Analysing The Feature of Meteorological Drought in North East Ethiopia: The Case of Meket and Wadla Districts in North Wello Zone, Amhara Region

The study was undertaken to investigate the magnitude, frequency and trends of drought incidence in North Wello Zone, northeast Ethiopia using monthly rainfall record for the period 1987 to 2017 of Meket and Wadla station. Standard precipitation index and Mann–Kendal test were used to analyse drought event and trends of drought occurrences, respectively. Drought Index Calculator used to analyse standard precipitation index. The coecient variation of the study area for Meket was (21.2%), while for Wadla it was (53%) which showed high inter-annual variability. It was established that both studied stations experienced drought episodes in 1987, 1988, 1991, 1992, 1994, 1998, 1999, 2001, 2006, 2014 and 2015, drought years in the history of Ethiopia. The year 2006 was the most severe and distinct-wide extreme drought episode in both studied stations which standard precipitation index values -2.14 at Wadla and -2.01 at Meket station. The frequency of drought number of years which experienced negative standardize precipitation index values in the total time series of 30 years observed for all time scale at both station is 50 percent and above. The drought magnitude of different time scale varied from slight to extreme severe in the studied stations. The Mann–Kendal trend test shows except two-month timescale at Wadla station, all timescales were not statistically signicant (P<0.05). Generally increasing tendencies of drought were observed during main rainy season and decreasing tendencies of drought during short rainy season and annual scale observed in the study area.


Introduction
Drought is a commonly used term, but the most complex and least realized of all the natural hazards affecting more people than any other hazards (Ashraf & Routray, 2013). Drought generally characterises as a short term meteorological occurrence, which stems from a shortage of rainfall for a long period of time comparing with its average and normal conditions (Alamgir et al., 2015). Furthermore, (Wilhite, 2000b) described drought as a normal, recurring phenomenon of climate that practically occur in all regions of the world.
There are four types of drought namely; meteorological, agricultural, hydrological and socio-economic droughts (Mishra & Singh, 2010;Tallaksen et al., 2004). Meteorological drought is de ciency of rainfall which can be observed immediately (PANU & SHARMA, 2002). Agricultural drought is measured in terms of de ciency in soil moisture, rainfall, ground water and reduction in crop yield (PANU & SHARMA, 2002;Wilhite, 2000a). Hydrological drought on the other hand is de ciency in water availability in surface and subsurface water reservoirs. Socio-economic drought is nal phase of drought that is caused by prolong shortage in agricultural production and food thus affecting overall economy. Recently, (Mishra & Singh, 2010) suggested adding groundwater drought as a fth category. Droughts, which originate from de ciency in precipitation over extended periods of time and affect approximately 60 per cent of the world's population, are the major constraints to viable rain-fed agriculture particularly in the ASALs (Huho & Mugalavai, 2010).The available estimates on drought impacts suggest that, during the period 1998-2017 following the ood, which affected a further 1.5 billion resulting in a huge toll to humanity, killing about 21,563 people (Bettin & Zazzaro, 2018). For the purpose of this study, meteorological drought was adopted where rainfall is commonly used for drought analysis.
Meteorological Drought is based on the degree of dryness or rainfall de cit and the length of the dry period (Kogan et al., 2019). Most often think about drought in relation to precipitation, assessing the degree of dryness (in comparison to a local or regional average) and the duration of the dry period. This is known as a meteorological drought, which is highly speci c to a region as average precipitation may vary considerably spatially. According to (Swain et al., 2017) Meteorological drought occurs when there is a prolonged time with less than average precipitation.
The magnitude and severity of meteorological drought that impacts on social and economic systems of any particular human society will be dependent on the underlying vulnerability and particular region exposed to the event, as well as climate and weather patterns that determine the frequency and severity of the event (Wilhite et al., 2007). In North Wello, prolonged and recurrent drought is the most typical corrupted events of climate change. Remarkably, drought cycle has been shortened than earlier that increases its risk (Oxfam, 2011). As a result, reproductive performance of livestock was reduced despite the fact that livestock mortality is increasing (Dias et al., 2016;Herrero et al., 2010) In North Wello, drought occurs every 1-2 years, compared to every 6-8 years in the past (Mohammed et al., 2018). This threat livestock production system which recurrently erodes the livestock asset before full recovery achieved (Abate & Angassa, 2016).
Over the past decades and recently there is quite a lot of studies with regard to drought which have been carried out in different parts of drought prone areas and in the study areas such as (Mengistu, 2015;Mwadalu & Mwangi, 2013;Opiyo et al., 2014). Most of these studies have been devoted to analyze the pastorals and agro pastorals coping strategies towards impacts of the droughts. Moreover, (Mengistu, 2016), studied the Impacts of Drought and Conventional Coping Strategies in Yabello and Dirre districts of Borana Community, Ethiopia, which the survey results show that drought has severely affecting the livestock resource basis of the pastoralists. (Opiyo et al., 2014) also studied on socio economic impact of drought, their coping strategies and government intervention in Marsabit County, Kenya and the results show that water resources, livestock market infrastructure and provision of human health services are affected by drought thus government intervention is needed. Though studies made so far are important they lack the important ingredients which are the extent of the problem by employing different indices including Standard Precipitation Index (SPI), Mann-Kendall's test for rainfall trend analysis (MK) and Coe cient of variation (CV)

