Spatiotemporal climate variability and extremes in Middle Awash Afar region Ethiopia: implications to pastoralists and agro-pastoralists food security

Abstract

Purpose

This study aims to examine the phenomenon of climate variability and its implications for pastoralists and agro-pastoralists food security in Amibara and Awash Fentale districts of the Afar region, Ethiopia.

Design/methodology/approach

The study relied on meteorological records of temperature and rainfall in the study area between 1988 and 2018. Besides, literature on the topic was reviewed to make caveats on the literal picture that comes from quantitative data, and that is the contribution of this study to the existing debate on climate change and variability. The spatiotemporal trend was determined using the Mann–Kendall test and Sen’s slope estimator, while variability was analyzed using the coefficient of variation and standardized anomaly index, and standardized precipitation index/standardized precipitation evapotranspiration index were applied to determine the drought frequency and severity.

Findings

The result reveals that the mean seasonal rainfall varies from 111.34 mm to 518.74 mm. Although the maximum and minimum rainfall occurred in the summer and winter seasons, respectively, there has been a decrease in seasonal and annual at the rate of 2.51 mm per season and 4.12 mm per year, respectively. The study sites have been experiencing highly seasonal rainfall variability. The drought analysis result confirms that a total of nine agricultural droughts ranging from moderate to extreme years were observed. Overall, the seasonal and annual rainfall of the Amibara and Awash Fentale districts showed a decreasing trend with highly temporal variations of rainfall and ever-rising temperatures, and frequent drought events means the climate situation of the area could adversely affect pastoral and agro-pastoral households’ food security. However, analysis of data from secondary sources reveals that analyzing precipitation just based on the meteorological records of the study area would be misleading. That explains why flooding, rather than drought, is becoming the main source of catastrophe to pastoral and agro-pastoral livelihoods.

Practical implications

The analysis of temperature and rainfall dynamics in the Afar region, hence the inception of all development interventions, must take the hydrological impact of the neighboring regions which appears to be useful direction to future researchers.

Originality/value

The research is originally conducted using meteorological and existing literature, and hence, it is original. In this research, we utilized a standardized and appropriate methodology, resulting in insights that augment the existing body of knowledge within the field. These insights serve to advance scholarly discourse on the subject matter.

Citation

Aytenfisu, A.T., Tolossa, D., Feleke, S.T. and Ayal, D.Y. (2024), "Spatiotemporal climate variability and extremes in Middle Awash Afar region Ethiopia: implications to pastoralists and agro-pastoralists food security", International Journal of Climate Change Strategies and Management, Vol. 16 No. 2, pp. 228-252. https://doi.org/10.1108/IJCCSM-11-2023-0140

1. Introduction

Climate variability poses significant threats to the achievement of sustainable development goals at both global and national levels. Developing countries, particularly those heavily reliant on nature for their economy and livelihoods, face practical and severe consequences from these phenomena (Filho et al., 2022; Pradhan et al., 2022; Birkmann et al., 2022). Ethiopia experiences rising temperatures, unpredictable rainfall patterns, recurrent droughts, irreversible land degradation, worsening structural poverty and chronic food insecurity (IFPRI and UNDP, 2019). Climate variability and extreme events could disrupt the national economy, exacerbating food insecurity, soil erosion, runoff and water resource depletion, impacting livelihoods (Radović and Iglesias, 2019; Shukla et al., 2019). The northeastern lowlands of Ethiopia confront multiple challenges including climate extremes, invasive alien species and pressure from population-induced rangeland degradation. These factors collectively weaken the ecosystem and biodiversity potential of the area, potentially worsening food insecurity (Birkmann et al., 2022; Ayal et al., 2023; IFPRI and UNDP, 2019; Shiferaw et al., 2021; IPCC, 2014).

The climate variables such as temperature, precipitation and humidity are inherently characterized by spatiotemporal variations (Doshi et al., 2023). These variations and associated impacts on the livelihood and food security of the exposed community could be explained by hydrometeorological spatial (local, regional and global) and temporal (daily, seasonal and annual) events distribution behavior (Niang et al., 2014). Drought is among the most recurring hydrometeorological hazards characterized by extended periods of falling precipitation below average and hence, causes pasture and water scarcity. Drought frequency, duration and intensity could vary significantly across regions and over time. The severity of drought and its effect could be exaggerated by non-climatic factors, e.g. rangeland degradation, demographic and institutional factors (Filho et al., 2021; Filho et al., 2020; Ayal et al., 2019). It seems clear that understanding how these variables change across seasons and years at the microlevel is essential for discerning long-term climate trends for adequate and proper intervention (Birkmann et al., 2022; Filho et al., 2022; Shukla et al., 2019).

Cognizant of this reality, this study examines the spatiotemporal trends, distributional behavior and drought situation using observational data. The temporal aspect of climate variability has been reconstructed at seasonal and annual scales. The spatial aspect of climate variability has been reconstructed based on meteorological data throughout the Amibara and Awash Fentale districts. Drought frequency and magnitude were also examined using established drought indices and investigated the patterns of various climate extremes. In so doing, the article attempts to contribute to a deeper understanding of climate variability in a localized setting and its real impacts and potential implications for ecosystems, societies and economies.

2. Methodology

2.1 Description of the study area

The study area is located in Afar administrative region (Zone 3), latitude between 9°06’ and 9°69’ N and longitude of 40°10’ and 40°46’ E (Figure 1), with areas of 1,046.41 km² and 2,007.05 km², respectively (Adnew et al., 2019). The topography of this area is generally flat, with a maximum altitude of up to 1,342 m above sea level.

