Analysis of climate variability and its linkage with large-scale climate oscillations: implications to pastoral and agropastoral livelihood resilience in Siti Zone, Somali Region, Ethiopia Open Access

Climate variability and extreme induced risks are among humanity’s most pressing challenges in the 21st century (McMichael et al., 2006). These risks can trigger secondary hazards, leading to widespread impacts, particularly in climate-sensitive sectors such as agriculture, health, water and the broader environment (Kumar et al., 2022; Mbow et al., 2019; Kim, 2012). Although climate change is a global issue, its effects are felt most at the local level, especially in poorer countries, notably in Africa, where adaptive capacity is weak and regions are frequently experiencing climate-related shocks (Malhi et al., 2021; Filho et al., 2021; Thornton et al., 2009; Brown et al., 2007; Conway and Schipper, 2011).

The impact of climate variability and extremes is particularly severe among pastoral and agropastoral communities, which entirely rely on traditional livestock management systems (Mekonnen, 2024; Abrham and Mekuyie, 2022; Habte et al., 2022). These communities are often inhabited in arid and semi-arid regions, where drought and its associated impacts are common (Mekonnen, 2024; Hilina et al., 2024). The Horn of Africa region experiences high rainfall variability and recurrent droughts, leading to repeated food crises and emergencies (Shanko and Camberlin, 1998). In the Horn of Africa, climate variability has played a major role in exacerbating food insecurity (Pricope et al., 2013). Furthermore, future projections suggest increased unpredictability in both the amount and distribution of rainfall, along with rising temperatures and more frequent and severe climate extremes, such as droughts and floods (IPCC, 2022). These changes are expected to exacerbate poverty and food insecurity throughout the Horn of Africa. This is particularly true in the Somali Region, where the arid and semi-arid climate favors livestock-dependent livelihoods (Filho et al., 2021; Meyer, 2015). As a result, pastoralism, which currently supports millions in the region, may encounter significant challenges, bringing it to a crossroads (Hilina et al., 2024). The changes in climate variables could increase vulnerability (Wennström, 2024; Salah et al., 2024) and impose a greater economic burden at both regional and national levels, diminishing its contribution to national development and jeopardizing the livelihoods of pastoral and agropastoral communities (Birkmann et al., 2022; Filho et al., 2020).

In the Horn of Africa and the Somali region, climate extreme events are strongly influenced by large-scale climate oscillations such as the El Niño-Southern Oscillation (ENSO), the Indian Ocean Dipole (IOD) and the North Atlantic Oscillation (NAO) (Williams and Funk, 2011). These oscillations can significantly change the amount, timing and distribution of rainfall, directly impacting water and pasture availability, which in turn affects the livestock system and food security (Aytenfisu et al., 2024; Mohamed, 2024). Typically, droughts and floods are the most common climate hazards associated with these oscillations. However, the magnitude and severity of these climatic extremes can vary over time and across different regions due to various factors (Korecha and Barnston, 2007). To fully understand the specific effects of these oscillations, micro-level research is essential given their diverse impacts in different areas.

The pastoral and agropastoral communities in Dembel and Aysha Woredas of the Siti Zone, Somali Region, Ethiopia, are highly vulnerable to climate variability and extreme events due to their dependence on climate-sensitive livelihoods. Frequent droughts, along with high-intensity rainfall and irregular dry and wet spells, are leading to rangeland deterioration and reducing its capacity to support livestock. Despite these challenges, efforts to support these communities are hampered by a limited understanding of spatiotemporal climate dynamics. This knowledge gap obstructs the development of context-specific adaptation measures needed to build resilience and ensure food security in pastoral and agropastoral communities (Hilina et al., 2024).

This study aims to address this gap by providing new insights into the situation of climate variability and extremes at the micro level. It focuses on analyzing the spatiotemporal variability and trends of rainfall, temperature and extremes of the past three decades. The findings are expected to offer valuable scientific data on historical climate variability, which is essential for enhancing the resilience of pastoral systems and improving food security in the region.

The Somali Region, located in southeastern Ethiopia, covers 376,073 km² and has an estimated population of 6 million, with an annual growth rate of 2.6% (World Bank, 2020). It shares borders with Somalia to the east and southeast, Kenya to the south, Djibouti to the north, to the northwest Afar region and west Oromia Region. It is divided into 11 zones. The region is notable for its extensive rangeland, with approximately 70% of its population engaged in pastoralism, raising cattle, camels and small ruminants. Siti Zone has a population of 750,320, with 286,493 and 463,827 residing in urban and rural areas respectively (CSA, 2017). Although crop farming has been practiced in the region for the past five decades, production has been severely impacted by frequent and severe droughts, along with declining rainfall, which has led to significant soil moisture deficits (Hilina et al., 2024; AbdinAsir et al., 2023). The map of the study area is provided in Figure 1.

The data used in this study were obtained from various sources to ensure a comprehensive analysis. Areal gridded, i.e. 4 km by 4 km minimum temperature, maximum temperature and rainfall data from 1990 to 2020 were obtained from the Ethiopian Meteorological Institute (EMI). This high-resolution data set offers superior data quality and effectively addresses the spatial and temporal gaps in Ethiopia’s national observation network (Esayas et al., 2018). The data quality was checked for its validity and reliability by EMI before being released (Dinku et al., 2018). Due to its enhanced accuracy and reliability, this gridded data set was chosen as the primary data source for the study.

