
,
Lucas-Picher
P.
,
Christensen
J. H.
,
Gobiet
A.
,
Saeed
F.
,
Hagemann
S.
2013
Downscaled climate change projections with uncertainty assessment over India using a high resolution multi-model approach
.
Sci. Total Environ.
468
,
S18
S30
.
Kurukulasuriya
P.
,
Mendelsohn
R.
2008
Crop switching as a strategy for adapting to climate change
.
Afr. J. Agric. Resour. Econ.
2
(
1
),
105
126
.
Liang
L.
,
Li
L.
,
Liu
Q.
2011
Precipitation variability in Northeast China from 1961 to 2008
.
J. Hydrol.
404
(
1
),
67
76
.
Manandhar
S.
,
Vogt
D. S.
,
Perret
S. R.
,
Kazama
F.
2011
Adapting cropping systems to climate change in Nepal: a cross-regional study of farmers’ perception and practices
.
Reg. Environ. Change
11
(
2
),
335
348
.
Mann
H. B.
1945
Nonparametric tests against trend
.
Econometrica
13
(
3
),
245
259
.
Mekasha
A.
,
Tesfaye
K.
,
Duncan
A. J.
2014
Trends in daily observed temperature and precipitation extremes over three Ethiopian eco-environments
.
Int. J. Climatol.
34
(
6
),
1990
1999
.
Merasha
E.
1999
Annual rainfall and potential evapotranspiration in Ethiopia
.
Ethiop. J. Nat. Resour.
.
Mondal
A.
,
Khare
D.
,
Kundu
S.
2015
Spatial and temporal analysis of rainfall and temperature trend of India
.
Theor. Appl. Climatol.
122
(
1–2
),
143
158
.
Mubiru
D.
2010
Climate Change and Adaptation Options in Karamoja
.
Food and Agriculture Organization
.
Muhire
I.
,
Ahmed
F.
2015
Spatio-temporal trend analysis of precipitation data over Rwanda
.
S. Afr. Geogr. J.
97
(
1
),
50
68
.
Murugan
M.
,
Mukund
V.
,
Ramesh
R.
,
Hiremath
M.
,
Josephrajkumar
A.
,
Shetty
P.
2008
Centennial rainfall variation in semi arid and tropical humid environments in the cardamom hill slopes, southern Western Ghats, India
.
Caspian J. Environ. Sci.
6
(
1
),
31
39
.
Nhemachena
C.
,
Hassan
R.
2007
Micro-level Analysis of Farmers Adaption to Climate Change in Southern Africa
.
International Food Policy Research Institute
,
Washington, DC
,
USA
.
Oliver
J. E.
1980
Monthly precipitation distribution: a comparative index
.
Prof. Geogr.
32
(
3
),
300
309
.
Osman
M.
,
Sauerborn
P.
2001
Soil and water conservation in Ethiopia
.
J. Soils Sediments
1
(
2
),
117
123
.
Pachauri
R. K.
,
Allen
M. R.
,
Barros
V. R.
,
Broome
J.
,
Cramer
W.
,
Christ
R.
,
Church
J. A.
,
Clarke
L.
,
Dahe
Q.
,
Dasgupta
P.
2014
Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
.
IPCC
,
Geneva
,
Switzerland
.
Parry
M. L.
2007
Climate Change 2007 – Impacts, Adaptation and Vulnerability: Working Group II Contribution to the Fourth Assessment Report of the IPCC
.
Cambridge University Press
,
Cambridge
.
Partal
T.
,
Kahya
E.
2006
Trend analysis in Turkish precipitation data
.
Hydrol. Process.
20
(
9
),
2011
2026
.
Pingale
S. M.
,
Khare
D.
,
Jat
M. K.
,
J.
2016
Trend analysis of climatic variables in an arid and semi-arid region of the Ajmer District, Rajasthan, India
.
J. Water Land Dev.
28
(
1
),
3
18
.