The climate of the area is hot and semi-arid. The mean annual temperature is estimated at 26.64°C. The mean annual rainfall is 468.69 mm. Precipitation is generally little, irregular and unpredictable and seems to follow a bimodal pattern in February–April and July–August (Figure 2).

2.2 Statistical variability test

Understanding the spatiotemporal trends and variability of rainfall and temperature is best achieved through the integration of observational data and climate models. Historical climate data, gathered from meteorological stations, satellites and remote sensing, provides valuable insights into past climate patterns. In addition, climate models are used to simulate both past and future climate conditions, enabling scientists to analyze trends and variability within a controlled environment. Various statistical techniques, including time series analysis and spatial interpolation, are commonly used to discern patterns and quantify changes in climate variables.

In this study, a range of statistical techniques has been applied to analyze climate data, primarily categorized into variability and trend analysis (Kiros et al., 2016; Koudahe et al., 2017).

Parameters such as mean, standard deviation, median, kurtosis, skewness, range and the coefficient of variance (CV) were used in the analysis process. The CV can be calculated using equation (1):

1 CV=σμ×100

where σ is the standard deviation, and μ is the mean, CV < 20, less variability; 20 < CV < 30, moderate variability; and CV beyond 30, high variability (Asfaw et al., 2018; Alemayehu et al., 2020)

2.3 Standard anomaly index

Standard anomaly index (SAI) was used to compute negative and positive rainfall and temperature fluctuation anomalies in the study area. The SAI indicates the distance between the data and its mean value (Alemayehu and Bewket, 2017). It is calculated as follows:

(1) SAI=(x−x¯σ)

where x is the rainfall or temperature data, x¯ and σ are the mean and the standard deviation of the data, respectively.

2.4 Temporal and spatial trend analysis of climate variables

Different types of tests are available to detect and estimate trends of climate variables. Trend detection can be performed using parametric and nonparametric methods. An example of the parametric method is the linear regression test, and nonparametric (also called rank-based) methods include the Mann–Kendall test (Mann, 1945; Kendall, 1948). The linear regression model was used to measure the pattern or trend of variables over a long period (Kiros et al., 2016). It is calculated using equation (2).

(2) Y=a+bxt

where Y indicates the trend value, a is the intercept, b is the slope of the trend and x_t is the time point.

Mann–Kendall or nonparametric test is applied to analyze the trends of the climate variables. The tests are distribution-free and can tolerate data outliers (Kendall, 1948; Mann, 1945; Yue et al., 2002). The mathematical equations for calculating Mann–Kendall statistics (S) and standardized test statistics Z are as follows (equations (4)–(7)]:

(3) S=∑i=1N−1∑j=i+1Nsgn(xj−xi)

(4) sgn(θ)={1 if (xj−xi)>00 if (xj−xi)=0−1 if (xj−xi)<0}

(5) V(S)=118[n(n−1)(2n+5)−∑k=1gtk(tk−1)(2tk+5)]

(6) Zs={S−1Var(S)S>00,       S=0S+1Var(S) S<0

where N is each number of data points, assuming (x_j – x_i) = θ, the value sgn(θ) was computed as the signum functioning, which is the number of data points, g is the number of tied groups (a set of data having the same value) and t_k is the number of data series in the k^th group. Z is the standard statistics test. The statistical significance level of the trend variation was evaluated using Z_s value. A positive Z_s value indicates an increasing trend, while a negative Z_s value shows a decreasing trend. The null hypothesis H0 should be rejected if |Zs|>Z(1−α2)at the α = 0.05 level of significance (Du et al., 2013). However, time-series climate data has positive autocorrelation, increasing the probability of significant output that indirectly leads to a false trend (Alexander et al., 2009; Sa’adi et al., 2019). Thus, this study applied autocorrelation tests of the time series before using the Mann-Kendall (MK) trend test.

2.5 Sen’s slope estimator

Sen’s slope (Sen, 1968) estimator was used to predict the magnitude of the trend. The nonparametric method can evaluate the change per unit of time. This technique assumes a linear trend in the time series. The slope (Q_i) of all pairs of data x can be calculated as

(7) Qi=xj−Xkj−k,i=1,2,…N,j>k

where x_j and x_k are data values of time j and k, respectively. If there are n values in the time series, then there are as many as n = n(n – 1)/2 slop estimates. The Sen’s slope estimator is defined as the median of the n values of Q_i. The values of slopes are ranked from the smallest to the largest, and Sen’s slope estimator Q_i is calculated as

If N is an odd observation: Qi=Q(N+1)/2 8

If N is even observation: Qi=12(QN/2+Q(N+2)/2) 9

In this study, each climate variable was analyzed based on four seasons; they were taken as Spring runs from March to May, Summer runs from June to August, Autumn runs from September to October and Winter runs from November to February.

2.6 Test for single change point (Pettiti test)

The Pettitt test has been used in several climate studies to detect abrupt changes in the mean distribution of the variable of interest. It is applied to detect a single change point in climate series with continuous data and gives information about the location of the shift. The test statistic U_t,T is evaluated for all random variables from 1 to T; then, the most significant change point is selected where the value of ⌊U_t,T⌋ is the largest (Jaiswal et al., 2015). To identify a change point, a statistical index U_t is defined as follows:

(10) Ut,T=∑i=1t∑j=1TSgn(xi−xj), 1≤t≤T

where similar to the MK test,

(11) Sgn(θ)={+1 θ>0 0  θ=0   −1   θ<0

The most probable change point is found where its value is (the break occurs in year k when). The test statistic Kn and the associated probability (p) used in the test are given as.