The variability of rainfall in Ethiopia has been linked to several large-scale ocean-atmospheric indices (Alemayehu and Bewket, 2017; Zeleke and Damtie, 2016). Among these indices, the ENSO stands out as a global climate phenomenon with significant impacts on hydroclimatic and, hence, natural resources (e.g. water, pasture) and livelihoods of the community depend on them. In addition, sea-level pressure (SLP), which reflects atmospheric pressure at sea level, provides valuable insights into atmospheric circulation patterns that can lead to extreme events such as floods and droughts. Variations in sea surface temperature (SST) can also induce changes in the heat-flux field, resulting in anomalies in atmospheric circulation and rainfall patterns (Copsey et al., 2006). In this study, the most relevant global climate indices were selected to assess their relationship with local drought and rainfall indices. These are:

• SST in the tropical region, including key indices like the Dipole Mode Index (DMI), which reflects the intensity of the IOD based on SST anomalies between the Western (10°S-10°N and 50°−70°E) and Southeastern (10°S-0° and 90°−110°E) regions. The Pacific Decadal Oscillation index is another critical measure, representing the leading principal component of monthly SST variability in the Northern Pacific (poleward of 20°N). In addition, the ENSO is represented by averaged Niño SST indices, including Niño 1+2, Niño 3 (900–150°W and 5°N–5°S), Niño 3.4 and Niño 4 (150°W–160°E and 5°N–5°S).

• Atmospheric pressure at sea level, or SLP, encompasses key indices such as the Southern Oscillation Index, which measures the pressure difference between Tahiti and Darwin. Other important indices include the North Pacific Index, representing area-weighted SLP over the region from 30°N to 65°N and 160°E to 140°W, the Trans-Polar Index and the NAO. This data was sourced from the National Oceanic and Atmospheric Administration (NOAA) and is available in NOAA's climate data catalog. This selection allows for a comprehensive analysis of the teleconnections between global climate phenomena and local climate variability in Ethiopia (Getachew, 2018).

The trends and distribution behavior of climate variables such as temperature and rainfall can be analyzed using various methods (Kiros et al., 2016; Koudahe et al., 2017). Key statistical measures, including mean, standard deviation (SD), median, kurtosis, skewness, range and coefficient of variation (CV) could be used. The CV, which assesses the relative variability of the data, is calculated using equation (1):

where CV is the coefficient of variation; σ is the SD; and μ is the mean of the sample. The CV helps to understand the consistency or stability of climate patterns over time and across locations. To better understand the degree of variability in climate conditions and tailor adaptation strategies accordingly, the CV is categorized into different classes. For instance, following Hare (2003), the variability is classified as follows: low variability (CV < 20%), moderate variability (20% ≤ CV ≤ 30%), high variability (CV > 30%), very high variability (40% < CV < 70%) and extremely high variability (CV > 70%) (Asfaw et al., 2018). Mapping the CV helps visualize the variability of climate variables more effectively.

The standard anomaly index (SAI) is used to quantify deviations of climate variables from their historical averages. It measures deviations in standardized units by comparing the actual value of a climate variable to its long-term mean (Alemayehu and Bewket, 2017). The SAI is calculated as follows:

where SAI is; X is the annual climate variable data, µ is the long-term annual mean; and σ is the long-term SD of the climate variable. In the case of rainfall, for example, the SAI with negative values represents periods of below-normal rainfall (dry period), whereas positive values reflect above-normal rainfall (wet period). Thus, the SAI can be used to detect climate extremes by identifying values that deviate significantly from the long-term mean. Extreme values are typically identified as those that exceed a certain threshold, often defined as a certain number of SDs from the mean. So, the SAI can be used for meteorological drought assessment based on the precipitation. For example, Agnew and Chappell (1999) provide four classes of drought conditions based on the Z-score of the SAI: 0 to −0.99 (mild drought); −1 to −1.49 (moderate drought); −1.50 to −1.99 (severe drought) and $≼−2.00$ (extreme drought). MCkee et al. (1993) and Viste et al. (2013) used the standardized rainfall anomaly index (RAI) to identify the weather situation as an extremely wet year when RAI > 2 and as an extremely dry or drought year when RAI < −2. Thus, the SAI can be used for the classification of dry and wet severity conditions.

Precipitation concentration index (PCI) is used to analyze the variability (heterogeneity pattern) of rainfall across different scales, such as annual or seasonal. The PCI values were calculated following the method outlined by Oliver (1980) and modified by De Luis et al. (2011), as follows:

where P_i the rainfall amount of the ith month.

According to Oliver (1980), PCI values below 10 indicate a uniform monthly distribution of rainfall (low precipitation concentration), values between 11 and 15 denote moderate concentration, values from 16 to 20 signify high concentration and values of 21 or higher represent very high concentration.

Several researchers agreed that there is no universal solution for serial correlation in the time series (Patakamuri et al., 2020). Trend tests will reduce the effect of positive autocorrelation in the data, increasing the probability of detecting trends when none truly exists and vice versa. This study used parametric and nonparametric methods to perform hydroclimate data and extreme indices trend detection. Parametric tests are stronger, but the data must be normally distributed. Nonparametric tests are “distribution-free” methods, which do not rely on assumptions that the data are drawn from a given probability distribution (Amrender Kumar et al., 2015). Hydroclimate data are nonparametric and the study applied mostly nonparametric models. The parametric method is the linear regression test and nonparametric (rank-based) methods include the Mann–Kendal (MK) (Kendall, 1975; Mann, 1945). Modified MK and trend-free pre-whitening MK trend tests.