Methodology
The study was conducted in Meket and Wadla districts of North Wello zone in Amhara regional state. North Wello zone is located in southern Ethiopia between 11° 20N'-11° 50' N and 38° 40'-39° 30' E, at a distance of 580 kilometers away from Addis Ababa, the capital of Ethiopia. The study area experiences a bi-modal monsoon rainfall type, where 40% of the 300-900mm annual rainfall occurs during September to October (Belg) and 60% between June to August (Kiremt) (Gissila et al., 2004). The same source further contends that the average annual temperature of the area is 26.5 o C. The principal feature of rainfall in most parts of North Wollo Zone is its seasonal character, poor distribution and variability from year to year. For the past decades, an erratic distribution of rainfall has been the major climatic factor affecting crop yields in the study area. Subsistence mixed (crop and animal) agriculture is the major means of livelihood in the area with an average farm size of about one hectare (ha).

Research Design, Data Sources and Methods of Analysis
The study was used mixed method, particularly the concurrent triangulation approach as research design. The purpose of mixed methods research is to build on the synergy and strength that exists between quantitative and qualitative research methods to understand a phenomenon more fully than is possible using either quantitative or qualitative methods alone (Gay et al., 2012).
Monthly rainfall data from two stations for the period (1987-2017) were obtained from Ethiopian National Meteorological Agency (NMA) and used as an input variable to calculate meteorological drought. To ll the missing values, data were generated following the rst order Markov chain model using INSTAT plus (v3. 6) Software ((Roger Stern, Derk Rijks, Ian Dale, 2006).