(12) Kt0=max⁡1≤y≤n|Ut,T|

and the significance probability associated with the value K_t is evaluated as

(13) p(t0)=2exp[−6kto2T3+T2]

where t₀ is concluded as a significant change point when  Pt0≤0.5. The value is then compared with the critical value (Pettitt, 1979). Given a certain significance level α, if p < α, we reject the null hypothesis and conclude that x_t is a significant change point at level α (Du et al., 2013).

2.7 Drought indices

Studying droughts involves assessing their historical occurrence and projected changes in a changing climate. Standardized precipitation index (SPI), the Palmer Drought Severity Index and the soil moisture anomaly index are widely used to quantify the severity, frequency and spatial extent of drought events at different scales. Moreover, advanced techniques like remote sensing and machine learning are increasingly being used to monitor and predict drought conditions. Different drought indices have been used to analyze drought characteristics in different areas (Morid et al., 2006). This is because a single index does not provide a complete picture of the spatiotemporal distribution of the drought characteristics of the area (Alsafadi et al., 2020; Temam et al., 2019; Tefera et al., 2019; Kourouma et al., 2022). In this study, the two multiscalar (i.e. multiple timescales) drought indices, including the SPI and the SPEI in 3- and 12-month time scales, were used to investigate the occurrence of drought and wet events in the area. The SPI index, the two-parameter Gamma distribution, has been suggested (McKee et al., 1993). The SPEI index is calculated with the three-parameter log-logistic distribution(Vicente-Serrano et al., 2010). Table 1 is used to categorize the dry and wet values of the indices.

2.8 Standardized precipitation index

The SPI evaluates drought in precipitation deficit, impacting groundwater availability, soil moisture streamflow and reservoir storage. Organization (2012) and McKee et al. (1993) developed the SPI to analyze precipitation departures from the average precipitation for a particular month or determining time scale. The SPI index is based on the cumulative probability of the considered precipitation. The SPI calculation consists of adjusting the Gamma probability density function to the frequency distribution of the precipitation of each of the rainfall stations. The Gamma probability density function is given by the equation (19).

(14) f(x)=1βαΓ(α)xα−1e−xβ x>0

where α > 0 is the shape of the parameter, β > 0 is the scale parameter, x > 0 is precipitation and г(α) is the Gamma function, defined as equation (2).

(15) Γ(α)=∫0∞xα−1e−xdx

The parameters α and β of the probability density function are estimated for each station by maximum likelihood (Wilks, 2011).

(16) α^=14A1+1+4A3

(17) β^=x^α^

(18) A=lnx¯−∑inxin

The resulting parameters are then used, and the cumulative probability of the occurrence of a precipitation event for the given month and time scale at the station is considered. The cumulative probability is given by:

(19) G(x)=∫0xg(x)dx=1βγΓ(γ)0∫0xxγ−1e−xβdx

where α > 0 is the shape parameter, β > 0 is the scale parameter, x is the precipitation and Γ (α) is the Gamma function, defined as

(20) G(x)=1Γ(γ)∫0xtγ−1e−tdt

because the gamma function is undefined for x = 0 and the precipitation distribution may contain zeros, the cumulative probability becomes:

(21) H(x)=q+(1−q)G(x)

where q is the probability of zero precipitation. H(x) is then converged into the standard normal SPI variable following approximation (Abramowitz and Stegun, 1965). The normal standardized distribution with null average and the unit variance was then obtained from the transformation of

(22) SPI={−(t−c0+c1t+c2t21+d1t+d2t2+d2t3) for 0<H(x)≤0.5t=ln(1H(x)2)+(t−c0+c1t+c2t21+d1t+d2t2+d2t3)   for 0.5<H(x)≤0.1t=ln(11−H(x)2)

H(x) is the cumulative probability of observed precipitation. The coefficient C₀, C₁, C₂ and d₁, d₂, d₃ are given as follows: C₀ = 2.515517, C₁ = 0.802853, C₂ = 0.010328 and d₁ = 1.432788, d₂ = 0.189269, d₃ = 0.001308. The SPI values are symmetrical and can identify both dry and wet conditions. The drought begins when the SPI value becomes negative and ends while returning to a positive value (wet events) (McKee et al., 1993).

2.9 Standardized precipitation evapotranspiration index

The SPEI is multiscalar and used to evaluate different types of global droughts (Labudová et al., 2017; Wang et al., 2014; Yang et al., 2019). The SPEI algorithm is mathematically similar to SPI, but it includes the effect of temperature in addition to precipitation. The SPEI is computed using precipitation (P) and PET as input variables, resulting in the climate water balance (P-PET) outputs (Vicente-Serrano et al., 2010). Including the widely used Penman–Monteith (PM) method, there are several methods have been introduced to calculate (PET) (Allen et al., 1998). However, the PM method requires complete meteorological data, which is challenging in many parts of the world. Thus, the PET can be estimated using the Hargraves (Hargreaves and Samani, 1985) method, which has a similar output as PM (Beguería et al., 2014). Hargraves method needs only precipitation, minimum and maximum temperature and is extraterrestrial radiation (R_a) (Senay et al., 2011). Hence, in this study, the Hargraves method was used to calculate the PET.

(23) PETHG=0.0023×(Tmean+17.8)×(Tmax−Tmin)×Ra

where PET_HG is potential evapotranspiration of the Hargreaves method (mm/day), R_a is the extraterrestrial radiation (mm/day), calculated theoretically as a function of latitude, T-mean is the average temperature (°C) and T-max and T-min are the maximum and minimum temperature (°C), respectively. Once PET was estimated, the water balance equation was used to calculate the monthly deficit (D_i)

(24) Di=Pi−PETi

where D_i is climatic water balance (CWB) in a given period (mm), P_i is monthly precipitation in a given period (mm) and PET_i is the monthly potential evapotranspiration (mm). The accumulated difference between P and PET in different time scales can be calculated as

(25) Dnk==∑i−0k−1(Pn−i−PETn−i)

where n ≥ k, k is a different time scale, and n is the number of calculations.