The linear regression model measures the pattern or trend of variables over a long period (Kiros et al., 2016). It is calculated using equation (3):

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

The MK trend test is sensitive to the autocorrelation present in a time series. To correct these effects, the modified Mann–Kendall (MMK) test is recommended (Yue and Wang, 2002). Using the standard MK test on highly autocorrelated hydrological data can lead to inaccurate trend estimates, either overstating or understating the actual trends, thereby compromising the validity of the trend test.

The MMK test adjusts for autocorrelation using the modified variance, calculated as:

where V(S)* is the modified variance and the correction factor n/n* is computed by:

where R_h is the autocorrelation coefficient of the ranked data.

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:

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 as many as n = n(n − 1)/2 slop estimates Q_i. 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:

Extreme climate change and intensive human activities can lead to sudden shifts in hydroclimate variables. The Pettitt test is widely used in climate studies to detect these abrupt changes in the mean distribution of a variable. This nonparametric, rank-based test identifies the exact timing of a shift in the mean within a time series and shows the shift’s direction.

The test statistic U_t,T is evaluated across all time points from 1 to T; with the most significant change point identified where the value of ⌊U_t,T⌋ is the largest(Jaiswal et al., 2015). The text statistic U_t is defined as:

where similar to the MK test:

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:

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

where t₀ is consider a significant change point if $Pt0≤0.5.$ This value is then compared with the critical value (Pettitt, 1979). If p < α, *where α is a given significant level), the null hypothesis is rejected, indicating that x_t represents a significant change point at the level α (Du et al., 2013).

Extreme indices were used to assess daily temperature and precipitation events that exceed or fall below specific thresholds. International research organizations, in collaboration with the Expert Team on Climate Change Detection and Monitoring Indicators, have proposed 27 indicators for extreme temperature and rainfall (WMO, 2009). This study focuses on the spatiotemporal variations and daily rainfall trends in the Aysha and Dembel Woredas. Table 1 presents detailed descriptions of the 12 extreme rainfall indicators that are specific to precipitation.

Various drought indices have been employed to analyze drought characteristics across different regions (Morid et al., 2006). The Standardized Precipitation Index (SPI) (McKee et al., 1993), the Palmer Drought Severity Index (PDSI) (Palmer, 1965) and the Standardized Precipitation Evapotranspiration Index (SPEI) (Vicente-Serrano et al., 2010) are widely used methods. Accordingly, in this study, two multi-scalar drought indices SPI and SPEI at three-month and 12-month timescales were used to investigate drought and flood events in the Aysha and Dembel Woredas. Like the SPI, the SPEI is multiscalar; however, it also incorporates temperature data to assess potential evapotranspiration (PET) and calculate the climate water balance (CWB) (Hänsel et al., 2019). Furthermore, the SPEI combines the PDSI’s sensitivity to changes in evaporative demand with the SPI’s ease of calculation and multi-temporal applicability (Vicente-Serrano et al., 2010). These indices were chosen for their effectiveness in assessing drought events in arid-dominated regions like East Africa (Ntale and Gan, 2003), see Table 2 below. In addition, using multiple drought indices provides a more comprehensive understanding of the spatiotemporal distribution of drought in the Aysha and Dembel Woredas (Temam et al., 2019; Alsafadi et al., 2020).

The correlation coefficient (r) ranges from +1 to −1. A value of +1 indicates a perfect positive correlation, meaning that as one variable increases, the other variable also increases predictably. A value of −1 indicates a perfect negative correlation, where an increase in one variable leads to a predictable decrease in the other. A value of 0 signifies no correlation between the variables. Pearson’s correlation coefficient, r, is calculated as:

where n is the number of observations, x and y are the variables and the $x¯$ and $y¯$ are the mean values of the variables, respectively. As suggested by Mukaka (2012), correlations of 0.1, 0.3 and 0.5 are typically considered small, moderate and large, respectively. However, these thresholds are often adjusted in various fields, with correlations between 0.3 and 0.7 regarded as moderate and those above 0.7 considered strong.

The climate of the study sites is significantly influenced by the tropical easterly and westerly winds originating from the Indian Ocean and the Atlantic Ocean. These climatic patterns result in a bimodal rainfall distribution, with two distinct rainy seasons. For instance, in Ayesha, the majority of the annual rainfall occurs during the Spring (March, April and May) and Summer (June, July and August) seasons, contributing approximately 37% and 34.4% of the total annual rainfall, respectively. In contrast, Autumn (September, October and November) contributes 20.3% to the annual rainfall, whereas Winter (December, January and February) accounts for just 8.3% of the yearly total. Similarly, Dembel receives the majority of its annual rainfall during Summer, with 37.4%, followed by Spring, which accounts for 30.1%. The remaining rainfall is spread across Autumn and Winter, contributing 22.1% and 10.3% of the total, respectively (see Figure 2). In both sites, seasonal rainfall is primarily concentrated in the Summer and Spring months, potentially leading to a moisture deficit during the remaining six months. The combination of less rainfall and pattern along with rising temperatures may negatively impact water and pasture availability, further intensifying rangeland degradation, weakening pastoral livelihoods and worsening food insecurity.

