Observed and Future Spatiotemporal Changes of Rainfall Extreme Characteristics and Their Dynamic Driver in June-August Season over Africa

Climate change has increased extreme events over globe and the most robust occurrences of concurrent drought and floods have become more common in Africa. This study focuses on the observed and projected analysis of rainfall extremes of consecutive dry day (CDD) and maximum monthly five day precipitation (RX5day) from Expert Team on Climate Change Detection and Indices (ETCCDI) in June-August season over Africa. The daily CORDEX Africa, reanalysis and CRU datasets were analyzed for extreme trends under RCP4.5 and RCP8.5 scenarios for the periods of 1980 to 2100. The spatiotemporal variability, trend, and magnitude of JJA seasonal rainfall performance exhibits a significant decreasing tendency over Eastern Africa compared to West Africa. The observed results of consecutive dry day (CDD) re-veal that increasing trend and moreover RX5day shows that promising positive trend. Both rainfall extremes are influenced by the combined effect of large scale indexes and appear to be correlated negatively and positively with ENSO, NAO and AO. The CORDEX ensemble mean projections of JJA seasonal exacerbate increasing of frequent and intensified drought extremes over Africa. Additionally, the future RX5day indicated that mixed trend and revealed that increasing 3.72%, 2.54% over West and decreases −16.12%, −22.47% over East Africa under RCP4.5 and RCP8.5 respectively. Generally, rainfall extremes of CDD are projected to increase and RX5day shows a mixed trend in the coming periods over Africa and calls for further verification by using high resolution datasets.


Introduction
There is strong evidence that enhanced greenhouse effect will result in not only changes in the mean weather conditions, but also in the increase of the variability of extreme climate [1] [2]. Climate change is already a reality in Africa and there are prolonged and intensified droughts in eastern Africa; unprecedented floods in western Africa and depletion of ecosystems in equatorial Africa [3].
Climate extremes exert a significant impact on the day to day economic development of Africa, particularly in traditional rain-fed agriculture and pastoralism and water resources at all scales [2]. Due to this floods and droughts can cause major human and environmental impacts on and disruptions to the economies of African countries, thus exacerbating vulnerability which is likely associated with regional climate change [4] [5] [6] [7]. Changes in rainfall extremes and patterns can have profound societal consequences, particularly across Africa where rainfall plays a crucial role in sustaining livelihoods and economic development and the change in rainfall across Africa have received much attention during the last 40 years [8]- [13]. Particularly, the Sahel regions of Africa were highly affected by continuous drought during the 1970s and 1980s due to its longevity and severity [8] [13] [14]. These changes in climate extremes of drought and flood affect human lives and have a large impact on society as a whole [15].
Knowledge on characteristics of climate extremes helps us to determine how the climate is changing, and also to identify the direction and prospect for the future. Many recent studies are based on extremes derived from precipitation [16] [17] [18] [19]. According to [2] report revealed that the 21 st century warming due to anthropogenic forcing will be large in Africa, which is likely to increase the number, duration and amplitude of extremes of rainfall, especially in arid and semiarid regions of Africa. Africa is often considered to be the most vulnerable continent to climate change and extremes because of its higher vulnerability and lower adaptive capacity [20]. Therefore analysis of changes in extreme climate in terms of rainfall is particularly relevant for Africa because agriculture and food security are vulnerable sectors by the variability, intensity and frequency of extreme climate [21] [22]. Limited availability of long records of daily climate data in some parts of the world, including Africa, hampers efforts to analyze the impacts of climate change and variability on the frequency and severity of climate extremes [23].
The study of extreme weather and climate events, especially in Africa is still limited, with the main challenge being lack of or access to observed station data of daily resolution [24] [25].
This study will provide an insight evidence of changes of basic rainfall extremes focusing on providing the most comprehensive analysis by using observed and projected rainfall extremes over Africa. Accordingly, our analysis is based on four ensemble Coordinated Regional Climate Downscaling Experiment (CORDEX) regional models for Africa and this paper focuses on a better understanding of characteristics of extreme rainfall in the changing climate over Africa. The analysis will examine the temporal and spatial correlation between the occurrences of high intensity extreme rainfall events with large scale climate indexes of El

Study Area
Africa is the second largest continent and it is bounded on the west by the At-  [27]. Agriculture is considered the largest main economic activity in Africa and it provides employment for approximately 60 percent of the African population, and more than 50 percent of GDP in some countries [28]. In Africa, the distribution of rainfall is very uneven in both space and time and the climate of much of the continent can be classified as arid and semi-arid, receiving less than 700 mm of precipitation each year [29] [30] [31]. Figure 1 indicates a description our study domains and for

Data
The study employs daily rainfall from four ensemble Coordinated Downscaling

Methods
Our study uses standard descriptive statistics, mean, standardized seasonal anomaly, linear regression and correlation to explore the linkage between circu-  [38]. A trend analysis for rainfall extremes was detected and carried out by using the Mann Kendall (MK) test as discussed in depth by [39]. The wind and rainfall anomaly is analyzed by using statistical method of empirical orthogonal function (EOF).