The function of log-logistic distributions gives better results than other distributions for obtaining SPIE series in standardized D with a mean of zero and standard deviation of one (Potop and Možný, 2011; Vicente-Serrano et al., 2010). The cumulative distribution function f(x) is given by

(26) f(x)=βα(x−γα)β−1[1+(x−γα)β]−2

where α, β and γ are scale, shape and location parameters, respectively, for D values in the range (γ < D < ∞). The probability distribution function of the D series, according to the log-logistic distribution, is given by

(27) F(x)=βα[1−(αx−γ)β]−1

The SPEI can be obtained as the standardized values of f(x).

(28) SPEI=W−C0+C1W+C2W21−d1W+d2W2+d2W3

(29) W=−2ln(p) for p≤0.5

where p is the probability of exceeding a determined D value, p = 1 − F (x), if p > 0.5, then p is replaced by 1 − p and the sign of the resultant SPEI is reversed; the constants are C₀ = 2.515517, C₁ = 0.8022853, C₂ = 0.010328, d₁ = 1.432788, d₂ = 0.189269 and d₃ = 0.001308. Detailed computing of SPEI has been widely described in Vicente-Serrano et al. (2010).

SPI/SPEI can be calculated on different time scales, such as 1–3-, 6–12- and 24-months and more (Mathbout et al., 2018). These time scales interpret the different effects of drought on the various types of water resources (McKee et al., 1993; Vilà et al., 2011). This study calculated the SPI and the SPEI indices using a 3–12-month time scale. The three-month time scale is crucial for understanding agricultural drought events, while the 12- and 24-month time scales are exclusively used to analyze hydrological drought events or annual water deficits (Di Giuseppe et al., 2019; Alsafadi et al., 2020; Potop et al., 2012).

2.10 Assessing drought characteristics

Drought characteristics can be expressed by the following essential features: duration, frequency, intensity, severity and spatial and temporal extent (Alamgir et al., 2015; Andreadis et al., 2005). The study evaluated the drought occurrence across months whereby measurements that fall between SPI/SPEI ≤ − 1 are further categorized as moderate, severe and extreme drought events. In this research, the assessment of mild drought values was excluded since the variation from the normal level is very slight. The duration, severity, intensity and frequency of the drought events were calculated based on Table 1:

• The frequency is the number of months in which the SPEI value meets a set value (Table 1) divided by the number of months in the entire series (Wang et al., 2014).

  (30) F=nN×100

where n is the number of months of drought events (SPI/SPEI < −1) that an index value meets a set drought criterion divided by the number of months in the entire series (N). Drought frequency (F) was used to assess the drought prevalence during the study period.

• Duration is monthly or more length drought episodes. Magnitude (M) is the cumulative sum of the index value based on the duration of drought occurrence:

  (31) M=∑i=1DurationIndex

• The intensity of a drought event is the magnitude divided by the duration. Events that have a shorter duration and higher severities will have larger intensities.

  (32) I=MagnitudeDuration

The inverse distance weighting method was used to visualize the spatial patterns of climate variability and change.

3. Results and discussion

3.1 Temporal trend of mean seasonal and annual analysis of climate variables

Table 2 depicts the mean seasonal and annual climate variability statistical description and the trend analysis of the study area from 1988 to 2018. The mean seasonal rainfall of the area ranges from 111.34 mm to 518.74 mm. The maximum and minimum rainfall occurred in the summer and winter seasons, respectively. However, in the area a decrease in seasonal and annual rainfall has been decreasing at the rate of 2.51 mm per season and 4.12 mm per year for the past three decades. Relatively, the highest amount of rainfall reduction was observed in the winter season. The rainfall distribution behavior was characterized by high fluctuations all year round. For instance, rainfall distribution was highly variable in spring (CV = 46%), summer (37.95%), autumn (63.75%) and winter (81.96%) (Table 2).

The seasonal and annual maximum and minimum temperatures in the study site vary across seasons. Both seasonal and annual minimum and maximum temperatures increased significantly. The mean seasonal maximum temperature variability ranges from 33.44°C to 34.95°C. The mean seasonal minimum temperature ranges between 18.35°C and 20.48°C. Generally, according to Table 2, the seasonal and annual rainfall of the Amibara and Awash Fentale districts observed a decrease in amount and highly variable in distribution, which means it could affect soil moisture and hence water and pasture availability of the study sites.

3.2 Standard anomaly index

In this study, analysis of the SAI of hydroclimate data shows the reference line bars above the mean (zero) as positive (surplus), and those below the line mean are negative (deficit) anomalies (Koudahe et al., 2017). The calculated value of SAI obtained was plotted against the year (1988–2018) for annual mean rainfall and maximum and minimum temperatures. Accordingly, in Figure 3, the mean annual rainfall anomaly depicts a negative trend and the coefficient of determination (R²) of 0.011. Both the mean minimum and maximum temperatures reveal an increasing trend R² 0.13 and 0.33, respectively. The reinforcing effect of declining rainfall and steadily increasing temperatures could worsen pasture and water availability and, hence, food insecurity in the pastoral areas. It seems clear that the reinforcing effect of decreasing rainfall and increasing temperatures exacerbated the availability and accessibility of pasture and water, leading to heightened food insecurity in the study sites.