Table 3 shows the seasonal and annual rainfall and temperatures trend in Aysha and Dembel Woredas from 1990 to 2020. The analysis shows Aysha Woreda experienced a decrease in seasonal and annual rainfall, except for the summer season. Specifically, rainfall during spring, autumn, winter and annually has decreased at the rate of 0.7, 0.16, 0.5 and 2.76 mm per year, respectively. Similarly, Dembel Woreda has seen a reduction in rainfall across all seasons, with the most significant decrease observed annually, followed by Spring (3.7 mm) and Winter (2.86 mm). The Autumn season recorded the smallest reduction in rainfall. However, the annual decrease in rainfall in Dembel, at a rate of 10.3 mm per year, was statistically significant. A microlevel climate study also confirmed that the Somali region has been experiencing a decreasing trend of seasonal rainfall (Hilina et al., 2024). Unlike Hilina et al. (2024), current trend analysis shows a decrease in annual rainfall in both sites.

It is important to highlight the existence of variations in seasonal rainfall trends between the Aysha and Dembel sites. For example, unlike Dembel, Aysha has experienced an increase in summer rainfall. In addition, the rate of annual rainfall decline has been more pronounced in Dembel. In contrast to the findings of Hilina et al. (2024), the current trend analysis shows a decrease in annual rainfall at both sites, indicating a lack of uniformity in rainfall patterns across the Somali region. The trend analysis confirms that both Woredas, particularly Dembel, have experienced severe moisture stress, leading to potential water shortages and reduced pasture availability. The drastic decline in rainfall during both of Dembel’s rainy seasons, coupled with rising temperatures, has had a serious impact on the livelihoods of pastoral communities in the area. These changes in rainfall patterns and temperatures directly threaten the sustainability of the pastoral system, further exacerbating the vulnerability of the pastoral communities.

Both sites have experienced statistically significant increases in maximum temperatures; however, the rate of increase varies by season and annually across the sites. For example, the maximum temperature has risen by 0.5°C in Ayasha and 0.7°C in Dembel. In addition, both annual and seasonal maximum temperatures are increasing at a faster rate in Dembel compared to Ayasha. In Ayasha, the minimum temperature has increased by 0.4°C, whereas Dembel shows a decrease of 0.13°C in minimum temperature. Both sites experience a decrease in minimum temperatures during the winter season. The pronounced increase in maximum temperatures, along with the mixed trends in minimum temperatures in Dembel, highlights the site’s high vulnerability to climate change and extreme.

Table 4 highlights the seasonal and annual rainfall change points. In Ayesha, an increase in seasonal rainfall change points was observed in Spring 2004, Summer 2014 and Autumn 1997, whereas Winter showed a significant decrease starting in 2008. Similarly, in Dembel, the trend analysis revealed a statistically significant decrease in rainfall in Spring 1996, Winter 2008 and annual 1997. In addition, statistically insignificant decreases were recorded for Autumn rainfall in 1998 and Summer rainfall in 2011.

Table 4 also shows that Aysha Woreda experienced statistically significant increases in maximum temperatures during Spring (1998), Summer (1992), Autumn (2011), Winter (1998) and annually (1998). A significant decrease in minimum temperatures was noted in the Winter of 2010, whereas increases in minimum temperatures were observed in the Summer and Autumn of 2012. Likewise, in Dembel, significant increases in maximum temperatures were recorded in Spring (1990), Summer (2007), Autumn (2001), Winter (2004) and annually (2004). Decreases in minimum temperatures occurred in Summer 1993 and Winter 2006.

The mix of increases and decreases in change points in seasonal and annual rainfall, minimum temperature and maximum temperature trends in study sites presents a challenge for pastoral and agropastoral communities (Salah et al., 2024; Filho et al., 2020; Mbow et al., 2019). Abrupt changes in rainfall and temperature in a short period, especially in regions heavily reliant on rangeland performance, can have severe adverse impacts compared to gradual shifts (Aytenfisu et al., 2024; Birkmann et al., 2022). These turning points often do not align with existing adaptation strategies, making it difficult for communities to adjust without adequate time for preparation. The results also show that the change points in both seasonal and annual rainfall and temperatures occurred at different times. This highlights the importance of micro-level climate analysis for developing more targeted and informed adaptation and mitigation measures, as the observed change points varied across different periods at both sites. Yitea et al. (2021) also emphasized the significance of micro-level rainfall and temperature change points in effectively implementing adaptation measures.

Table 4 depicts the distribution of seasonal and annual rainfall, maximum temperature and minimum temperature in Aysha and Dembel Woredas from 1990 to 2020. In Ayesha, the total long-term average annual rainfall was 685.98 mm, with a SD of 133.8 and a CV of 66.7. In Dembel, the total long-term average annual rainfall was 562.5 mm, with a SD of 168.6 and a CV of 109.

Both sites exhibit variability in annual rainfall. However, interannual rainfall variability is greater in Dembel (SD = 168.6) compared to Aysha (SD = 133.8). In addition, the differing CV (109) for Dembel and CV (66.7) for Aysha highlight the varying degrees of overall rainfall distribution behavior between the two study sites (Kiros et al., 2016; Koudahe et al., 2017; Asfaw et al., 2018). This extreme degree of rainfall variability poses significant threats to the pastoralist way of life in Dembel.

