The Response of Geopotential Height Anomalies to El Niño and La Niña Conditions and Their Implications to Seasonal Rainfall Variability over the Horn of Africa

In this study, we unveil atmospheric circulation anomalies associated with the large-scale tropical teleconnections using National Center for Environmental Prediction (NCEP) reanalysis dataset. Composite analyses have been performed to know the impact of large-scale tropical circulations on the Horn of Africa. The composite analysis performed at the geopotential height of 850 Mb and 200 Mb, and precipitation rate (mm/day) during six strong El Niño and La Niña episodes revealed that the large-scale tropical variability induced climate anomalies in space and time. A substantial decrease in upper-level height (200 Mb) has been observed in the study area during El Niño composite years as compared to the La Niña years. During El Niño conditions, the upper-level divergence initiates low-level vertical motion, thereby enhancing convection, however, during La Niña composite years, nearly contrasting sit-uations are noticed in Belg (February to May) season in Ethiopia. However, geopotential height anomalies at 850 Mb are above-normal during the strong El Niño years, suggesting suppressed convection due to vertical shrinking and enhancement of divergence at the lower level. Compared to the Belg (Febru-ary to May), geopotential anomalies were generally positive during the Kiremt (June to September) season, thereby suppressing the rainfall, particularly in Southern Ethiopia and Northern Part of Kenya. In contrast, an increase in rainfall was observed during the Belg season (February to May).


Introduction
The climate variability at any regional level is predominantly controlled by lower and upper-level atmospheric circulations which in turn are prominently mediated by the nearby and distant Oceanic thermal responses. Likewise, the climate in the Horn of Africa (HOA) varies from arid to tropical monsoon conditions due to the seasonal modulation of upper and lower atmospheric circulation patterns triggered by teleconnection forces [1]. Different scientific reports documented that the climate variability across the HOA and Eastern equatorial Africa (EEA) is mainly influenced by the large-scale seasonal atmospheric circulation as well as the warm waters of the Indian and Pacific Ocean [1] [2]. Besides season climate variations, the climate of the region varies over a much longer period. Internal and external forcing of the climate system results in Decadal and longer-term variability in the climate system [3] [4] [5]. [1] [6] further noted that, on the centennial, multi-decadal, and decadal time, the main mode of climate variability is observed in the Atlantic, Indian, Pacific, and the Southern Ocean and they result in a significant influence on regional as well as global atmospheric circulation.
According to [7], tropical large-scale circulation features dominating the Horn of Africa (HOA) are mainly controlled by a continent-ocean temperature gradient, a seasonal reversal of winds caused by the hemispheric scale circulations driven by convections in tropical oceans and complex topographic orientations in the region. Different studies uncovered the upper and lower level atmospheric circulation patterns are associated with seasonal rainfall variability over the HOA during the dominant rainy seasons [8] [9] [10] [11] [12]. The recent study that investigated changes in the mean state of rainfall over East Africa asserted that rainfall in the region exhibits considerable variability across spatial and temporal extent and this variability is caused by the complex interactions between different large scale atmospheric and oceanic features acting at regional and global scales [13]. Sea Surface Temperature in the equatorial Pacific Ocean is characterized by ENSO (El Niño Southern Oscillation) episodes [10]. Even though the equatorial Pacific Ocean are distant from the HOA, they are significantly correlated with climate variations (temperature, rainfall, upper and lower level circulation patterns, humidity, etc.) over the region, however, the extent and sign of correlations and seasonality are not uniform across the region [14] [15] [16].
Owing to topographic complexity in the region, the northern Ethiopia high lands and northwestern parts of the HOA have boreal summer monsoon from June to September (JJAS) locally known as "Kiremt" and it accounts for 50% to 80% of the annual rainfall over the region [16] [17]. Whereas, the equatorial part of the Great Horn of Africa (GHA) has two rainy seasons, the long rainy season from March to May (MAM) and the short rainy season from October to December (OND) [17] [18]. A diagnostic study carried out by [19] on monsoon dynamics in the region verified that usually two distinct monsoons patterns are The model study performed at the regional level also captured observed circulation anomaly patterns, with divergence at the lower level and convergence at the upper level during rainfall deficit years, but for the wet years, divergence (convergence) was simulated at the upper (lower) level [20]. Similarly, the spatial and temporal variability of rainfall over Ethiopia during summer (JJAS), locally known as Kiremt, is well captured by the satellite-based observations and, model simulation data [21]. According to [21], most parts of the country are dominated by the intensity and position of upper  [23].
In this study, we tried to explore the association between upper and lower level geopotential height anomalies and strong ENSO episodes during selected composite years, and observe enhanced/suppressed convective activities in response to the change in circulation pattern. Our study area encompasses the HOA with special reference to Ethiopia because due to its complex landscape nearly all types of climate zone in the region are observed in Ethiopia. The highly complex terrain features combined with the myriad synoptic systems that produce rainfall variability, has resulted in a very diverse climate that spans eight different climate zones that range from warm to humid highland climate, and Ethiopia encompasses seven of the eight climate zones in the region [24] [25]. This study was mainly focused on two seasons in Ethiopia locally known as Kiremt (June to September) and Belg (February-May). The study has attempted to discern the response of upper and lower geopotential height anomaly to ENSO phenomena on and associated seasonal modulation of precipitation patterns in the study area.