3.3 The spatial trend of mean seasonal analysis of climate variables

The Kriging method is widely used for spatial analysis using geographic information systems (GIS). In this study, spatial analysis was applied using the distance-weighting approach through interpolating the data by including information from specific locations. Figure 4 shows a spatial analysis MK trend test using the Z-value. Accordingly, in the spring, the rainfall decreased in the northern part of the Amibara district, and an increasing trend was observed in the Awash Fentale district (Figure 4). The spatial analysis result shows that rainfall has been significantly decreasing in the summer season throughout the study sites. However, the summer season rainfall reduction was most pronounced in the southernmost of the Awash Fentale district and northwest of the Amibara district for the past three decades.

Figure 4 shows the study sites have experienced a mix of increasing and decreasing autumn seasonal rainfall. For instance, the southwestern part of the Amibara district received an increase in autumn rainfall while s slight rainfall increment recorded in the southeastern part of the Awash Fentale district. However, it is important to note that except for the southwestern of Amibara and the southeastern part of Awash Fentale the rest parts of the study sites experienced a decrease in rainfall.

The winter season rainfall also observed a significant decreasing rate (Z < −1.96) throughout the study sites. In the winter season, the magnitude of the decreasing rainfall is very significant in the southern part of the Awash Fentale district. The spatial seasonal rainfall trend analysis (depicted in Figure 4) reveals that rainfall has been increasing throughout Awash Fentale and the southern part of the Amibara district in the spring season. The northern part of the Amhibara district has been experiencing a remarkable reduction in rainfall in the spring season (see Figure 4).

Figure 5 shows the spatial trend of seasonal maximum temperature increased for the past three decades except in the southwest and northeastern parts of the Amibara district. In the southwest and northeast parts of Amibara district during the winter season, maximum temperature exhibited reduction. Hence, the lowest and pick spring season maximum temperature increment was observed in the southern part of Awash Fentale and the northern part of Amibara district, respectively. The spatial analysis confirms that during the spring and summer seasons, maximum temperature increased from the south to the northern part of the study sites. Whereas maximum temperature increased from north to south during autumn season. Thus, during the autumn season, a pick maximum temperature was recorded in the southern part of Awash Fentale district. The highest winter season maximum temperature was observed in northern Amibara and the southern part of Awash Fentale districts (Figure 5).

Figure 6 depicts the nature of spatial variations in the seasonal minimum temperature trend of the study area from 1988 to 2018. Seasonal minimum temperature has been increasing in the four seasons at different scales. During the spring season, the pattern of minimum temperature has been increasing from south to north of the two districts. However, unlike the maximum season temperature increment, the autumn season minimum temperature increased from north to south while it lacks a pattern during the summer and winter seasons. The spatial minimum temperate pick of the summer season was observed in the southern part of Awash Fentale and the northeastern part of Amibara district. The slow minimum temperature increment rate of the summer season was exhibited along the northern and southern tips of Awash Fentale and the southern part of Amibara districts temperature lack pattern (Figure 6). On the whole, however, the data leads to the unmistakable conclusion that the entire study area experienced an increasing trend in the minimum temperature across all seasons due to reinforcing factors including expansion of agricultural land, settlement and invasive species due to land use land cover change. Thus, the expansion of agricultural land, settlement and invasive species could increase local temperatures and moisture availability by altering land surface characteristics, modifying natural ecosystems and contributing to localized climate changes (Shukla et al., 2019). The cumulative effect of these factors can result in a noticeable effect on pasture and water availability and hence pastoral household food security (Ayal et al., 2023).

3.4 Change points (PETIT) test

Recognizing change points is a statistical technique that plays a vital role in spotting climate jumps in the climatological data period (Palaniswami and Muthiah, 2018). This study used the Pettitt test method to detect the possible change point position and evaluate its statistical significance. The Pettitt test was applied to the annual mean rainfall, minimum temperature and maximum temperature. Accordingly, Figure 7 revealed that the mean annual rainfall experienced a significant decrease in 1999 (p = 0.037). The mean annual maximum temperature showed a significant increment in 2001 (p = 002). And the mean annual T-min had its significant mean change at the year of 1999 (p = 0.012).

3.5 The frequency, distribution and the magnitude of drought events

This study calculated the frequency of the drought events from moderate to extreme drought value based on the threshold value (SPI/SPEI ≤ −1) for the whole study period. From Table 3 below, it is evident that during the study period, the two districts experienced 13% to 37% moderate and above seasonal drought (SPI_3 and SPEI_3-months) frequency. For SPI_12 and SPEI_12 months, the area experienced severe hydrological drought frequency events from 43% to 47%. The peak extreme drought events were observed in all seasons except in the summer season of SPI_3.

3.6 The mean temporal evaluation of drought events

Figure 8 below shows the temporal distribution of drought occurrence over the study area. To evaluate the temporal variation of drought events in the area, those which are consecutively less than a threshold value (SPI/SPEI ≤ −1) and have a duration of more than two months have been selected. Accordingly, in the basin for agricultural drought (SPI_3), categorized from moderate to extreme droughts occurred in 1999, 2000, 2001, 2007, 2008, 2011, 2013, 2014 and 2017. For agricultural drought (SPEI_3), moderate to extreme droughts were found in 1999, 2001, 2003, 2007, 2008, 2011, 2012, 2014 and 2017. For hydrological drought (SPI_12), moderate to extreme droughts were found in 1988, 1989, 1999, 2000, 2001, 2002, 2003, 2007, 2008, 2010, 2013, 2016 and 2017. Hydrological drought events (SPEI-12) were identified in 1988, 1989, 1999, 2000, 2001, 2002, 2007, 2008, 2010, 2012, 2013, 2016 and 2017. Generally, in the study period, extreme drought was observed in 1989, 2000, 2001, 2002, 2007, 2010, 2012, 2014 and 2017. In sum, the data analysis on drought events (Figure 8.) revealed the occurrence of seasonal and hydrological drought events at a significant level.