Table 5 also indicates that intraseasonal rainfall distribution is highly variable, with greater variability observed in Dembel compared to Ayasha (see Table 5 SD). The CV value explains that seasonal rainfall distribution ranges from highly to extremely variable in both sites. This erratic nature of seasonal rainfall can complicate adaptation responses and undermine the resilience of pastoral communities. However, unlike the annual rainfall distribution, the CV reveals that seasonal rainfall variability is relatively higher in Ayasha than in Dembel. For example, during the spring and summer seasons, Ayasha experienced significant rainfall variability, with CV values of 90.4 and 70.5, respectively, while Dembel exhibited lower variability, with CV values of 45 and 43.1. The existence of pronounced rainfall variability in Aysha was also evident in the autumn and winter seasons. The annual mean maximum temperatures for Aysha and Dembel were 33.4°C and 31.4°C, respectively. In Dembel, the distribution of seasonal maximum and minimum temperatures was relatively uniform when compared to Aysha (see Table 5).

In addition to examining temporal rainfall trends, spatial analysis provides clearer insight into the extent of rainfall changes across various sites and hence, to identify hotspot areas during specific periods. Figure 3 illustrates the decline in seasonal rainfall across the study sites. During the Spring, rainfall decreased by 1.22–3.71 mm per year. In Aysha Woreda, rainfall reductions ranged from 1.22–2.06 mm per year, whereas Dembel Woreda experienced a more significant decrease of 3.71 mm per year during the same period. The northwest, western and southern parts of Aysha saw a reduction of 2.06 mm per year, whereas the central, eastern and northern regions experienced a relatively smaller decline in rainfall.

In the Summer season, Aysha Woreda exhibited varying levels of rainfall reduction, ranging from 0.12 to 1.8 mm per year. Dembel Woreda, on the other hand, saw a consistent decrease of 1.8 mm per year. The southernmost region of Aysha experienced a reduction of 0.66 mm per year, whereas the northwest, western and southern parts experienced a decline of 2.06 mm per year. Meanwhile, the central, eastern and northern parts recorded a smaller reduction of 0.12 mm per year.

During the Winter season, Aysha Woreda saw rainfall reductions ranging from 0.84 to 2.28 mm per year, whereas Dembel Woreda’s rainfall decreased by 2.86 mm per year. The southernmost part of Aysha experienced the largest reduction of 2.28 mm per year, with the northwest, eastern and southern areas showing a decrease of 1.43 mm per year. The central and northern parts experienced a relatively smaller reduction of 0.84 mm per year.

Autumn recorded the smallest seasonal rainfall reduction, ranging from 0.06 to 1.01 mm per year. Dembel Woreda’s rainfall decreased by 1.01 mm per year, whereas the southern part of Aysha experienced a reduction of 0.73 mm per year. The northwest and southern areas of Aysha saw a decrease of 0.29 mm per year, whereas the central and northern parts experienced the smallest reduction, with only 0.06 mm per year.

As shown in Figure 4, the rate of maximum temperature increase was higher in Dembel compared to Aysha over the study period. In Dembel, the maximum temperature increased by 0.37°C in Winter, 0.30°C in Spring and 0.28°C in Autumn. The rate of temperature increase was slower during summer, with only a 0.17°C rise, compared to the other seasons in Dembel and Ayesha. The rate of maximum temperature change in Aysha lacks uniformity across the study site during the study period. For instance, the southern part of Aysha exhibited the highest temperature increases, except during the Summer. During Winter, Spring and Autumn, maximum temperatures rose by 0.32°C, 0.25°C and 0.25°C, respectively, whereas the Summer increase was 0.23°C. The northern, central and eastern parts of Aysha showed a slower rate of temperature increase compared to the south, west and northwest during Winter (0.20°C), Spring (0.19°C) and Autumn (0.18°C). Notably, during the Summer, the northwestern part of Aysha saw a faster rate of maximum temperature rise at 0.27°C.

As depicted in Figure 5, Dembel and Aysha Woredas experienced a mix of cooling and warming seasonal minimum temperatures. Dembel experienced a decrease in minimum temperature during Summer (0.34°C) and Winter (0.38°C). Similarly, in Aysha, the minimum temperature reduction during winter ranged from 0.46°C to 0.63°C. However, most of Aysha Woreda experienced an increase in minimum temperature during Summer (0.16°C), whereas the southern and northwestern edges of Aysha showed a reduction in minimum temperature, ranging from 0.02°C to 0.34°C, over the study period. Overall, Dembel experienced a faster rate of minimum temperature increase (0.63°C) during Autumn and a decrease (0.34°C) during Summer. In contrast, Aysha exhibited more rapid warming (1.15°C) during Spring and cooling (0.63°C) during Winter. In Dembel, minimum temperatures increased by 0.75°C in Spring and 0.63°C in Autumn, whereas decreases were observed in Winter (0.38°C) and Summer (0.34°C). Like the maximum temperature trend, the rate of change in minimum temperatures in Aysha was not uniform across the study area. The highest increase in minimum temperature occurred in the central, northern and eastern parts of Aysha during the Spring (1.15°C). However, the same locations experienced a decline in minimum temperatures during the Summer (0.16°C) and Winter (0.63°C).