Drought Index Calculator (DrinC) which was developed by the Laboratory of Reclamation Works and Water Resources Management, National
Technical University of Athens was used to analyze (SPI) standard precipitation index (Tigkas et al., 2015) Standard Precipitation Index Computation SPI is used to identify the meteorological drought or de cit of precipitation for multiple timescales in the studied stations . The SPI is a z-score and represents the drought event departure from the mean, expressed in standard deviation units. Although SPI can be calculated from 1 month up to 72 months, 1-24 months is the best practical range of application (Guttman, 1999; World Meteorological Organization, 2012). The SPI values were computed at three time-scales, as used (Mohammed et al., 2018) 2 months (SPI-2), 3 months (SPI-3) and 12 months or annual (SPI-12). The SPI-2 was used to assess drought during Belg season, SPI-3 was used to assess droughts during Kiremt seasons, which represent the shorter and main rainy seasons, respectively, and SPI-12 was used to assess the annual drought.
A drought occurs when the SPI is consecutively negative and its value reaches an intensity of -1 or less and ends when the SPI becomes positive. For each month of the calendar year, new data series were created, with the elements equal to corresponding rainfall moving sums (Degefu & Bewket, 2015). Then, the SPI value provides a comparison of the rainfall over a speci c period with the rainfall totals from the same period for all the years included in the historical record (Mohammed et al., 2018;Shahid, 2008).To calculate the SPI, a long-term precipitation record at the desired station is rst tted to a probability distribution (e.g. gamma distribution), which is then transformed into a normal distribution so that the mean SPI is zero (Edwards, 1997;. It is expressed mathematically as follows: Where SPI ij is the SPI of the i th month at the j th timescale, X ij is rainfall total for the i th month at the j th time scale, µ ij and a ij are the long-term mean and standard deviation associated with the i th month at the j th time scale, respectively. Drought Index Calculator (DrinC  Table 1. Where S is the Mann-Kendal's test statistics; xi and xj are the sequential data values of the time series in the years i and j (j >i) and N is the length of the time series.
The variance of S, for the situation where there may be ties (that is equal values) in the x values, is given by: Where, m is the number of tied groups in the data set and ti is the number of data points in the ith tied group. For n larger than 10, ZMK approximates the standard normal distribution (Partal & Kahya, 2006;Yenigün et al., 2008) and computed as follows: The presence of a statistically signi cant trend is evaluated using the Z MK value. In a two-sided test for trend, the null hypothesis Ho should be accepted if at a given level of signi cance. Z1-α/2 is the critical value of Z MK from the standard normal table. For example, for 5% signi cance level, the value of Z1-α/2 is 1.96.
MK test, used by many researchers for trend detection due to its robustness for non-normally distributed data, was applied in this study to assess trends in the time series data (Kendall, 1957). The signi cance level of the slope was estimated using Sen's method. Sen's slope (Q) estimates methods account for seasonality of the precipitation data. This method uses a simple non-parametric procedure developed by sen's (1968) to estimate the slope. The nonparametric tests are used to detect trends but don't quantify the size of the trend or change. Hence, magnitude of the observed trend can be estimated with sen's slope estimator when signi cant (Paulo et al., 2012).

Coe cient of variation
The rainfall variability for Meket and Wadla meteorological stations was calculated using the Coe cient of variation (CV) to evaluate the interannual variability of seasonal and annual rainfall is computed as: Where  The analysis of 12month timescale drought was recorded across the study areas in 1987, 1988, 1991, 1992, 1994, 1998, 1999, 2001, 2006, 2014 and 2015 although there was varying by degree of severity. The total number of moderate droughts at 12month timescale was the 13 months at Wadla and 9 months at Meket. This study con rms that the frequency of occurrences of droughts in the main rainy season was the highest at Wadla station. Generally, the rainfall pattern in the studied stations exhibits certain characteristics that a drought year are followed by another two or three dry years vis-à-vis the wet years. This study is in line with the ndings of (Girma et al., 2016; Hadgu et al.,2013)

Trends of Drought Occurrences
Trends of drought occurrences for 2 months (Short rainy season), 3 months (Main rainy season) and 12 months (annual) timescales were shown in Figures 7 and 8. SPI values of January, February and March were considered to represent the drought conditions from June, July and August (Main rainy season), September to October (Short rainy season) and January-December (annual rainfall), respectively.
The computed SPI values for 3 months timescale during main rainfall season revealed that the occurrence of negative rainfall anomalies or frequent droughts were observed in the 1990s and 2000s in Wadla station while in Meket station frequent droughts were observed in 2010s. The main rainfall season trend for 3 months timescale shows decreasing in Meket and increasing in Wadla station (Figure7& 8). The short rainfall season for 2 months timescale in Wadla station showed increasing trends, which indicates declining of occurrences of droughts from time to time across the study stations while the Meket station showed decreasing trends which implies increasing the occurrences of droughts in the area. The computed SPI value for 12 months (annual) timescale showed increasing trend at both studied station. This implies decreasing of incidences of drought at 12 months (annually).
MK (Mann Kendall) test used by many researchers for trend detection due to its robustness for non-normally distributed data, was applied in this study to assess trends in the time series data (Kendall, 1957).
The MK(Mann Kendall) trend test showed decreasing changes in SPI values in Meket stations, which implies increasing the tendency of drought incidences at main rainy season, while Wadla station showed increasing trend that implies declining of the tendency of drought incidences ( Table  2.) On the other hand, SPI values at the 2month timescale during short rainy season showed increasing changes at both stations, similar increasing trend of drought was also observed at 12-month timescale in both stations. The trend analysis shows except at the two-month time scale in Wadla station there was no statistical signi cance (p < 0.05) of any positive or negative trend of meteorological drought severity and frequency for both stations. Generally, detection of trends using nonparametric methods, including Sen's method and the Mann-Kendall test, showed increasing tendencies of drought during main rainy season and decreasing tendencies of drought during short rainy season and annual scale in the study region.