3.7 Implications of climate variability and extremes to food security

From the discussion of the climate data, it has been clearly pointed out that the study area has undergone climate variability, which varied from moderate to extreme levels. That, in turn, reveals that the ecological setting suitable to the age-old agronomic practices has been disrupted. The fluctuation behavior of rainfall and temperature, as well as drought severity would certainly affect the agricultural calendar, soil moisture and pasture availability as well as livestock production and productivity and hence, food security. In sum, the climate variability discerned from metrological data is not something abstract phenomenon but a potent reality that has environmental, socioeconomic, as well as cultural manifestations being observed in the study area. Therefore, it is of a paramount importance to demonstrate the catastrophic impacts of climate variability on people and the physical environment they inhabit.

For a long time, the Afar region rangeland has been characterized by a vegetation cover markedly dominated by different savanna bush, grass and open savannas and woody savannas. Studies reveal that the land use land cover change (LULC) in the Afar region is characterized by 58% Barren land/wasteland (11,501, 000 ha), 3% Grasslands (644, 000 ha), 8% Rainfed-SC-croplands (1,493 000 ha), 6% Irrigated-SC-croplands (1,263 000 ha) and 23% Forest/shrub lands/grasslands (4,531 000 ha) (Gumma et al., 2022). The physical landscape of the Afar region has never been characterized by dense forest cover, even along the Awash River basin. The character of the Afar region rangeland relates to the dominance of open and woody savannah grasslands ideally suitable for grazing and crop cultivation, mainly maize, sorghumand vegetables. That pattern is the norm in the entire lowlands of the Afar region while a few high altitudes exhibit connected woodlands. Therefore, the long-established ecological condition has been exploited by the Afar people both for crop cultivation and animal husbandry (Tessema et al., 2021; Tadese et al., 2019).

However, in recent decades, the Afar pastoralists and agro-pastoralists are confronted with the phenomenon of vegetation cover dynamics which has a stark correlation with the patterns of climate variability. In the same vein, climate variability characterized by increasing temperature trends has made the occurrence of drought events more frequent and devastating. Not surprisingly, studies in the lowlands of Eritrea, which have a similar ecological setting with the lowlands of the Afar, revealed a similar alarming trend of loss of vegetation cover as analyzed based on seasonal and annual time scales. Analysis of the normalized difference vegetation index showed decreases in the LULC shares of croplands and shrublands. The same study attributed the phenomenon to low moisture caused by drought conditions in the semi-arid region. As is the case elsewhere, the northeast African region is undergoing considerable alteration of vegetation cover and the relative distribution of different species induced by climate variability that favors the incidence of drought, which disturbs the regeneration and growth pattern of plants (Simon et al., 2019).

The scientific literature concurs on the conventional view that one of the driving factors for the dwindling of natural habitats and biodiversity has to do with the emergence in and colonization of a given environment by invasive plant species. For instance, the glaring disappearance of shrubland biomes is attributable to the invasive plant known as mesquite (Shiferaw et al., 2021). As far as such a phenomenon is the result of natural processes, it has a lot to do with human-induced factors. Strikingly, the occurrence of a devastating invasive plant in the study area and its astonishing expansion is found to be the result of climatological and environmental enabling environment as well as human activities. In this regard, the most peculiar invasive plant prevalent in the study area is known as Prosopis juliflora.

To start with, it would be misleading to argue that the origin and expansion of Prosopis juliflora in the study area are the result of a spontaneous environmental reaction triggered by the arid environment, decline of rainfall and ever-rising temperatures. Originally, the plant was introduced by human intervention meant to be the best solution to alter bare landmass with lash green vegetation cover in the mid-1970s. The plant was introduced to cover bare lands with green vegetation, to produce firewood, to increase livestock feed and for the purpose of enhancing soil stability (Adnew et al., 2019). In retrospect, however, the measure undertaken without a feasibility study proved to be a catalyst for compounding a preexisting problem of water scarcity. This is because the plant exhausts surface and groundwater throughout the years, especially during the dry season. The unintended effect of introducing Prosopis juliflora to the Afar region is that it has posed a formidable threat to the achievement of sustainable development and climate change adaptation (Shiferaw et al., 2021).

Although the appearance of Prosopis juliflora in the Afar region has never been a natural response to climate variability, its rapid expansion is attributable to increasing trends in temperature and moisture stress. For a long time, the global concern on the impact of Prosopis juliflora has been limited to South America, Central America, South Europe, North Africa and the Caribbean region, dominated by savannahs grasslands (Kyuma et al., 2016). This study found that the environmental setting that allowed the fast spread of Prosopis juliflora in those regions exists in the study area and on a more permissive scale. This is because the minimum temperature and frost weather condition that contribute to arresting the expansion of Prosopis juliflora in those regions do not exist in the study area. In fact, the typical weather condition in Afar, high temperature and little precipitation, has provided the most suitable ecological conditions for the Prosopis juliflora to thrive and expand beyond expectations. The ecological contrast between North African countries and the Afar region explains the sharp contrast in the nature of the problem posed by Prosopis juliflora. Whereas Prosopis juliflora in North African countries is limited to the original place of plantation (Dakhil et al., 2021), the climatic and environmental factor in the Afar region have offered the plant the right conditions for its unfettered expansion.