The observed increase in maximum temperatures and the mix of warming and cooling in minimum temperatures in Dembel and Aysha Woredas have important implications for pastoral resilience. Sharp seasonal rises in both maximum and minimum temperatures, coupled with rapid cooling of minimum temperatures and highly variable rainfall, can heighten livestock stress, weaken immunity and lower productivity (Gupta et al., 2023; West, 2003). These temperature fluctuations could also negatively impact rangelands, leading to shortages of pasture and water, which would further weaken pastoral community’s resilience to climate variability. The result highlights the need for adaptation strategies to help pastoral communities manage temperature extremes and sustain their livelihoods.

Figure 6 shows the deviation of annual rainfall from the long-term average. In Ayesha, from 1990 to 2020, 16 years experienced below-normal rainfall, whereas 11 years saw above-normal rainfall. Notable below-normal rainfall occurred in 2012, 2013 and 2015, with the highest positive anomalies recorded in 2018, 2017, 1996 and 2019. A persistent below-normal rainfall trend was observed from 2000 to 2016, whereas rainfall was above normal between 1993 and 1999, and again from 2017 to 2020.

In Dembel, 14 years had below-normal rainfall, whereas 12 years saw above-normal rainfall during the same period. Significant below-normal rainfall occurred in 2015, 2016 and 2017, whereas positive anomalies were recorded in 1996, 1997 and 1998. In Dembel, the first decade had mostly above-normal rainfall, whereas the second and third decades saw more below-normal rainfall. These prolonged periods of below-normal rainfall in both Aysha and Dembel can lead to water and pasture scarcity, whereas periods of above-normal rainfall could result in flooding – both conditions pose significant stress on livestock and threaten the sustainability of pastoral livelihoods.

The PCI offers key insights into rainfall distribution patterns. As shown in Figure 7, Aysha Woreda experienced very highly concentrated rainfall during the years 1993, 1997, 2001, 2003, 2004, 2005, 2006, 2011, 2016, 2017 and 2018. In addition, rainfall was highly concentrated in 1990, 1991, 1994, 1995, 1996, 1999, 2000, 2011, 2013 and 2016. For the remaining years within the study period, rainfall was moderately distributed, except for 1993, which stands out for its normal distribution (Oliver, 1980). These distribution patterns suggest that a substantial amount of rainfall makes runoff instead of percolating into the ground for groundwater replenishment, leading to moisture stress throughout the study period. Similarly, the Dembel site exhibited a range of rainfall distributions, from normal to very high. For example, in 2000, 2011, 2012, 2015 and 2016, Dembel experienced very high concentrations of rainfall. Aside from the years 1993, 2009 and 2019, which showed normal rainfall distribution, the remaining years experienced moderate to highly concentrated rainfall patterns.

Figure 8 illustrates the trends in consecutive dry days (CDD), consecutive wet days (CWD), heavy rainfall events (R20mm), the maximum one-day rainfall (Rx1day), five-day maximum rainfall (Rx5days) and very wet days (R95P) in Aysha and Dembel from 1990 to 2020. In Ayesha, CDD increased from the southwest to the northeastern part of the Woreda, with the highest observed increase being 2.78 days per year. The southern tip of Aysha saw a smaller increase of 0.99 days per year. In Dembel, CDD uniformly increased by 0.99 days per year. Regarding CWD, the central part of Aysha experienced a decreasing trend of 1.17 days per year, whereas an increasing trend was observed in the northwest and southern parts. In Dembel, CWD increased slightly, at a rate of 0.19 days per year.

A significant increase in R20mm was observed in Ayesha, with the central region showing a marked rise in heavy rainfall days, whereas the southern areas experienced a decrease. In Dembel, CWD showed a decrease of 0.17 days per year, with variations observed from the southeast to the western part of the Woreda. Rx1day (maximum one-day rainfall) exhibited an increasing trend from the northwest to the northeastern parts of Ayesha. In the central and northeastern parts of Ayesha, Rx1day (1.05 days per year), Rx5days (1.19 days per year) and R95P (5.18 days per year) all increased. However, in Dembel, these indicators showed a decreasing trend, with Rx1day, Rx5days and R95P decreasing by 1.22, 1.65 and 6.14 days per year, respectively.

In Figure 9, TXx observed the warmest day of the year, the highest temperature ever recorded is displayed. Its increment ranges from 0.1°C and 0.25°C/year. This rating makes it easier to identify Aysha were hotter than Dembel. TNx (Maximum Temperature of Warmest Night) the highest temperature ever measured on the warmest night of the year is shown in figure. The increasing TNx ranges 0.1°C to 0.36°C/year. TXn (Minimum Temperature of Coldest Day) the lowest temperature ever recorded was observed. I show the increasing rate that ranges of 0.1°C to 0.36°C/year. This rating facilitates the identification of areas that are enduring milder cold occurrences. TNn was an indicator of critical to comprehending the effects of cold stress at night. TX10p (Cool Days) the more cool days shown increasing tendency from −0.23 days/year to 0.41 days/year.TN10p (Cool Nights): The number of nights annually where the low falls below the baseline period’s 10th percentile is depicted in Figure 9. Reduced numbers (in blue) show more cool evenings, whereas higher values (in red) show fewer. It shown increasing from −0.34 days/year to 0.38 days/year.

According to Figure 10, both PET figures of Aysha and Dembel done by Hargreaves analysis that consider maximum and minimum temperatures increment, which drives the global warming. Thus, PET results shown below were observed the increasing trend in the two Woredas. The result depicted in this figure is alien from the results obtained from the anomaly figures and PCI. In this case, the maximum PET peak point was shown in 2015.