Randell
H.
,
Gray
C.
2016
Climate variability and educational attainment: evidence from rural Ethiopia
.
Glob. Environ. Change
41
,
111
123
.
Ranger
N.
,
Hallegatte
S.
,
Bhattacharya
S.
,
Bachu
M.
,
Priya
S.
,
Dhore
K.
,
Rafique
F.
,
Mathur
P.
,
Naville
N.
,
Henriet
F.
2011
An assessment of the potential impact of climate change on flood risk in Mumbai
.
Clim. Change
104
(
1
),
139
167
.
Rashid
M.
,
Beecham
S.
,
Chowdhury
R.
2013
Assessment of statistical characteristics of point rainfall in the Onkaparinga catchment in South Australia
.
Hydrol. Earth Syst. Sci. Discuss.
10
(
5
),
5975
6017
.
Rashid
M. M.
,
Beecham
S.
,
Chowdhury
R. K.
2015
Assessment of trends in point rainfall using Continuous Wavelet Transforms
.
82
,
1
15
.
Seleshi
Y.
,
Zanke
U.
2004
Recent changes in rainfall and rainy days in Ethiopia
.
Int. J. Climatol.
24
(
8
),
973
983
.
Simane
B.
,
Zaitchik
B. F.
,
Mesfin
D.
2012
Building climate resilience in the Blue Nile/Abay Highlands: a framework for action
.
Int. J. Environ. Res. Public Health
9
(
2
),
610
631
.
Singh
P.
,
Kumar
V.
,
Thomas
T.
,
Arora
M.
2008
Changes in rainfall and relative humidity in river basins in northwest and central India
.
Hydrol. Process.
22
(
16
),
2982
2992
.
Solomon
S.
2007
Climate Change 2007 – the Physical Science Basis: Working Group I Contribution to the Fourth Assessment Report of the IPCC
.
Cambridge University Press
,
Cambridge
.
Stocker
T. F.
,
Qin
D.
,
Plattner
G.-K.
,
Tignor
M.
,
Allen
S. K.
,
Boschung
J.
,
Nauels
A.
,
Xia
Y.
,
Bex
B.
,
Midgley
B.
2013
IPCC, 2013: Climate Change 2013: the Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
.
Cambridge University Press
,
Cambridge
.
Suarez
J.
,
Suarez
P.
,
Bachofen
C.
2012
Center Task Force Report: Games for a New Climate: Experiencing the Complexity of Future Risks
.
.
Suryavanshi
S.
,
Pandey
A.
,
Chaube
U. C.
,
Joshi
N.
2014
Long-term historic changes in climatic variables of Betwa Basin, India
.
Theor. Appl. Climatol.
117
(
3–4
),
403
418
.
Wagesho
N.
,
Yohannes
E.
2016
Analysis of rainfall variability and farmers’ perception towards it in Agrarian Community of Southern Ethiopia
.
Analysis
6
(4).
Xu
Z.
,
Li
J.
,
Liu
C.
2007
Long-term trend analysis for major climate variables in the Yellow River basin
.
Hydrol. Process.
21
(
14
),
1935
1948
.
Yavuz
H.
,
Erdoğan
S.
2012
Spatial analysis of monthly and annual precipitation trends in Turkey
.
Water Resour. Manage.
26
(
3
),
609
621
.
Yue
S.
,
Pilon
P.
,
Phinney
B.
2003
Canadian streamflow trend detection: impacts of serial and cross-correlation
.
Hydrol. Sci. J.
48
(
1
),
51
63
.
Zhao
W.
,
Yu
X.
,
Ma
H.
,
Zhu
Q.
,
Zhang
Y.
,
Qin
W.
,
Ai
N.
,
Wang
Y.
2015
Analysis of precipitation characteristics during 1957–2012 in the semi-arid Loess Plateau, China
.
PLoS ONE
10
(
11
),
e0141662
.