Study Area
The study area is shown in the rectangular black box in Figure 1; below encompasses Ethiopia, Somalia, Kenya, Uganda, Rwanda, Tanzania, and South and North Soudan partially [24] [26], however, our analysis more focused on Ethiopia due to most of the climate zones in the region found there. The region is known for its complex topographic features with the lowest part below sea level found in the north-Eastern part of Ethiopia (Denkel Depression) and Mount Kilimanjaro in Tanzania with 5895 m above sea level. According to [1], the equator

Data
The data from the National Center of Environmental Prediction (NCEP) reanalysis data set has been used by the climate diagnostic center (https://www.esrl.noaa.gov/psd/). [7] [28] pointed out that the reanalysis dataset was derived from historical observations that have been quality controlled and modeled with a modernized, fixed version of the NCEP global data assimilation system. As fully described in [29], the NCEP uses a frozen state-of-the-art global data assimilation system and a database complete as possible. They further noted that the reanalysis project involves, the recovery of the land surface, ship, rawinsonde, pibal, aircraft, satellite, and other data; quality controlling and assimilating these data with a data assimilation system that is kept unchanged over the reanalysis period 1957-96 [29]. This dataset has been used rigorously by numerous researchers to predict the weather, water, climate extremes and to further understand and diagnose the state of the climate system at regional as well as to the globe extent [

Methodology
Composite analysis has been used to explore the relationship between temporal viabilities for precipitation rate (mm/day) and, upper and lower level geopotential height (Mb) in the study area to remote oceanic and atmospheric properties. The composite analysis involves identifying and averaging one or more categories of fields selected according to their linkage to key climate conditions [19]. Results of composites are then used to generate hypotheses for patterns that may be associated with the individual scenarios [20] [31]. The key conditions selected for this study are the strongest El Niño and La Niña years as mentioned in Table 1; the intensity of geopotential height and precipitation rates long-term mean, seasonal mean, and anomalies have been investigated across selected years.
According to Diro et al. [23], composite analysis has an advantage over individual case studies because compositing emphasizes commonly occurring features while smoothing more random fluctuations. The composite analysis is also better than correlation because it allows the study of non-linearity [23].

Composite Analysis of Key Meteorological Variables
The previous studies carried out in the region, especially in Ethiopia revealed that the total rainfall amount was mainly contributed by Kiremt (JJAS) and Belg

Variations in Geopotential Height during the Strong El Niño Years
Previous studies done by various researchers explored that the climate of East

Geopotential Height at 850 Mb during the Strong El Niño Years
As one can see from Figure 2, below, clear contrast has been observed in the long-term mean, a composite mean, and the composite anomaly of geopotential

Geopotential Height at 200 Mb during the Strong El Niño Years
Regarding the upper-level height anomaly at 200 Mb, Figure 3

Geopotential Height at 850 Mb during the Strong La Niña Years
Note that, long-term mean fields are replicated to compare it with composite The cyclone activities and frequency in the Indian Ocean basin have a crucial impact on the convection activities and moisture advection in the southern as well as northern and central parts of Ethiopia [41]. Supporting this premise, [42] argued that, the cyclones that develop in the southwest Indian Ocean (SWIO) usually travel west then southwest, and finally recurve to the southeast before reaching East Africa and, they further noticed that the cyclone/depression can indirectly affect the weather pattern and condition in Ethiopia.

Geopotential Height at 200 Mb during the Strong La Nino Year
The composite anomalies observed in Figure 5(c) and Figure 5  than Kiremt (JJAS) season. A similar observation was documented by [41] justifying that, Belg rainfall is much more influenced by cyclone activities than Kiremt rainfall, which occurs outside the cyclone season of the Southwest Indian Ocean. [41] also noted that, on a daily basis, rainfall activities during the Belg

Precipitation Rate during the Strong El Niño and La Niña Years
From our previous discussion on rainfall variability in space and time, we noticed that a complex relationship has existed during the ENSO period in a historical record. The rainfall anomaly index analysis used by [9] has revealed this fact. Supporting this, the study carried out by [30] applying harmonic analysis to El Niño Southern Oscillation (ENSO) composite of the 6-months Standardized Precipitation Index (SPI) and rainfall anomaly for the 1900-1996 period has unveiled that ENSO response in East Africa rainfall is region and season dependent, and the influence of El Niño is stronger and opposite than that of La Niña.
The composite precipitation rate (mm/day) map presented below in Figure 6   [46]. But, Figure 7(c), and d show nearly the opposite precipitation rates in both seasons when compared to the El Niño composite presented in Figure 6(c), and Figure 6(d), however magnificent anomaly is seen during El-Nino episodes than La Niña.
The seasonal and spatial contrast of precipitation rate due to the modulation of ENSO events has a huge impact on vegetation biomass productivity and growth [47] [48]. According to [47], the transition from El Niño to La Niña con-

Summary and Conclusions
This