The findings obtained from CWBs of Aysha and Dembel Woredas observed a decreasing trend that supported the result obtained from PET in the whole study period. The result of CWD indicates the decreasing trend of the water balance in both Woredas. Figures 10–14 illustrate high variability in climate extremes observed across the study Woredas. The analysis reveals that these areas have been subjected to irregular rainfall patterns and a pronounced rise in temperatures over the study period. This temperature increase has, in turn, led to a notable rise in PET and a consequent decline in the overall water balance. Such climatic irregularities underscore the growing challenges posed by climate variability, highlighting the urgent need for adaptive measures to mitigate the adverse impacts on water resources, agriculture and livelihoods in these regions.

Figure 12 presents a comprehensive drought analysis for Ayesha Woreda using the SPI and the SPEI at both three-month (SPI-3, SPEI-3) and 12-month (SPI-12, SPEI-12) timescales. The three-month indices (SPI-3 and SPEI-3) are typically used to capture short-term or seasonal droughts, particularly those affecting agriculture, whereas the 12-month indices (SPI-12 and SPEI-12) are more indicative of long-term or hydrological drought conditions that influence water resources and ecosystem health. The analysis reveals that Ayesha Woreda experienced prolonged seasonal drought conditions over several years, as indicated by SPI-3 and SPEI-3 values. However, there were notable exceptions in 2017 and 2018, during which both indices showed significant positive anomalies, reflecting extreme wet conditions that likely led to flooding. This abrupt shift suggests a strong seasonal climate variability during those years, potentially linked to regional climatic drivers such as ENSO or the IOD. Similarly, the SPI-12 and SPEI-12 indices demonstrate the longer-term hydrological drought trends in the area. The results indicate that Ayesha Woreda endured severe and persistent hydrological drought from 2012 to 2016. This prolonged dry period was subsequently followed by unusually high rainfall levels in 2017 and 2018, marking a transition from drought to extreme wet conditions. These hydrological shifts have significant implications for water availability, rangeland productivity and overall livelihood resilience in the region.

Figure 13 illustrates the seasonal and annual drought patterns in Dembel Woreda, as assessed using the SPI of the rainfall data. The SPI-3 index, which reflects short-term or agricultural drought conditions, indicates that the years 1994, 2013 and 2015 were among the driest seasons experienced in the area. Conversely, the years 1993, 1996, 1997 and 2010 showed wetter seasonal conditions, suggesting better rainfall performance and reduced drought stress during those periods. The SPI-12 analysis, which represents long-term or hydrological drought conditions, reveals that the years 2015, 2016, 2017 and 2018 were characterized by extreme and persistent drought in Dembel Woreda. These consecutive years of hydrological drought likely had significant impacts on water availability, agricultural productivity and local livelihoods. In contrast, the results derived from the SPEI-12, which incorporates both precipitation and PET, show a different pattern. According to SPEI-12, the years 1993, 1995, 1997 and 1998 were the wettest years in Dembel Woreda. This discrepancy between SPI and SPEI highlights the influence of temperature and evapotranspiration in modifying drought intensity and duration, underlining the importance of using both indices for a comprehensive drought assessment.

Figure 14 shows the analysis of both agricultural and hydrological drought conditions in Ayesha Woreda, using the SPEI at three-month and 12-month timescales (SPEI_3 and SPEI_12, respectively). The SPEI is particularly valuable as it incorporates the effects of global warming by factoring in not only precipitation but also other key climatic variables such as PET and the CWB of the area. The results reveal that Ayesha Woreda has experienced both extreme drought and flooding events over the past two decades. According to the agricultural drought index (SPEI_3), the area suffered from persistent and extreme dryness between 2011 and 2015, with a notable recurrence in 2019. In contrast, extreme seasonal flooding was recorded in 2017 and 2018. Regarding hydrological drought (SPEI_12), severe drought conditions were observed in 2012, 2014 and 2015, whereas extreme flood events were again evident in 2017 and 2018.

Figure 15 presents the analysis of both agricultural and hydrological drought conditions in Dembel Woreda, using the SPEI at three-month and 12-month timescales (SPEI_3 and SPEI_12, respectively). These indices help differentiate between short-term agricultural drought and long-term hydrological drought by incorporating both precipitation and PET data, making them useful tools for assessing climate-related stress in the area. The SPEI_3 results indicate that the years 2011–2015, as well as 2019, were characterized by extreme agricultural drought conditions. On the other hand, the years 1990, 1991, 1993, 1997, 1999, 2018 and 2019 were identified as periods of extreme flooding, highlighting significant seasonal climatic variability. Similarly, the SPEI_12 analysis reveals that Dembel Woreda experienced prolonged hydrological droughts between 2011 and 2015. In contrast, the years 2017, 2018 and 2019 were marked by extreme flooding, reflecting considerable shifts in the long-term water balance of the region.

Variability in atmospheric circulation patterns significantly influences local and regional rainfall variations, and there is a connection between climate change and global circulation models (Trenberth, 2011). Figure 16 illustrates the correlation between rainfall extreme indices in Aysha Woreda (CDD, CWD, Rx1day, Rx5day, PRECPTOT and SDII) and global climate indices (SLP, DARWIN, TAHITI, IOD, SST and NINO1.2).