Rainfall variability
The coe cient of variation of the seasonal and annual rainfall of the stations is presented in Table 3 indicated that rainfall variability at Meket CV=21.2% while, Wadla station has CV=53%. The region has moderate and very high Rainfall variability respectively. Main rainy season (June, July, August) rainfall contributed the highest Percentages 52.2% and 48.3% of rainfall to annual rainfall at Meket and Wadla respectively and short rainy season (September and October) also contribute 21.4% and 24.7% at Meket and Wadla station respectively. This result agreed with the ndings of (Viste et al., 2013) and (Girma et al., 2016), who reported that main seasons contributed the highest contribution to annual rainfall .
The main rainfall season, coe cient of variation range was 32.6% at Meket station which shows high variability while 71.2% at Wadla station which implies very high variability in the study area ( Table 3). The analysis of coe cient of variation for short season (September to October) at Meket station was 48.3% which shows very high variability and 94.4% at Wadla station which is extremely variable. This shows that variability in both areas is higher than main seasonal rainfall which agreed with many other authors (Girma et al., 2016;Hadgu et al., 2013). The short rainy season rainfall at both stations shows the highest variation of rainfall distribution with the highest coe cient of variation, followed by the main rainy season and annual rainfall respectively. This is consistent with previous studies, (Ayal et al., 2018), that conclude that greater rainfall variability is experienced during the small rainy season than the main rainy season and annual rainfall. General, the study site experiences moderate to extremely high inter annual rainfall variability.

Conclusion
This study investigated the frequency, magnitude and trends of drought over north Wello during the short and main rainy seasons and the annual period for the period 1987-2017. The study was able to identify the major drought years within the study period with drought severity. The assessment of meteorological drought showed the occurrence of slight to extreme severe in the historical drought episodes in the studied stations.
The meteorological data results revealed greater rainfall variability is experienced during the small rainy season and followed by the main rainy season and lastly annual rainfall at the both station which in turn call for close follow up attention so as to provide timely response to the vulnerable community. Generally, the implication of this study is that drought events and their negative effects are highly localized in the study area and there is a need for local-scale planning for drought management and response. Local climate studies, such as the present study, should be strengthened since they are useful to provide ne scale climate information for drought risk management and adaptation purposes. Moreover, it may contribute decision makers in planning actions for management of drought and reducing the impact of drought/dryness in the area.

ACKNOWLEDGEMENT
The author would like to thank the National Meteorological Agency and those of Meket and Wadla meteorological stations for the provision of meteorological data and for their cooperation related to the study Author's contributions M A was responsible for all activities of the research process such as the design, data compilation and entry, data analysis, and interpretation of results as well as writing up of the manuscript.

Funding
This study was nancially supported by Bahir Dar University.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations Competing interests
The authors declare that they have no con ict of interests Month timescale SPI at Wadla station Month timescale SPI at Wadla station Month timescale SPI at Meket station Month timescale SPI at Wadla station Trends of SPI value (2, 3 and 12 months) at Meket station.