The colonization of the Afar region by Prosopis juliflora is attributable to the mutually reinforcing factors intrinsic to the nature of the plant itself, as well as the enabling ecological and climatological scenarios. The plant tolerates high temperature. Its deep-reaching roots enable the plant to overcome moisture stress at the expense of underground water. It is naturally endowed with high efficient water use. Its wide canopies prevent other vegetation types from accessing sun light. Prosopis juliflora increases soil fertility by augmenting fine soil particles useful for its expansion. The allelochemicals produced by the plant undermine the survival of other species by diminishing their seed germination as well as growth. Because of such character traits, the traditional vegetation composition of the Afar region has become an easy prey to the predatory invasive plant known as Prosopis juliflora (Dakhil et al., 2021; Kyuma et al., 2016).

The rapid expansion of Prosopis juliflora in the Afar region is assisted by domestic animals. When domestic animals consume the fruit of Prosopis juliflora, the seeds carried in the droppings of animals will be dispersed to long distances, and everywhere they are dropped, the plant grows. Moreover, the nature of climate variability characterized by seasonal and annual temperature increases coupled with the alkaline nature of the soil in the study area has created the right condition for the unmitigated expansion of Prosopis juliflora. Hence, the region is confronted with the rapid shrinking of savannah grasslands and bush forests that for long constituted the right ecological condition on which the erstwhile pastoralist livelihood of the Afar people depended (Dakhil et al., 2021). Therefore, the findings of the study neatly tally with a previous finding in Kenya, which argues that a climate variability characterized by erratic and dwindling rainfall coupled with increases in the annual mean temperature nurtures the expansion of Prosopis juliflora thereby exacting the feed resource available to the livestock population (Kyuma et al., 2016).

There is an emerging discussion about policy response toward Prosopis juliflora. As part of that effort, the pros and cons of the plant are being sorted out. On the positive side, the plant is a good source of food for humans and animals especially during periods of food and feed stress. It is a good source of firewood, biofuel energy, charcoal and timber production (Adnew et al., 2019; Abeje et al., 2019; USAID, 2018). Reports show that local people use the plant to construct their houses and sheds. The plant is also positively evaluated for its benefit regarding flood protection, carbon sequestration, windbreak, treatment for soil salinity and landscape greening. To the contrary, the plant has damaging aspects: undermining the ecosystem services, fostering food insecurity by reducing the volume of milk and meat production. The more Prosopis juliflora depletes animal pasture, animals lose body weight and their market price gets smaller (Adnew et al., 2019). Livestock health is being affected by the plant, which causes a disease local people, “Armko,” which arises when animals consume the raw seeds of Prosopis. On balance, the adverse effects of Prosopis juliflora on ecological resources, socioeconomic conditions, development projects, urban areas and national parks outweigh its controversial benefits (Adnew et al., 2019).

Apparently, there seems to be a contradictory argument in the literature about the chronic shortage of precipitation on the one hand and the recurrent problem of flooding. As discussed above, the analysis of climatic data in this study spanning 30 years offers a compelling argument for climate variability characterized by increment of temperature and decrement of precipitation. Our argument is that the temperature and rainfall measurements in metrological stations offer a misleading impression about the water resources available in the Afar region. For a balanced assessment of water resources available to the Afar region, one must look at the fact that the lowlands of Afar are natural absorbents of heavy rainfall water from the Amhara, Tigrai and central Ethiopia (USAID, 2018). As a result, the Afar region is naturally bound to be a victim or beneficiary of climate phenomena in the Ethiopian highlands. For instance, the massive flooding that occurred in 2010 had damaged considerable crops, infrastructure, irrigation systems and livestock. In the Afar region, about 67,000 people were affected by the flood (HOUSE, 2011). Similarly, a flood incident in 2020 impacted 67,885 individuals, displacing 40,731 people and putting 32,839 people at risk. Further, the adverse impact of the flood incident involved the evacuation of 4,915 people, the destruction of 100 education and health centers and the loss of 16,626 livestock (OCHA, 2020). Table 4 suggests that in the Afar region, recurrent flooding rather than drought is the primary factor responsible for crop damage compared to other regional states.

Finally, the impact of climate variability and the expansion of Prosopis juliflora need to be discussed in light of the strong drive of the government to make the Afar region a vital target of food security through wheat production. In view of the growing demand for wheat, the policy direction of the government to substitute wheat import through local production, the government has made the Afar region one of the priority areas for irrigation-based wheat production. The government is trying to cover 1.5 million hectares in the coming five years with wheat production (Tadesse et al., 2022). In view of the clear danger posed on this ambitious project by the Prosopis juliflora invasive plant, the need to overcome the threat of climate variability on food security requires an all-inclusive approach.

4. Conclusions and recommendations

The study has established the characteristics, manifestations, trends and magnitude of climate variability as witnessed in the Awash Fentale and Amibara districts of the Afar National Regional State. The two main sources of data were measurements of temperature and precipitation taken in situ that span 30 years of temporal horizon. For analysis of the impact of climate variability, the study investigated contemporary reports of pertinent organizations and published articles. Based on analysis of statistical results and periodic reports, it is concluded that the study areas have witnessed extreme climate variability where the main pattern is neither substantial nor directional increases or decreases in measurements of temperature and rainfall. Rather, the peculiar feature of climate variability in the two districts is erratic and highly unpredictable temperature and precipitation. The erratic and unpredictable characteristics of climate variability have been further compounded by the indelible impact of weather conditions in northern and central Ethiopia on the lowlands of the Afar region. Since the Afar region is geographically situated to be on the receiving end of excess precipitation in central and northern Ethiopian highlands, analysis of precipitation merely based on meteorological records is proved to be superficial.