Seasonal Rainfall Characteristics
Understanding the spatiotemporal variability, trend, and changes in precipita-  Table 2). Additionally noticeable variability of JJA seasonal rainfall observed over West African regions. This result is in agreement with [40] who observed that in some portions of African countries the seasonal rainfall performance depicts both increase and decrease but in general the rainfall trends are weak. Due to the variability of the seasonal rainfall during 1980-2017 severe to extreme drought occurred over each region and such types of rainfall extremes affect varies socio-economic activities over Africa (Table 2). Changes in extreme events of droughts and floods have already been observed due to deficit and heavy precipitation events globally including Africa [41]. As indicated by

Observed Seasonal Extremes
The selected observed rainfall extremes of consecutive dry days (CDD) and maximum rainfall five day precipitation (RX5day) are computed based on ETCCDI definitions for each study region (Table 1). CDD (maximum number of consecutive days with rainfall intensity < 1 mm/day) and the result has been shown an increasing trend during 1980-2017. Furthermore, significant enhancement of CDD was observed starting from 1995 and onwards over both regions ( Figure   3(a), Figure 3(b) and Table 5 The result of RX5day revealed that slightly promising positive trend mainly over R1 compared to R2 regions ( Figure 3(c), Figure 3(d) and Table 5). Overall it is seen that the observed consecutive dry day's and RX5day trend shows that increasing trend and which is conducive for the occurrence of extremes like drought and sudden flood over our study area. [46] noted that there has been an increase in rainfall extremes across African regions in latest years. Climate extremes are one of the grand challenges proposed by the World Climate Research Programme (WCRP) and documented by the World Meteorological Organization [47].
The Sequential Mann-Kendall (SMK) test was used to determine the abrupt changes in CDD and RX5day (Figures 4(a)-(d)). According to [48], abrupt shifts refer to the alteration of climate variables from one stable state to another.
When these shifts exceed a certain threshold, due to external forces which aggravate a change of state at a frequency governed by the climate systems, it signifies an abrupt change in the climate variable [49]. Abrupt change is considered   To this end, Figure 4 shows the change points of consecutive dry days and monthly maximum 5-day precipitation during 1980-2017 the study period ( Figure   4(a) and Figure 4(b)). During summer season the CDD was rapidly changing starting from 1995 over both regions. Moreover, RX5day result shows that enhancement during the period of 2000 and onwards over R1 and R2 (Figure 4(c) and Figure 4(d)).

Palmer Drought Severity Index In JJA Season
Extreme precipitation events of drought and flood have major impacts on various socio-economic activities [50]. The Palmer drought severity indicator shows that extreme to severe drought phenomena occurred during 1980-2017 over Africa. Moreover, the intensity and frequency are high over Eastern Africa ( Figure 5). Similarly, the selected anomalies years indicated that drought and flood extreme events are affecting various socio-economic activities over Africa (Table 2).   Previous works of [53] depict that a complex influence of SSTs is responsible for rainfall patterns and trends in the Sahel at different frequency scales, including SSTs-ENSO in the Indian Ocean [54] and the Mediterranean Sea [55]. Furthermore, land-surface-moisture feedbacks play an important role [56].

2) Association of CDD and RX5day extremes with large scale SST anomalies
Understanding the clear relation extremes with various large scale climate indexes is very important to minimize the impact of drought and flood events.
The CDD and RX5DAY have been significantly associated with several large scale climate indexes which include ENSO3.4, SOI, NAO, AO and IOD ( Figure   7, Figure 8 and Table 3). As pointed out in Figure 7 the standardized anomaly of CDD and RX5DAY shows enhancement after 1995 and is highly governed and associated by large scale indexes of ENSO, SOI, NAO, AO and IOD. The