The analysis shows that CDD has a positive correlation with SLP and a negative correlation with DARWIN. CWD exhibited an insignificant correlation with global indices. Rx1day is positively correlated with IOD, SST and NINO1.2 (El Niño region). Similarly, Rx5day, PRECPTOT (total precipitation) and SDII (simple daily intensity index) also show significant positive correlations with IOD, SST and NINO1.2. However, Rx5day and PRECPTOT have a negative correlation with SLP.

In Aysha Woreda, SLP is correlated with most rainfall indices, whereas DARWIN shows mixed positive and negative associations, indicating its variable influence on different rainfall extremes. The TAHITI index generally has moderate correlations with rainfall indices, suggesting a lesser impact. A positive IOD is linked to greater precipitation extremes, particularly in Rx1day. Most rainfall indices exhibit moderately positive relationships with SST, indicating that higher SSTs are associated with increased precipitation extremes. Likewise, NINO1.2 shows positive correlations with most rainfall indicators, suggesting that higher NINO1.2 values (El Niño conditions) are linked to more extreme rainfall events.

Figure 17 shows the correlation between rainfall extreme indices in Dembel Woreda (CDD, CWD, Rx1day, Rx5day, PRECPTOT and SDII) and global climate indices (SLP, DARWIN, TAHITI, IOD, SST and NINO1.2). In Dembel, CDD has a negative correlation with DARWIN and a significant positive correlation with IOD and NINO1.2 (El Niño region). Rx1day and Rx5day both show significant positive correlations with DARWIN.

Increased precipitation extremes in Dembel Woreda, particularly in Rx1day, Rx5day, PRECPTOT (total precipitation) and SDII are closely linked to higher SSTs, a positive IOD and El Niño conditions (NINO1.2). Fewer CDDs are associated with higher pressure in DARWIN, indicating a negative correlation. Mixed correlations are observed between SLP and other indices, with somewhat positive correlations with CDD but neutral relationships with other rainfall extreme indices.

The observed decline in rainfall and rising temperatures are exacerbating moisture stress, posing significant threats to water and pasture resources that are essential for sustaining pastoral livelihoods. This trend highlights the urgent need for feasible adaptation strategies, including sustainable water and pasture management practices, as well as the implementation of climate-responsive crop farming. These findings have critical policy implications, highlighting the importance of integrating climate adaptation measures into regional development planning. Policymakers must prioritize investment in water conservation infrastructure, such as rainwater harvesting systems and efficient irrigation technologies, to address water scarcity. In addition, the promotion of ecosystem management, such as rangeland restoration and improved grazing management, can help enhance the resilience of pastoral systems.

The results also call for policies that strengthen the capacity of vulnerable communities to cope with climate extremes. This includes providing access to climate information services, fostering community based early warning systems and ensuring the availability of financial safety nets, such as insurance schemes tailored to pastoral and agropastoral settings. Collaborative efforts between governments, development partners and local communities are essential to design and implement interventions that address both immediate needs and long-term sustainability goals. Policymakers can safeguard livelihoods, ensure food security and reduce the vulnerabilities to climate change by implementing these measures. This will strengthen community resilience and promote long-term sustainability through adaptive strategies.

The study sites demonstrated erratic patterns in both seasonal and annual rainfall distribution, accompanied by an increase in annual maximum temperatures. Notably, there was a mixed trend observed in minimum seasonal temperatures, with some seasons experiencing increases, whereas others showed declines. In both Aysha and Dembel, maximum temperatures are rising more rapidly than minimum temperatures, indicating a potential shift in the local climate dynamics. Interestingly, in Dembel, the annual minimum temperature decreased, diverging from the prevailing global trend, which typically indicates an increase in minimum temperatures.

Rainfall variability emerged as a significant concern in both locations, characterized by fluctuations that ranged from moderate to extremely variable. This variability, combined with the magnitude of seasonal and annual rainfall trends, demonstrates a lack of spatial and temporal uniformity across the sites. Dembel has exhibited substantial reductions in both annual and seasonal rainfall, as well as higher variability compared to Ayesha. These climatic challenges have profound implications for the resilience of pastoral livelihoods in the region. The inconsistency in rainfall patterns can lead to unpredictable water availability, which is essential for livestock sustenance. Pastoral communities depend heavily on consistent rainfall for pasture regeneration and water sources and any significant deviation can threaten their food security and economic stability.

The increase in maximum temperatures further exacerbates these challenges, as higher temperatures can lead to increased evaporation rates, reducing soil moisture and stressing vegetation. When combined with decreased minimum temperatures, particularly in Dembel, this can disrupt the growth cycles of critical forage plants, further diminishing the resources available for livestock. Given these dynamics, the vulnerability of both Aysha and Dembel to climate change is underscored, highlighting the urgent need for adaptive strategies to enhance the sustainability of pastoral systems. Strategies such as improved water management, diversification of livelihoods and the introduction of climate-resilient forage species could help mitigate the adverse effects of climate variability. Moreover, community-based approaches that involve local knowledge and participation can enhance resilience by empowering pastoralists to make informed decisions in response to changing environmental conditions. By addressing these vulnerabilities and implementing adaptive strategies, pastoral communities can better navigate the challenges posed by climate change and work toward securing their livelihoods in an increasingly uncertain climate landscape.