From the above conclusion, it is recommended that policy design and implementation about the management of climate variability while ensuring food security best suited to the Afar natural ecological landscape and the traditional livelihood practices must take elements and conditions of the weather phenomenon on the entire north Ethiopian region. Because the Afar region gets more precipitation from the flow of floods in central and north Ethiopia, agronomic technologies fare better from the application of floodwater technologies rather than the propagation of rainfall-based agricultural technologies. Therefore, the ongoing policy drive for irrigation-based wheat production is highly commended by the findings of this research. Finally, it should be pointed out that the prospect of harnessing climate change-friendly livelihoods for an ideal level of food security requires the concerted and coordinated efforts of all stakeholders toward the eventual elimination of the invasive plant called Prosopis juliflora.

Figures

Figure 1.

(a) Location of study area relative to regional divisions of Ethiopia; (b) the whole districts of Afar region and (c) Awash Fentale Woreda and Amibara Woreda Awash in Afar region

Figure 2.

Mean monthly rainfall, mean monthly T-max and mean monthly T-min of the study area

Figure 3.

Standard anomaly index of mean annual rainfall and temperature from 1988 to 2018 of the study area

Figure 4.

Spatial trend of seasonal rainfall in the study area from 1988 to 2018

Figure 5.

Spatial trend of seasonal maximum temperature in the study area from 1988 to 2018

Figure 6.

Spatial trend of the seasonal minimum temperature of the study area from 1988 to 2018

Figure 7.

Changing point and the mean variation for mean annual rainfall, maximum and minimum temperature from 1988 to 2018

Figure 8.

Trend and variation of different time-scale drought events in the study area for SPI_3. SPEI_3 represents seasonal/agricultural drought, and SPI_12 and SPEI_12 represent annual/hydrological drought in the study period from 1988 to 2018

Table 1.

Classification of the severity of dry/wet events on the calculation of SPI/SPEI

Categories	SPI/SPEI values
Extreme drought	Less than −2
Severe drought	−1.99 to −1.50
Moderate drought	−1.49 to −1.00
Near normal	−0.99 to 0.99
Moderately wet	1.00 to 1.49
Severely wet	1.50 to 1.99
Extremely wet	More than 2

Sources: Li et al. (2015) and Woldegebrael et al. (2020)

Table 2.

Statistical summary of seasonal and annual (rainfall, maximum and minimum temperature)

Statistics	Season	Mini	Max	Mean	SD	Kendall’s tau	Sen slop	p-value	CV*100
Rainfall	Spring	29.42	218.20	111.85	54.58	−0.02	−0.18	0.92	48.00
	Summer	120.90	518.74	246.04	92.39	−0.23	−2.51	0.07	37.9455
	Autumn	19.53	251.08	72.81	46.42	−0.01	−0.12	0.95	63.75
	Winter	3.95	111.34	37.98	30.87	−0.40	−1.61	0.00	81.96
	Annual	220.66	939.61	468.69	133.70	−0.23	−4.12	0.08	28.52
T-max	Spring	27.57	39.99	34.95	2.18	0.24	0.07	0.06	Very high
	Summer	32.89	39.27	36.40	1.47	0.38	0.08	0.00	>>
	Autumn	30.22	36.33	34.13	1.36	0.24	0.05	0.06	>>
	Winter	31.42	36.14	33.44	1.22	0.02	0.00	0.92	>>
	Annual	32.00	37.39	34.73	1.11	0.20	0.03	0.11	>>
T-min	Spring	15.45	23.44	19.46	2.02	0.45	0.15	0.00	>>
	Summer	14.51	24.33	20.48	2.68	0.38	0.15	0.00	>>
	Autumn	13.52	22.28	18.35	2.03	0.43	0.10	0.00	>>
	Winter	12.70	18.29	15.94	1.47	0.18	0.04	0.15	>>
	Annual	15.49	21.50	18.56	1.76	0.38	0.13	0.00	>>
T-mean	Annual	23.78	28.84	26.65	1.22	0.37	0.07	0.00	>>

Source: Authors’ own creation using National Metrology Institute of Ethiopia Data

Table 3.

Drought events frequency, magnitude and peak value in Awash Fentale and Amibara districts (1988 to 2018)

	Tim scale	Duration	Frequency (%)	Magnitude	Peak value
Seasonal drought	Spring	SPI_3	13	−19.5	−4.24
		SPEI_3	23	−34.5	−2.08
	Summer	SPI_3	20	−30	−1.82
		SPEI_3	37	−55.5	−2.11
	Autumn	SPI_3	14	−21	−4.14
		SPEI_3	20	−30	−2.27
	Winter	SPI_3	15	−22.5	−2.83
		SPEI_3	27	−40.5	−2.57
Hydrological drought	Annual drought SPEI_4	SPI_12	43	−64.57%	−3.67
		SPEI_12	47	−70.5	−2.71

Source: Authors’ own creation using National Metrology Institute of Ethiopia Data

Table 4.

Crop damage by geographic area and source

Region	Dominant hazard in cropland damage	Seasonal dynamic of hazard	Dominant hazard contributing to loss of cattle
Amhara	Hail 100%	August–September corresponding to the Kiremt season	Hailstorms
Oromia	Fire 82% Floods 14%	Fires: September flood: March and September	Drought
SNNPR	Drought 13% Flood 12%	Floods: September to December	Flood
Afar	Drought 13% Floods 47%	Drought: March and September Floods: July to November	Floods
Tigray	Drought 58% Floods 35% Hailstorm 8%	Drought: September Floods: June to December Hailstorm: July to December	Hailstorm

Source: USAID (2018)