ENSO3.4, AO and IOD index has positively correlated with CDD over R1 and R2
and whereas NAO correlates negatively over both study regions ( Table 3). The extent to which these large scale indexes influence the seasonal cumulative precipitation over Africa has been widely studied by [30] [57] [58]. The Southern Oscillation Index (SOI) and North Atlantic Oscillation (NAO) correlate positively and negatively with RX5day over each region respectively (Table 3). Previous studies of [59]  3) Atmospheric circulation features associated with extremes Figure 9 shows the first EOF modes of 150 hPa v-wind and geo-potential height anomalies over Africa in JJA season. The mid-latitude flow which indicates that a wavelike structure propagating towards African continent approximately extends from northern Atlantic and Europe which have negative signals over eastern African portions (Figure 9). The EOF wind anomalies revealed that higher variability and the variance of the first mode is 37.4% (Figure 9(a) and    and it shows that positive anomalies over R1 than R2 (Figure 9(b) and Figure   9(d)). Previous works of [63] showed that the drought and flood episodes in the   (Figures 10(a)-(c)).
The African Easterly Jet (AEJ) at 700 hPa located between 10˚N and 15˚N which resulted from thermal difference between the Saharan and Atlantic Ocean and it is sustained by the contrasting moist convection to the south and a dry convection to the north and the strengthening of the AEJ is usually connected with a decrease precipitation for the African summer rainfall and this situation exacerbate  Figure 10(e), Figure 11(b) and Figure 11(e)). Atmospheric circulation is an important component of the African summer monsoon [64]. It drives the monsoon inland from the coast and contributes to vertical moisture transport, a crucial requirement for African summer rainfall [65]. The JJA rainfall in the Sahel region exhibits significant variability and the region is particularly vulnerable to major droughts such as occurred in the 1970s and 1980s [66].
The anomaly field of velocity potential/divergence (convergence) associated with the composite wet and dry years are analyzed during June-August seasonal rainfall over Africa ( Figure 12). The shaded lines of Omega over East Africa show that negative trend at all pressure levels and while weak positive over West Africa (Figures 12(a)-(c)). The wet events are characterized by convergence  (divergence) at low level (upper level) winds respectively (Figures 12(a)-(c)). Mainly over our study domains, thus the composite velocity potential during wet years is associated with rising/upward motion over West Africa (R1) and the dry years indicates that which are opposite to the wet years it is associated by divergence (convergence) over the study region (Figures 12(a)-(c)). The composite dry years are therefore characterized with sinking/downward motion mainly over Eastern Africa (R2).

June-August Seasonal Rainfall Performance
Seasonal variations were investigated to provide more detail about the changes in seasonal rainfall over Africa using the empirical orthogonal function (EOF). Understanding the future changes in the seasonal cycle of rainfall over Africa is crucial for establishing appropriate mechanisms to minimize the impact of future rainfall extremes like drought and flood. The analysis of the results obtained by the rainfall changes is presented in Figure 13 under RCP4.5 and RCP8.5 climate scenarios. The results of EOF indicate that the future JJA seasonal rainfall revealed spatial and temporal variability is high for the coming periods over Africa ( Figure 13). The EOF plot of first-third mode revealed that the June-August    Figure 13 indicates that the seasonal rainfall varies by 24.9%, 9.1% and 7.3% respectively under RCP8.5 climate scenario. The projected spatiotemporal seasonal rainfall performance depicts that strong variability under RCP8.5 compared to RCP4.5 scenario. Previous studies of [67], found that strong decrease in precipitation by 2100 under the RCP8.5 scenario. Thus, the results of this investigation depict that the projected rainfall over the study area is towards decreasing trend. Our results are in agreement with [68] who reported that regional climate model, have predicted a significant reduction of precipitation at the end of the century over Africa. Additionally, IPCC reported that Africa will likely experience longer and more intense droughts in the near future [2] [69].

Projected CDD and RX5day Change
Projecting the future climate extremes is important for providing the general information on the trend, intensity, frequency, characteristics and changes of climate extremes and their impact over socio-economic activities [1]. The consecutive dry days over Africa shows that significantly increasing trend in the coming periods under both scenarios and with the higher under RCP8.5 ( Figure  14, Table 4 and Table 5). The projected mean increment of CDD range shows   Figure 14). Moreover, the CDD percentage change shows that increases by 25.11%, 28.02% and increases by 26.49%, 31.66% over R1 and R2 respectively under RCP 4.5 and RCP 8.5 scenarios. This phenomenon indicates that a significant change of precipitation and tending to drying trend over Africa (Table 4). The current result is quite good agreement with [70] who reported that precipitation decreases more than 50% at the end of 21st century and leads to increase of hydro-meteorological disasters in JJA over Africa. Similarly [71] indicated that under global warming the occurrence of extreme climate disasters is predicted to become more frequent with droughts being one of the most severe disasters.
The projected maximum monthly five day precipitation amount (RX5day) revealed that mixed trend of decreasing and increasing under both RCP 4.5 and RCP 8.5 scenarios (Figure 15, Table 4 and  Africa. The frequency of extreme precipitation events is projected to increase globally in the coming periods, including Africa [73] [74], which will also increase the risks of drought, floods, landslides, agricultural and environmental disasters [75].

Conclusions
This paper describes trends of two basic rainfall extremes over Africa by using daily rainfall of historical and projected datasets from CORDEX Africa and CRU Overall the observed and projected extreme rainfall of CDD shows that significantly increasing trend and RX5day shows mixed trends of increasing and decreasing. The variability of these extreme events may lead towards the increment of intensified drought cases and are most likely to devastate and reduce the socio-economic activities of the study area. The intensity and frequency of extreme rainfall events like drought and flood are projected to increase under climate warming [79] [80] [81]. Generally, the variation of rainfall extremes like CDD is high and calls for further verification by using high-resolution datasets.